Effect of Pedestrian Movement onMIMO-OFDM Channel Capacity in an
Indoor Environment
by
Jishu Das Gupta
A Thesis
Submitted for the Degree of
Doctor of Philosophy
School of Engineering Systems
Queensland University of Technology, Brisbane
July, 2010
i
Certificate ii
Statement of Original Authorship
The work contained in this thesis has not been previously submitted to meet re-
quirements for an award at this or any other higher education institution. To the best
of my knowledge and belief, the thesis contains no material previously published or
written by another person except where due reference is made.
———————————————————
Signature of Candidate
————————————
Date
Certificate
Certificate iii
Certificate
Acknowledgement iv
Acknowledgement
During my PhD research work in QUT, I have received enormous amount of
support by various people and institutions. First and foremost, I would like to
thank my supervisor, Dr Karla Ziri-Castro and Dr. Hajime Suzuki for their con-
stant encouragement, support, direction, and technical contributions, without which
the completion of this thesis was not possible. I also wish to thank my co-supervisor
Dr Bouchra Senadji for her unprecedented support.
I acknowledge the support of many staff members from CSIRO ICT centre who
helped me to conduct the indoor radio propagation measurement. My heartfelt
thanks goes to Mr. Mark Barry and Mr. Mark Dwyer from High Performance
Computing Group for their support relating HPC operation. Though not related
to the content of the thesis, I cannot help but spare this space to show my grati-
tude to the staff of the Research Portfolio Office, Faculty of Build Environment and
Engineering including Ms. Diane Kolomeitz, Ms. Elaine Reyes, Mrs. Christine
Percy who have supported me directly or indirectly with a comfortable and efficient
researching environment.
I would like to express my appreciation for the financial support I have received
from QUT in the form of Postgraduate Research Scholarship and my principal su-
pervisor Dr. Karla Ziri-Castro for Supervisor Scholarship, without which this thesis
work would have not existed.
Last but not least I would like to take the opportunity to thank my lovely wife
Rupa and son Ariyan, who have constantly supported me in every possible way to
achieve my goal.
Acknowledgement
Acknowledgement v
Acknowledgement
Dedication vi
Dedication
To my Wife, Rupa
and
Son, Ariyan.
Dedication
Dedication vii
Dedication
Abstract viii
Abstract
The rapid growth of mobile telephone use, satellite services, and now the wire-
less Internet and WLANs are generating tremendous changes in telecommunica-
tion and networking. As indoor wireless communications become more prevalent,
modeling indoor radio wave propagation in populated environments is a topic of
significant interest. Wireless MIMO communication exploits phenomena such as
multipath propagation to increase data throughput and range, or reduce bit error
rates, rather than attempting to eliminate effects of multipath propagation as tradi-
tional SISO communication systems seek to do. The MIMO approach can yield
significant gains for both link and network capacities, with no additional transmit-
ting power or bandwidth consumption when compared to conventional single-array
diversity methods. When MIMO and OFDM systems are combined and deployed
in a suitable rich scattering environment such as indoors, a significant capacity gain
can be observed due to the assurance of multipath propagation. Channel variations
can occur as a result of movement of personnel, industrial machinery, vehicles and
other equipment moving within the indoor environment. The time-varying effects
on the propagation channel in populated indoor environments depend on the differ-
ent pedestrian traffic conditions and the particular type of environment considered.
A systematic measurement campaign to study pedestrian movement effects in
indoor MIMO-OFDM channels has not yet been fully undertaken. Measuring chan-
nel variations caused by the relative positioning of pedestrians is essential in the
study of indoor MIMO-OFDM broadband wireless networks. Theoretically, due
to high multipath scattering, an increase in MIMO-OFDM channel capacity is ex-
pected when pedestrians are present. However, measurements indicate that some
reductions in channel capacity could be observed as the number of pedestrians
approaches 10 due to a reduction in multipath conditions as more human bodies
absorb the wireless signals. This dissertation presents a systematic characteriza-
tion of the effects of pedestrians in indoor MIMO-OFDM channels. Measure-
Abstract
Abstract ix
ment results, using the MIMO-OFDM channel sounder developed at the CSIRO
ICT Centre, have been validated by a customized Geometric Optics-based ray trac-
ing simulation. Based on measured and simulated MIMO-OFDM channel capacity
and MIMO-OFDM capacity dynamic range, an improved deterministic model for
MIMO-OFDM channels in indoor populated environments is presented. The model
can be used for the design and analysis of future WLAN to be deployed in indoor
environments.
The results obtained show that, in both Fixed SNR and Fixed Tx for determinis-
tic condition, the channel capacity dynamic range rose with the number of pedestri-
ans as well as with the number of antenna combinations. In random scenarios with
10 pedestrians, an increment in channel capacity of up to 0.89 bits/sec/Hz in Fixed
SNR and up to 1.52 bits/sec/Hz in Fixed Tx has been recorded compared to the one
pedestrian scenario. In addition, from the results a maximum increase in average
channel capacity of 49% has been measured while 4 antenna elements are used,
compared with 2 antenna elements. The highest measured average capacity, 11.75
bits/sec/Hz, corresponds to the 4x4 array with 10 pedestrians moving randomly.
Moreover, Additionally, the spread between the highest and lowest value of the the
dynamic range is larger for Fixed Tx, predicted 5.5 bits/sec/Hz and measured 1.5
bits/sec/Hz, in comparison with Fixed SNR criteria, predicted 1.5 bits/sec/Hz and
measured 0.7 bits/sec/Hz. This has been confirmed by both measurements and sim-
ulations ranging from 1 to 5, 7 and 10 pedestrians.
Abstract
CONTENTS x
Contents
1 Introduction 1
1.1 Indoor Wireless Communication Services . . . . . . . . . . . . . . 1
1.2 Wireless Channel Characterization . . . . . . . . . . . . . . . . . . 5
1.2.1 Deterministic Modeling . . . . . . . . . . . . . . . . . . . 6
1.2.2 Ray Tracing Simulation . . . . . . . . . . . . . . . . . . . 7
1.2.3 Empirical Modeling . . . . . . . . . . . . . . . . . . . . . 10
1.3 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.4 Objective/Goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.5 Contribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.6 Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2 Theory and Background 17
2.1 SISO Single Carrier System and Channel Modeling . . . . . . . . . 17
2.2 MIMO Single Carrier System and Channel Modeling . . . . . . . . 19
2.3 SISO Multi-Carrier System and Channel Modeling . . . . . . . . . 23
2.4 MIMO Multi Carrier System and Channel Modeling . . . . . . . . 28
2.5 Channel Temporal Variation . . . . . . . . . . . . . . . . . . . . . 30
2.6 MIMO-OFDM Channel Capacity . . . . . . . . . . . . . . . . . . . 33
2.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3 Literature Review 37
3.1 MIMO-OFDM System . . . . . . . . . . . . . . . . . . . . . . . . 37
CONTENTS
CONTENTS xi
3.1.1 MIMO-OFDM General Concept . . . . . . . . . . . . . . . 38
3.1.2 MIMO-OFDM History . . . . . . . . . . . . . . . . . . . . 42
3.1.3 MIMO-OFDM in Practice . . . . . . . . . . . . . . . . . . 43
3.2 Pedestrians and the Indoor Channel . . . . . . . . . . . . . . . . . 46
3.3 MIMO-OFDM Testbeds . . . . . . . . . . . . . . . . . . . . . . . 49
3.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4 Measurement Equipment and Scenarios 55
4.1 Measurement Equipment . . . . . . . . . . . . . . . . . . . . . . . 56
4.1.1 General Description . . . . . . . . . . . . . . . . . . . . . 56
4.1.2 Technical Specifications . . . . . . . . . . . . . . . . . . . 58
4.2 Measurement Locations . . . . . . . . . . . . . . . . . . . . . . . . 61
4.2.1 LOS Deterministic Burst Mode: Room 386 . . . . . . . . . 64
4.2.2 LOS Random Burst Mode: Room 52C . . . . . . . . . . . . 64
4.3 Measurement Procedure . . . . . . . . . . . . . . . . . . . . . . . 65
4.3.1 LOS Deterministic Burst Mode Measurement Procedure . . 67
4.3.2 LOS Random Burst Mode Measurement Procedure . . . . . 67
4.4 Data Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
4.4.1 Deterministic Measurement Scenarios . . . . . . . . . . . . 70
4.4.2 Random Measurement Scenarios . . . . . . . . . . . . . . . 71
4.4.3 Total Measured Data . . . . . . . . . . . . . . . . . . . . . 71
4.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
5 Simulation Software and Scenarios 73
5.1 Simulation Software . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.2 Simulated Locations . . . . . . . . . . . . . . . . . . . . . . . . . 80
5.2.1 LOS Deterministic Simulation: Room 386 . . . . . . . . . 80
5.2.2 LOS Random Simulation: Room 52C . . . . . . . . . . . . 81
5.3 Simulation Scenarios . . . . . . . . . . . . . . . . . . . . . . . . . 81
5.3.1 Deterministic LOS Burst Mode . . . . . . . . . . . . . . . 81
CONTENTS
CONTENTS xii
5.3.2 Random LOS Burst Mode . . . . . . . . . . . . . . . . . . 82
5.4 Data Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.4.1 Deterministic Simulation Scenarios . . . . . . . . . . . . . 83
5.4.2 Random Simulation Scenarios . . . . . . . . . . . . . . . . 83
5.4.3 Total Simulated Data . . . . . . . . . . . . . . . . . . . . . 84
5.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
6 Analysis of Results for Deterministic Scenarios 87
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
6.2 MIMO-OFDM Channel Measurements . . . . . . . . . . . . . . . 90
6.2.1 Average Channel Capacity . . . . . . . . . . . . . . . . . . 93
6.2.2 Channel Capacity Cumulative Distribution Function . . . . 100
6.2.3 Channel Capacity Dynamic Range . . . . . . . . . . . . . . 100
6.3 MIMO-OFDM Channel Simulations . . . . . . . . . . . . . . . . . 102
6.3.1 Average Channel Capacity . . . . . . . . . . . . . . . . . . 102
6.3.2 Channel Capacity Cumulative Distribution Function . . . . 105
6.3.3 Channel Capacity Dynamic Range . . . . . . . . . . . . . . 107
6.4 Measurements Vs. Simulations . . . . . . . . . . . . . . . . . . . . 108
6.4.1 MIMO-OFDM Channel Capacity . . . . . . . . . . . . . . 108
6.4.2 MIMO-OFDM Channel Capacity Dynamic Range . . . . . 110
6.4.3 Capacity Dynamic Range vs Number of Pedestrians . . . . 113
6.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
7 Analysis of Results for Random Scenarios 121
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
7.2 MIMO-OFDM Channel Measurements . . . . . . . . . . . . . . . 124
7.2.1 Average Channel Capacity . . . . . . . . . . . . . . . . . . 124
7.2.2 Channel Capacity Cumulative Distribution Function . . . . 129
7.2.3 Channel Capacity Dynamic Range . . . . . . . . . . . . . . 132
7.3 MIMO-OFDM Channel Simulation . . . . . . . . . . . . . . . . . 133
CONTENTS
CONTENTS xiii
7.3.1 Average Channel Capacity . . . . . . . . . . . . . . . . . . 134
7.3.2 Channel Capacity Cumulative Distribution Function . . . . 137
7.3.3 Channel Capacity Dynamic Range Analysis . . . . . . . . . 140
7.4 Measurement Vs. Simulation . . . . . . . . . . . . . . . . . . . . . 141
7.4.1 MIMO-OFDM Channel Capacity . . . . . . . . . . . . . . 141
7.4.2 MIMO-OFDM Channel Capacity Dynamic Range . . . . . 147
7.4.3 Capacity Dynamic Range vs Number of Pedestrians . . . . 150
7.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
8 Conclusions and Future Work 157
8.1 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . 159
8.2 Research Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . 163
8.2.1 Journals . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
8.2.2 Conferences . . . . . . . . . . . . . . . . . . . . . . . . . . 164
8.3 Future Research Topics . . . . . . . . . . . . . . . . . . . . . . . . 165
8.3.1 Real time MIMO-OFDM Channel Modeling for Realistic
Environment . . . . . . . . . . . . . . . . . . . . . . . . . 165
8.3.2 Quality Improvement using Controlled Scattering Fixture . . 166
CONTENTS
LIST OF TABLES xiv
List of Tables
4.1 Statistical Facts of the Project (Det:Deterministic, Ran:Random,
Mes:Measurement, MO:MIMO-OFDM channel, SO: SISO Single
Carrier channel) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.1 Statistical Facts of the Project (Det:Deterministic, Ran:Random,
Sim:Simulation, MO:MIMO-OFDM channel, SO: SISO Single Car-
rier channel) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
6.1 Measured MIMO-OFDM Channel Capacity for Deterministic Fixed
SNR and Fixed Tx Power . . . . . . . . . . . . . . . . . . . . . . . 98
6.2 Measured MIMO-OFDM Channel Capacity Dynamic Range for
Deterministic Fixed SNR and Fixed Tx Power(90%) . . . . . . . . 101
6.3 Simulated MIMO-OFDM Channel Capacity Dynamic Range for
Deterministic Fixed SNR and Fixed Tx Power . . . . . . . . . . . . 104
6.4 Simulated MIMO-OFDM Channel Capacity Dynamic Range for
Deterministic Fixed SNR and Fixed Tx Power (90%) . . . . . . . . 107
6.5 Measured and Simulated MIMO-OFDM Channel Capacity for De-
terministic Fixed SNR and Fixed Tx Power . . . . . . . . . . . . . 110
6.6 Linear and Quadratic Regression for Different deterministic Mea-
sured and Simulated Scenarios (Sim: Simulation, Mes: Measure-
ment, Lin: Linear Regression, Qua: Quadratic Regression, FSNR:
Fixed SNR, FTX: Fixed Tx) . . . . . . . . . . . . . . . . . . . . . 114
LIST OF TABLES
LIST OF TABLES xv
6.7 Average Linear and Quadratic Regression for Deterministic Mea-
sured and Simulated Scenarios (FSNR: Fixed SNR, FTX: Fixed Tx) 115
7.1 Average Measured MIMO-OFDM Channel Capacity for Random
Scenarios in Fixed SNR and Fixed Tx Power . . . . . . . . . . . . 127
7.2 Measurement Average Channel Capacity Dynamic Range for Ran-
dom Fixed SNR and Fixed Tx with Different Numbers of People . . 132
7.3 Average Simulated Channel Capacity for Random Scenarios with
Fixed SNR and Fixed Tx Power (using middle 90 percent samples) . 136
7.4 Simulated Average Channel Capacity Dynamic Range for Random
Scenarios with Fixed SNR and Fixed Tx . . . . . . . . . . . . . . . 140
7.5 Linear and Quadratic Regression for Different Random Measured
and Simulated Scenarios (Sim: Simulation, Mes: Measurement,
Lin: Linear Regression, Qua: Quadratic Regression, FSNR: Fixed
SNR, FTX: Fixed Tx) . . . . . . . . . . . . . . . . . . . . . . . . . 152
7.6 Average Linear and Quadratic Regression for Different Random
Measured and Simulated Scenarios (FSNR: Fixed SNR, FTX: Fixed
Tx) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
LIST OF TABLES
LIST OF FIGURES xvi
List of Figures
1.1 Change in Mobile Phone Subscriber Number [1] . . . . . . . . . . . 2
2.1 Mathematical Model of the Channel [2] . . . . . . . . . . . . . . . 18
2.2 A Schematic Representation of a MIMO Communication Scheme . 20
2.3 Diagram of a MIMO Wireless Transmission System [3] . . . . . . . 21
2.4 OFDM Signal Plot . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.5 OFDM spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.6 Temporal CW Envelope Fading for a Medium Size Office Building.
Carrier frequency is 915 MHz and both antennas were stationary
during the measurements. (a) Antenna separation 10 m; (b) antenna
separation 20 m. (Measurements and processing by David Tholl of
TRLabs.)[2][nsec=second] . . . . . . . . . . . . . . . . . . . . . . 31
2.7 Example of Measured Temporal Variation of 4×4 MIMO-OFDM
Fixed SNR Channel Capacity with 1 Pedestrian (2 samples per sec-
ond). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.8 Sample of Measured CDF of 4×4 MIMO-OFDM Fixed SNR Chan-
nel Capacity with 1 Pedestrian. . . . . . . . . . . . . . . . . . . . . 36
3.1 Simulated Scenarios (Top View) for Pedestrians Trajectories [4] . . 48
4.1 MIMO-OFDM Channel Sounder . . . . . . . . . . . . . . . . . . . 57
4.2 Details Front Panel View of Transmitter and Receiver . . . . . . . . 60
4.3 A Schematic Diagram of the MIMO-OFDM Testbed . . . . . . . . 61
LIST OF FIGURES
LIST OF FIGURES xvii
4.4 Floor Plan of CSIRO ICT Centre. Measurement Sites, Rooms 386
and 52C, are Highlighted. . . . . . . . . . . . . . . . . . . . . . . . 62
4.5 Experimental Floor Plans . . . . . . . . . . . . . . . . . . . . . . . 63
4.6 Experimental Setup at Room 386 . . . . . . . . . . . . . . . . . . . 65
4.7 Schottky Room (52C), used for Random Trajectory Experiments . . 66
4.8 Schottky Room (52C) Arrangement Showing Tx and Rx Location . 69
4.9 Randomly Moving People between Tx and Rx in Room 52C . . . . 70
5.1 Simulated Comparison for Reflection order Analysis (Fixed SNR) . 76
5.2 Simulated Comparison for Reflection Order Analysis (Fixed Tx) . . 77
5.3 Repeated Simulation Comparison Analysis . . . . . . . . . . . . . 78
5.4 Deterministic Model Room and Pedestrian Block . . . . . . . . . . 79
5.5 Random Model Room and Pedestrian Block . . . . . . . . . . . . . 79
6.1 The 6m Preset Trajectory for Deterministic Measurement Scenarios 89
6.2 A Sample of 4x4 Relative Received Power for Fixed Tx Scenario [5] 91
6.3 A Sample of the 4x4 MIMO-OFDM Sub-Channels when pedestrian
is blocking LOS path[5] . . . . . . . . . . . . . . . . . . . . . . . . 92
6.4 Capacity Analysis for Deterministic Scenarios (4× 4) . . . . . . . 94
6.5 Capacity Analysis for Deterministic Scenarios (3× 3) . . . . . . . 95
6.6 Capacity Analysis for Deterministic Scenarios (2× 2) . . . . . . . 96
6.7 Measured CDF Analysis for Deterministic Fixed SNR and Fixed Tx 99
6.8 Simulated Average Capacity with Different Number of Pedestrians
and Antennas. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
6.9 Channel Capacity CDF Plots for Simulated Deterministic Fixed SNR
and Fixed Tx Scenarios . . . . . . . . . . . . . . . . . . . . . . . . 106
6.10 Average Channel Capacity Comparison for Measured and Simu-
lated Fixed SNR and Fixed Tx . . . . . . . . . . . . . . . . . . . . 109
LIST OF FIGURES
LIST OF FIGURES xviii
6.11 Measured and Simulated Dynamic Range Variation with Different
Numbers of Pedestrians and Antennas. SM: Fixed SNR, measure-
ment. SS: Fixed SNR, simulation. TM: Fixed Tx power, measure-
ment. TS: Fixed Tx power, simulation. . . . . . . . . . . . . . . . 111
6.12 Percentage Dynamic Range Variation with Different Number of
Pedestrian and Antennas. SM: Fixed SNR, measurement. SS: Fixed
SNR, simulation. TM: Fixed Tx power, measurement. TS: Fixed Tx
power, simulation. . . . . . . . . . . . . . . . . . . . . . . . . . . 112
6.13 Linear Regression for Deterministic Fixed SNR . . . . . . . . . . . 116
6.14 Quadratic Regression for Deterministic Fixed SNR . . . . . . . . . 117
6.15 Linear Regression for Deterministic Fixed Tx . . . . . . . . . . . . 117
6.16 Quadratic Regression for Deterministic Fixed Tx . . . . . . . . . . 118
7.1 Area for Random Human Movement in the Measurements Site Room
52C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
7.2 Measured MIMO-OFDM Channel Capacity for Random Scenarios
Vs Numbers of People using Fixed SNR criteria . . . . . . . . . . . 125
7.3 Measured MIMO-OFDM Channel Capacity for Random Scenarios
Vs Numbers of People using Fixed Tx criteria . . . . . . . . . . . . 126
7.4 Measured Average Channel Capacity for Random Scenarios . . . . 128
7.5 Measured CDF Analysis for Random Scenarios in Fixed SNR and
Fixed Tx(0-3 ppl) . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
7.6 Measured CDF Analysis for Random Scenarios in Fixed SNR and
Fixed Tx (0,3,5,10 ppl) . . . . . . . . . . . . . . . . . . . . . . . . 131
7.7 Simulated Average Capacity for Random Scenarios with Different
Numbers of Pedestrians and Antenna Combinations . . . . . . . . . 135
7.8 Simulated CDF for Random Scenarios in Fixed SNR and Fixed Tx
with 0 to 3 pedestrians) . . . . . . . . . . . . . . . . . . . . . . . . 138
7.9 Simulated CDF for Random Scenarios in Fixed SNR and Fixed Tx
with 0, 3, 5 and 10 pedestrians) . . . . . . . . . . . . . . . . . . . . 139
LIST OF FIGURES
LIST OF FIGURES xix
7.10 Measured and Simulated Average Channel Capacity for Random
Scenarios in Fixed SNR . . . . . . . . . . . . . . . . . . . . . . . . 142
7.11 Measured and Simulated Average Channel Capacity for Random
Scenarios in Fixed Tx . . . . . . . . . . . . . . . . . . . . . . . . . 143
7.12 MIMO-OFDM Channel Capacity Dynamic Range Variation with
Different Number of Pedestrians and Antennas for Random Sce-
narios in Fixed SNR. (a)Simulation (b)Measurement . . . . . . . . 145
7.13 MIMO-OFDM Channel Capacity Dynamic Range Variation with
Different Number of Pedestrians and Antennas for Random Sce-
narios in Fixed Tx. (a)Simulation (b)Measurement . . . . . . . . . 146
7.14 Dynamic Range Variation with Different Number of Pedestrians
and Antennas for Random Scenarios using (a) 2 × 2 (b) 3 × 3 (c)
4× 4 arrays. SM: Fixed SNR, measurement. SS: Fixed SNR, sim-
ulation. TM: Fixed Tx power, measurement. TS: Fixed Tx power,
simulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
7.15 Normalized Dynamic Range Variation with Different Number of
Pedestrians and Antennas for (a) 2 × 2 (b) 3 × 3 (c) 4 × 4 antenna
combinations. SM: Fixed SNR, measurement. SS: Fixed SNR, sim-
ulation. TM: Fixed Tx power, measurement. TS: Fixed Tx power,
simulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
7.16 Linear Regression for Random Scenarios in Fixed SNR . . . . . . 153
7.17 Quadratic Regression for Random Scenarios in Fixed SNR . . . . . 154
7.18 Linear Regression for Random Scenarios in Fixed Tx . . . . . . . . 155
7.19 Quadratic Regression for Random Scenarios in Fixed SNR . . . . . 156
LIST OF FIGURES
Nomenclature xx
Nomenclature
3GPP Third Generation Partnership Project
3G Third Generation
4G Fourth Generation
ADC Analog to Digital Converter
ADSL Asymmetric Digital Subscriber Line
AD Analog to Digital
AOA Angle Of Arrival
AOD Angle Of Departure
BICM Bit Interleaved Coded Modulation
BS Base Station
CDF Cumulative Distribution Function
CORDIC Coordinate Rotation Digital Computer
CP Cyclic Prefix
CSIRO Commonwealth Scientific and Industrial Research Organization
CW Carrier Wave
DAB Digital Audio Broadcasting
Nomenclature
Nomenclature xxi
DAC Digital to Analog Converter
DA Digital to Analog
DC Direct Current
DPSK Differential Phase-Shift Keying
DSL Digital Subscriber Line
DSP Digital Signal Processing
DVB-T Digital Video Broadcasting Terrestrial
DVB Digital Video Broadcasting
FDM Frequency Division Multiplexing
FDTD Finite Difference Time Domain
FFT Fast Fourier Transform
FPGA Field Programable Gate Array
FRTT Frustum Ray Tracing Technique
GO Geometric Optics
GSM Global System for Mobile
HIPERLAN High Performance Radio LAN
HPC High Performance Computer
i.i.d. independent, identically distributed
IFFT Inverse Fast Fourier Transform
IF Intermediate Frequency
ISI Inter symbol Interference
Nomenclature
Nomenclature xxii
LA Link Adaption
LDPC Low Density Parity Check
LOS Line-of-Sight
LSD List Sphere Decoder
LTE Long Term Evolution
MAN Metropolitan Area Network
MASCOT Multiple Access Space Time Code
MIMO Multiple Input Multiple Output
MISO Multiple-Input Single-Output
MSE Mean-Square Error
MS Mobile Station
MU-MIMO Multi-User MIMO
NLOS Non Line-of-Sight
OFDM Orthogonal Frequency Division Multiplexing
PC Personal Computer
PDF Probability Density Function
PER Packet Error Rate
PN Phase Noise
QAM Quadrature Amplitude Modulation
QoS Quality of Service
QPSK Quadrature Phase-Shift Keying
Nomenclature
Nomenclature xxiii
RCS Radar Cross Section
RF Radio Frequency
Rx Receiver
SDM Space Division Multiplexing
SIMO Single-Input Multiple-Output
SISO Single Input Single Output
SNR Signal-to-Noise Ratio
STC Space-Time Coding
TOA Time Of Arrival
Tx Transmitter
Wi-Fi Wireless Fidelity
WiMAX Worldwide Interoperability for Microwave Access
WLAN Wireless Local Area Network
ZF Zero Forcing
Nomenclature
1
Chapter 1
Introduction
1.1 Indoor Wireless Communication Services
In 1901, Guglielmo Marconi established the first wireless communication link be-
tween two points across the Atlantic. Since then, wireless communications have
experienced a constant increase in the number of subscribers as can be observed
from the increase in the number of subscribers to mobile phone services shown
in Fig .1.1 [1], as this technology can offer freedom of mobility and accessibility.
Ubiquitous connectivity (i.e., connectivity anytime and everywhere) to the inter-
net, an existing intranet, or to other data services is highly demanded. A significant
amount of wireless communications are carried in indoor environments such as resi-
dential, commercial, and office buildings. This is one of the fastest growing areas of
technology backed by solid commercial potential. In a typical indoor environment,
the transmitter and receiver can be at any location. For example a wireless modem
can be positioned at one end of the house, and a person with a hand held device can
be anywhere else in the house. A Personal Computer (PC) can be linked to internet
using a Wireless Fidelity (Wi-Fi) connection from different locations in an indoor
environment. Additionally, Bluetooth devices can be utilized in conjunction with a
laptop or PC to access certain utilities. Considering all possible mobility scenarios,
presence of interference and change in the location of the antennas and/or objects
Chapter 1 Introduction
1.1 Indoor Wireless Communication Services 2
in an indoor environment, the information from Transmitter (Tx) to Receiver (Rx)
needs to reach the required location with high speed and reliability.
Figure 1.1: Change in Mobile Phone Subscriber Number [1]
From an end user perspective, the growing need and demand of indoor wireless
systems will contribute to indoor wireless data traffic overtaking indoor wired data
traffic [6]. As it can be inferred from the increasing trend in demand for augmented
capacity, data rates, and data services due to the tremendous momentum in wireless
technology created both by the successful deployment of wireless data systems such
as Wireless Local Area Network (WLAN) in 1970 and Global System for Mobile
(GSM) (including the quest for cheaper, smaller and more power-efficient handsets)
in 1982. More recently, the demand for multimedia services such as video-on-
Chapter 1 Introduction
1.1 Indoor Wireless Communication Services 3
demand, and video conferencing are directly adding terabytes of data volume in
the wireless data traffic, as communication services are expected to be available
anytime and everywhere.
In an indoor environment WLANs provide users with the wireless version of the
physical communication medium, required for constructing local area networks of
computers. This typically comprises a transceiver attached to each computer of a
network that communicates with each other and/or with the server base station. In
addition, to freeing users from being tethered at the fixed network access point, the
WLANs have also been recognized as a cost effective solution to the expensive task
of altering already deployed network cables in a building. Especially in situations
where fast deployment of LAN is required, the WLAN is a viable option. A major
challenge of WLANs is that its available data-rate is small, which is typically one
tenth of alternative wired LANs. It is obvious that the main goals in developing next
generations of wireless communication systems are increasing the link throughput
(i.e., bit rate) and the network capacity. Since equipment cost is high and the avail-
able frequency spectrum is limited, to fulfil the high data rate goal, future systems
should be characterized by improved spectral efficiency.
Earlier, research in information theory has revealed important improvements in
spectral efficiency that can be achieved when multiple antennas are applied at both
the transmitter and receiver side, in rich-scattering environments using both nar-
rowband and wideband channels [7, 8]. Also it has been shown that when MIMO
systems are deployed in suitable rich scattering environments, a significant ca-
pacity gain can be observed due to the assurance of multipath propagation [7, 9].
MIMO systems can basically be divided into two groups: Space-Time Coding
(STC) [10] and Space Division Multiplexing (SDM) [7]. STC increases the robust-
ness of the wireless communication system by transmitting different representations
of the same data stream on the different transmitter branches, while SDM achieves
a higher throughput by transmitting independent data streams simultaneously and
at the same carrier frequency. In order to enhance the performance STC uses an
Chapter 1 Introduction
1.1 Indoor Wireless Communication Services 4
advanced signal processing algorithm at the receiver to combine the signals origi-
nated from the different transmitters. In case of SDM, advanced signal processing
algorithms at the receiver recover the parallel streams of data that are mixed-up
in the air. The latter technique usually requires multiple receive antennas, too, to
ensure adequate performance. In MIMO system, advanced signal processing algo-
rithms at the receiver combine the signals originated from the different transmitters
to enhance the performance as well as recover the parallel streams of data that are
mixed-up in the air. The highest average spectral efficiency gains are achieved
when the individual channels from every transmit antenna to every receive antenna
can be regarded to be independent. In practice this is the case in rich-scattering
environments, preferably with Non Line-of-Sight (NLOS) path between transmitter
and receiver. In general, MIMO can be considered as an extension to any Sin-
gle Input Single Output (SISO), Single-Input Multiple-Output (SIMO) or Multiple-
Input Single-Output (MISO) system. A typical SISO single carrier system, with one
transmitter and one receiver, transmits information over a single data channel. This
data transmission is subject to a complex, random and time-varying indoor radio
propagation channel. Considering all the different possibilities of reflection, refrac-
tion and scattering, the transmitted signal most often reaches the receiver by more
than one path, resulting in a phenomenon widely known as multipath. The typi-
cal WLAN standards IEEE 802.11 [11], IEEE 802.11b [12] and IEEE 802.11a/g
[13] are usually deployed in indoor environments, while the probability of having
no direct communication path between transmitter and receiver is high. Therefore,
rich-scattering indoor environment conditions significantly favor the use of MIMO
system for WLAN. The bandwidth efficiency is increased by a factor proportional
to the number of transmitting/receiving antennas compared to that of a single trans-
mitter to single receiver system.
MIMO together with conventional bandwidth efficient modulation schemes such
as Orthogonal Frequency Division Multiplexing (OFDM) have become available in
commercial WLAN products. IEEE standard 802.11 Working Group, is incorporat-
Chapter 1 Introduction
1.2 Wireless Channel Characterization 5
ing a Multiple Input Multiple Output Orthogonal Frequency Division Multiplexing
(MIMO-OFDM) based scheme aiming to achieve higher bandwidth efficiency [14].
In a strong multipath condition, different components of multipath add construc-
tively and destructively. Thus, the resulted received signal can vary as a function of
frequency, location and time. These variations are collectively referred to as fading
and can lead to severe distortion of the received signal. OFDM can mitigate this
problem efficiently. In OFDM a wideband frequency-selective fading channel is
split up into multiple orthogonal narrowband frequency-flat fading channels (i.e.,
subchannels or subcarriers) of which each can be equalized in a trivial way. Com-
bined with coding, this principle also results in robustness against narrowband inter-
ference. Moreover, the ability to include a proper guard interval between subsequent
OFDM symbols provides an effective mechanism to handle Inter symbol Interfer-
ence (ISI) caused by severe multipath propagation. It is important to consider that,
to provide a strong and reliable service for an indoor WLAN using MIMO-OFDM
system, characterization of the wireless channel is crucial. Wireless system design
strongly based on channel characterization can steer the outcome to achieve high
throughput using same frequency bandwidth, as well as can improve the reliability
of the transmitted data.
1.2 Wireless Channel Characterization
The design and successful implementation of any wireless communication system
strongly depends on a detailed understanding of the channel characteristics. Some
parameters that characterize the wireless channel include received signal, statisti-
cal characteristics of fading, and impulse response [15, 16]. In order to maximize
the use of limited radio frequency bandwidth, the system needs to be carefully de-
signed to suit the characteristics of the radio channel in which will operate. System
designers usually wish to characterize the channel parameters of a given wireless
environment, in order to find an optimum solution in terms of system performance
Chapter 1 Introduction
1.2 Wireless Channel Characterization 6
and cost. The wireless channel characteristics determine the choice of frequency
bands, modulation, coding and diversity schemes. For decades, a constant growth
in the wireless technology market has attracted many researchers to contribute to
the improvement of wireless channel characterization [4, 16–19]. The most reliable
method for channel characterization is to conduct measurements on the site of inter-
est, as the MIMO-OFDM channel capacity has been found to be strongly dependent
on the local scattering environment for Line-of-Sight (LOS) situations [15]. In ad-
dition, exactly how much correlation would geometry of the environment cause to
MIMO sub-channels in indoor LOS environments is still largely unknown. For ex-
ample, the inverse power law exponent of averaged signal power in an office build-
ing environment varies from 2 to 6, within which the designer is left to choose value
for his/her building of interest. If channel parameters can be predicted analytically
with the use of computational resources, the method can be very useful in providing
accurate channel information. To this end, so called deterministic modeling, which
is discussed in the following section, has been introduced.
1.2.1 Deterministic Modeling
In the deterministic modeling approach, detailed information such as the structure
and construction material of a building, or the location and field pattern of antennas
are coupled with electromagnetic theory to predict analytically and computationally
radio wave propagation inside buildings. The deterministic modeling approach is
expected to provide the following advantages:
• Accurate prediction of radio channel characteristics at various frequencies
inside many dissimilar buildings, which can replace expensive measurements.
• Detailed evaluation of channel performance by different system parameters
in a particular indoor environment.
• Prediction of very site-specific channel characteristic information, which can
be used to determine system parameters that are site-specific in nature, such
Chapter 1 Introduction
1.2 Wireless Channel Characterization 7
as the location, characteristics, and number of base stations.
In order to achieve the aforementioned advantages, the following two features are
considered to be essential:
• High prediction accuracy: Accurate enough to be used as the system param-
eter evaluation tool.
• Low computational requirement: Inexpensive enough so that the replacement
of expensive measurement is justifiable.
The deterministic modeling approaches range from simple path loss models in-
corporating attenuation through building objects [20] to full-wave analysis such
as the Finite Difference Time Domain (FDTD) method [21]. In general, simple
models only provide a gross average of channel characteristics, while its compu-
tational requirement is kept to a minimum. The full wave analysis enables users
to extract detailed and reliable channel characteristics, however the computational
requirement of the techniques is very high, making it only suitable for predictions
in simple environments.
Using the deterministic modeling approach, it is desired to generate a compre-
hensive representation of channel characteristics in a particular indoor environment.
Channel characteristics can be presented in many forms by analyzing the data col-
lected using measurements as well as conducting simulations. The collected mea-
sured information in the form of received power can be utilized to estimate average
channel capacity. In addition, difference between maximum and minimum channel
capacity can also be considered valuable information for channel characterization.
As this difference is proportional to the de-correlation of the sub-channels due to
the blocking of the LOS path by a larger number of pedestrian or other obstacles.
1.2.2 Ray Tracing Simulation
Among the prediction techniques for deterministic modeling, ray tracing technique
has been most utilized in the last decade. The ray tracing technique is based on
Chapter 1 Introduction
1.2 Wireless Channel Characterization 8
the Geometric Optics (GO) , which represents an approximated solution for the
Maxwells equations [22]. GO consists of a set of laws that are based on the light-
like nature of the wave propagation, which is strictly valid only as long as the di-
mension of interacting objects in the modeled building environment is large enough
compared to the wavelength. While the formulae for prediction of channel param-
eters are largely the same among the reported works that utilized the ray tracing
technique, various ray tracing algorithms have been proposed which aim to deter-
mine valid GO propagation paths efficiently and accurately. The conventional ray
tracing algorithms for indoor radio propagation channel prediction can be catego-
rized into four groups [22]:
1. image model based technique
2. ray launching based technique
3. ray tube tracing based technique
4. frustum ray tracing based technique
These techniques can be summarized as follows:
• The image model based ray tracing technique creates images of transmitter
or receivers by referencing them along side building objects. It can determine
the exact GO propagation path with no additional path error. However, its
computational complexity can be expressed as y× nr, where y is the number
of receiving points, r is the number of multiple reflections (reflection order),
and n is the number of objects in the model, which can become unmanageably
high when applied to a complex building model. In addition, its calculation
time is linearly proportional to the number of receiving points.
• The ray launching based technique casts rays uniformly to all directions from
the transmitter, irrespective of the location of observation points. The re-
ception of rays at an observation point is achieved by the use of a reception
Chapter 1 Introduction
1.2 Wireless Channel Characterization 9
circle or sphere. The computational burden of the algorithm can be expressed
as x × 2r where x is the number of rays launched from the transmitter. The
complexity of the building model has lesser impact on the computational load
in this case, making it suitable for prediction in a complex building structure.
However, the ray launching based technique introduces a unique error derived
from the use of the reception area algorithm. The error may be less obvious
when prediction is to be performed at single or isolated observation points,
but becomes apparent in area-type prediction.
• The ray tube tracing based technique casts ray tubes instead of rays uniformly
in all directions from the transmitter, and thus achieves a similar calculation
efficiency to that of ray launching based techniques. The advantage of using
ray tubes is to eliminate the error associated with the use of the reception area
algorithm. Multiple observation points can be found inside a ray tube, making
it suitable for channel characteristic map prediction. However, the ray tube
tracing technique has an inherent problem in determining correct GO shadow
and lit boundary.
• The frustum ray tracing technique, adopts a very different approach from the
conventional methods, utilising a fast line-clipping algorithm instead of the
conventional time-consuming ray intersection test. The result is to achieve
both calculation efficiency and prediction accuracy. With those techniques, it
becomes feasible to obtain a thorough GO solution at a large area constitut-
ing a channel characteristic map with many observation points (in the order
of 103 to 105) in a complex building environment (with the number of objects
less than 103) within less than approximately 2 hours of calculation time on
a typical PC. The methods provide the designers of any indoor wireless com-
munication services an inexpensive means of simulating and characterizing
the radio channel at various sites.
Chapter 1 Introduction
1.2 Wireless Channel Characterization 10
In this thesis, all conducted measurement were replicated by frustum ray tracing
technique. The simulation allows to design an indoor environment specifying a
given size as well as permeability and conductivity of the building materials. Within
the given room a human body model has been implemented with moving capability.
The body model can be moved randomly as well as in a deterministic fashion. In
addition, a function for avoiding collisions between different body blocks, while
they are moving in a random fashion, has also been incorporated.
1.2.3 Empirical Modeling
Any prediction method needs to be verified by measurement results, in order to
prove its applicability for designing a given system. Several authors [23, 24] present
tutorials on the overview of empirical modeling in the area of MIMO systems and
the operation of MIMO wireless communication systems. In [17], author presented
an indoor MIMO-OFDM channel measurements using 4× 4 MIMO sub-channels,
117 OFDM sub-carriers, and 6400 receiving antenna array locations per local area,
which provide a statistical relationship between the characteristics of multipath
propagation. In addition, the performance of MIMO systems had been evaluated
in [25] considering multipath propagation. It is well established that the perfor-
mance of MIMO systems is dictated by the nature of propagating channels and the
resulting fading correlation due to multipath propagation features, as well as the
antennas’ mutual coupling. The findings have revealed that the achievable indoor
MIMO capacity is a function of the dominant propagation mechanisms, including
the number of effective multipaths. The time varying effects on the propagation
channel within populated indoor environments depend on different pedestrian traf-
fic conditions, and is related to the particular type of environment considered. As
shown in [26], investigations have been carried out through measurements of the
indoor narrowband propagation channel at 5.2 GHz. Accurate modeling of prac-
tical MIMO-OFDM channels is important for designing and optimizing this trans-
mission scheme. Sufficiently rich multipath signal propagation has been found in
Chapter 1 Introduction
1.3 Motivation 11
MIMO channels operating within indoor environments [15, 18].
Empirical modeling of channel variations caused by the relative positioning of
pedestrians is essential in the study of indoor MIMO-OFDM broadband wireless
networks. Since pedestrian movement potentially causes scattering and consequent
temporal variations in MIMO-OFDM channel capacity. Different pedestrian traffic
conditions within populated indoor environments cause time varying effects related
to the particular type of environment considered. In [4, 16, 26, 27] authors have
reported empirical modeling of human body shadowing effects in an indoor envi-
ronment for narrowband channels. So far, to the best of author’s knowledge, none
of the reported studies regarding MIMO-OFDM systems in indoor environments
have conducted a systematic measurement campaign to characterize human body
movement effects in wideband MIMO-OFDM channels.
In this thesis, the channel fading characteristics considering human body move-
ment effects in wideband MIMO-OFDM channels are empirically characterized.
Findings show, MIMO-OFDM channel capacity in a local area is strongly depen-
dent on the local scattering environment in the case of a LOS situation .
1.3 Motivation
The work in this thesis is motivated essentially by fundamental questions regarding
site-specific modeling of indoor MIMO-OFDM channels in the presence of pedes-
trians. The main research questions that motivated this thesis are:
• How human body is affecting the MIMO-OFDM channel characteristics in
an indoor environment?
• How the average MIMO-OFDM channel capacity is behaving when more
people are introduced in an indoor environment?
• How the average MIMO-OFDM channel capacity is behaving when more
numbers of antenna elements are deployed in an indoor environment?
Chapter 1 Introduction
1.3 Motivation 12
• How the average MIMO-OFDM channel capacity dynamic range changes
with the number of pedestrian in an indoor environment?
• How the average MIMO-OFDM channel capacity dynamic range changes
with the number of antenna elements in an indoor environment?
• Increase robustness of the simulation by incorporating realistic populated in-
door environment?
Due to the growing mobile telephone use, satellite services, and wireless In-
ternet and WLANs, a significant change, namely the use of multiple antennas, has
been introduced in the telecommunications and networking technologies. With in-
creasing need of indoor wireless communications, it has become a prevalent interest
to the researchers to design more effective and efficient indoor radio wave propa-
gation model for indoor populated environments. The ongoing changes in the field
led researchers to look more in depth at the efficiency of the systems.
Generally, SISO communication seeks to eliminate effects of multipath propa-
gation to improve the quality of the communication link. Whereas MIMO-OFDM
wireless systems exploit the multipath propagation phenomenon to increase data
throughput and range, or reduce bit error rates. The MIMO-OFDM approach can
yield significant gains for both link and network capacities, with no additional en-
ergy or bandwidth consumption when compared to conventional single-array di-
versity methods [7, 9]. In a suitable rich scattering environment such as indoors,
MIMO-OFDM systems can introduce a significant capacity gain, due to the assur-
ance of multipath propagation. Temporal channel variations can occur as a result of
movement of personnel, industrial machinery, vehicles and other equipment mov-
ing within the indoor environment. Recent studies have been performed to simulate
pedestrian traffic effects on MIMO channels [4, 16, 26, 27]. However, a systematic
measurement campaign to study pedestrian movement effects in wideband MIMO-
OFDM channels has not yet been fully undertaken and suitable simulation tool is
not existing. As human bodies are a common factor in most indoor environments,
Chapter 1 Introduction
1.4 Objective/Goals 13
this thesis aims to characterize the effect of pedestrian movement in the indoor
MIMO-OFDM channel.
1.4 Objective/Goals
A fundamental issue in designing personal indoor wireless radio systems is gath-
ering knowledge about propagation characteristics over different types of environ-
ments and buildings. The experimental study that has been carried out in this thesis
aims to characterize the 5.2 GHz indoor MIMO-OFDM wireless channel in the
presence of pedestrians. This frequency has been chosen by recent indoor WLAN
standards, such as IEEE 802.11n.
Although a number of indoor wireless models have been proposed in the lit-
erature, the temporal channel variations due to human body shadowing effects in
indoor MIMO-OFDM channels have never been investigated. This thesis aims to
develop and validate a novel deterministic channel model, based on experimental
measurements, of human body effects on LOS MIMO-OFDM indoor propagation
channels at 5.24 GHz. This model will provide an insight into the characterization
of the significant human body effects in the indoor communication environment.
The objective is to provide a systematic characterization of the time varying effects
of the human body shadowing and scattering on indoor MIMO-OFDM channels.
The presence of a human body is highly realistic in an indoor environment and
can contribute to significant multipath generation. Hence, a noticeable variation in
channel capacity and dynamic range is expected. Such information is highly signif-
icant to the WLAN engineers. Moreover, an effective WLAN design can improve
quality of service as well as optimize bandwidth utilization.
This thesis will present two systematic measurement campaigns which are unique
to the best of our knowledge. Although several researchers report measurements re-
lating the indoor and outdoor channels, none of them characterize MIMO-OFDM
channels considering the presence of human bodies. The outcome of the thesis
Chapter 1 Introduction
1.5 Contribution 14
will provide researchers and WLAN designers with accurate channel information
to predict and design more efficient WLAN systems.
1.5 Contribution
The main contribution of this thesis is as follows;
1. Novel Deterministic model for a 5.24 GHz MIMO-OFDM system in presence
of pedestrian in indoor environments.
2. Systematic measurement campaigns for a 5.24 GHz MIMO-OFDM system
in presence of up to 10 pedestrians in indoor environments.
3. Characterization of measured 5.24 GHz MIMO-OFDM received power in
presence of up to 10 pedestrians in indoor environments.
4. Validation of proposed Deterministic model for indoor 5.24 GHz MIMO-
OFDM systems with measurement considering up to 10 pedestrians.
Additionally, a systematic study of the theoretical concept of MIMO-OFDM
channel capacity and channel capacity dynamic range has been presented. Exten-
sive literature review shows that the MIMO-OFDM system is likely to be the best
solution for the next generation of wireless communication. Although many re-
searchers have contributed to the MIMO channel modeling, very few incorporated
OFDM. Moreover, human body movement effects have never been fully investi-
gated. Our findings can contribute to the recent focus of building an effective
MIMO-OFDM channel model. We have presented simulated results considering
different human movement scenarios and validated these results by measurements.
Both results show similarity and describe human body effects for indoor environ-
ments. In addition, the implemented simulations also allow researchers to place
from none to a room full of human bodies (including a validation for intersecting
bodies) in dynamically adjustable room environments, considering all present ma-
terials.
Chapter 1 Introduction
1.6 Organization 15
1.6 Organization
The content of this thesis is organized as follows:
In Chapter 2, basics of SISO, MIMO, MIMO-OFDM, channel estimation and
channel capacity have been established. The theory behind the MIMO-OFDM anal-
ysis has been incorporated. A discussion on channel temporal variations is intro-
duced with theoretical interpretation. The importance of variation in channel char-
acteristics, due to the external effects such as fluorescent lights have been illustrated.
Chapter 3 presents a review of other researchers’ work in relation to the gradual
development of MIMO-OFDM systems. It also narrates work conducted in relation
to the MIMO-OFDM general concept, the history as well as introductory problems
of MIMO systems, with the the incorporation of OFDM. In addition, an analytical
review of pedestrian effects in indoor environments, with the reflection of other
investigators work has been presented. Several testbeds from around the world have
been detailed; as well, several modeling techniques relating to the MIMO-OFDM
system have been discussed in detail.
In chapter 4 and 5, detailed description of the measurements and simulation
techniques have been presented. Chapter 4 incorporates the measurement locations
description, measurement scenarios and setup. It also describes the detailed con-
figuration of the Commonwealth Scientific and Industrial Research Organization
(CSIRO) ICT Centre developed MIMO-OFDM channel sounder used in the exper-
iments. Chapter 5 focuses on specifications of the simulation technique that have
been used for this thesis. In this chapter, the simulation software, procedures and
scenarios considered for the investigation have been detailed.
Chapter 6 is dedicated entirely to the analysis of deterministic experimental sce-
narios. This chapter presents the measured and analyzed results of MIMO-OFDM
average channel capacity , capacity Cumulative Distribution Function (CDF) and
MIMO-OFDM capacity dynamic range. In addition, comparisons with simulations
have also been included and the results discussed and analyzed. Finally we have
compared the measured and simulated results and characterized the human body
Chapter 1 Introduction
1.6 Organization 16
effects for indoor environments.
Chapter 7, discusses the findings related to the random experimental scenarios.
This chapter depicts the measurement results for random human body movement in
an indoor environment ranging from none to ten. Simulated results have also been
presented in comparison with the measured findings, to validate the simulation tool.
Chapter 8 summarizes the outcome of this thesis and presents future research
directions.
Chapter 1 Introduction
17
Chapter 2
Theory and Background
This chapter aims to deliver background research and theory behind the SISO,
MIMO and MIMO-OFDM systems. A brief illustration of SISO, MIMO and single
/ multi carrier OFDM is also presented in this chapter.
2.1 SISO Single Carrier System and Channel Model-
ing
In a typical indoor environment, a single transmitter and a single receiver can be
placed in any location. For example an internet wireless modem can be positioned
in one end of the house and person with laptop can be at other end. Also two
laptops can be connected using a Wi-Fi connection from different locations in an
indoor environment. Additionally, Bluetooth devices can be utilized in conjunction
with laptop or PC to access certain utilities. Irrespective to the random mobility
and location of the antennas and/or objects in an indoor environment, radio signal
needs to be transmitted from the transmitter to the receiver. Due to the reflection,
refraction and scattering, the signal from the transmitter arrives at the receiver using
different propagated path, widely known as multipath. A typical SISO single carrier
system, with one transmitter and one receiver, transmits information over a single
data channel. This data transmission is subject to a complex, random and time-
Chapter 2 Theory and Background
2.1 SISO Single Carrier System and Channel Modeling 18
Figure 2.1: Mathematical Model of the Channel [2]
varying indoor radio propagation channel.
The indoor propagation channel can be modelled by considering each path of
the multipath in the three-dimensional space. In accordance to that, the channel is a
linear time-varying filter with the impulse response given by [2]:
h(t, τ) =
N(τ)−1∑
k=0
ak(t)δ[τ − τk(t)]ejθk(t)) (2.1)
where t is the observation time and τ is the application time of the impulse, N(τ)
is the multipath component, {ak(t)},{τk(t)}, {θk(t)} are the random time varying
amplitude, arrival time and phase sequences, δ is the delta function. The channel
can be completely characterized by these path variables.
Fig. 2.1 shows the wideband mathematical model of the channel. Due to the
generality of the model it can be used to obtain the response of the channels to
the transmission of any transmitted signal s(t) by convolution of s(t) with h(t)
and adding noise. To describe multipath fading channels Turin [28] suggested the
time invariant version of the signal propagation model. The proposed model has
been used successfully in mobile radio applications. For time invariant stationary
channel 2.1 reduces to:
h(t) =N−1∑
k=0
akδ(t− t(k))eiΘk . (2.2)
The output x(t) of the channel to a transmitted signal s(t) is therefore given by
x(t) =
∫ ∞
−∞s(τ)h(t− τ)dτ + n(t) (2.3)
Chapter 2 Theory and Background
2.2 MIMO Single Carrier System and Channel Modeling 19
where n(t) is additive Gaussian noise. From the depicted mathematical model, if
the signal s(t) = Re {s(t)exp[jω0t]} transmitted through this channel environment,
the received signal will be x(t) = <{ρ(t) exp[jω0t]} where
ρ(t) =N−1∑
k=0
aks(t− tk)ejθk + n(t). (2.4)
When a single unmodulated carrier (constant envelope) is transmitted in a multi-
path environment, due to vector addition of the individual multipath components, a
rapidly fluctuating Carrier Wave (CW) envelope can be experienced by a receiver in
motion. This phenomenon can be triggered by the vector addition of the individual
multipath components when single modulated carrier is transmitted in a multipath
environment. To avoid this narrow-band result from the model, we let s(t) of 2.4
equal to 1. Excluding noise, the resultant CW envelope R and phase φ for a single
point in space are thus given by
<φ =∞∑
k=0
akeθk . (2.5)
Through frequent sampling of the channel’s impulse response and using the wide-
band impulse response model, a narrow-band CW fading results for the receiver in
motion can be generated.
2.2 MIMO Single Carrier System and Channel Mod-
eling
The transmission over wireless links formed by multiple antennas equipped at both
the transmitter and receiver end is known as MIMO wireless communication. The
key advantages of employing multiple antennas lie in the more reliable performance
obtained through diversity and the higher data rate achievable through spatial mul-
tiplexing. Various schemes that employ multiple antennas at the transmitter and
receiver are being considered to improve the range and performance of communica-
Chapter 2 Theory and Background
2.2 MIMO Single Carrier System and Channel Modeling 20
Figure 2.2: A Schematic Representation of a MIMO Communication Scheme
tion systems. By far the most promising multiple antenna technology is recognized
as the MIMO system [29].
Fig. 2.2 depicts a schematic representation of a MIMO communication scheme,
which shows the distribution of antenna array in the MIMO systems at both the
transmitter and receiver end.
An important fact to note is that unlike traditional means of increasing throughput,
MIMO systems do not increase bandwidth in order to increase throughput. They
simply exploit the spatial dimension by increasing the number of unique spatial
paths between the transmitter and receiver. However, to ensure that the channel ma-
trix is invertible, MIMO systems require an environment rich in multipath [25, 30].
Fig. 2.3 is a diagram of a MIMO wireless transmission system. The transmitter and
receiver are equipped with multiple antenna elements. Coding, modulation, and
mapping of the signals onto the antennas may be realized jointly or separately.
Consider a wireless communication system with Nt transmitting (Tx) and Nr re-
ceiving (Rx) antennas. The idea is to transmit different streams of data on the differ-
ent transmitting antennas, but at the same carrier frequency. The stream on the i -th
transmitting antenna, as a function of the time t, will be denoted by si(t). When
a transmission occurs, the transmitted signal from the i -th Tx antenna might find
different paths to arrive at the j -th Rx antenna, namely, a direct path and indirect
Chapter 2 Theory and Background
2.2 MIMO Single Carrier System and Channel Modeling 21
Figure 2.3: Diagram of a MIMO Wireless Transmission System [3]
paths through a number of reflections. This principle is called multipath. Suppose
that the bandwidth B of the system is chosen such that the time delay between the
first and last arriving path at the receiver is considerably smaller than 1/B. In this
case the system is called a narrowband system. For such a system, all the multi-
path components between the i -th Tx and j -th Rx antenna can be summed up in
one term, say hij (t). Since the signals from all transmitting antennas are sent at the
same frequency, the j -th receiving antenna will not only receive signals from the
i -th, but from all Nt transmitters. This can be denoted by the following equation
xi(t) =Nt∑p=1
hij(t)si(t). (2.6)
To capture all Nr received signals into one equation, the matrix notation can be
used. With
s(t) =
s1(t)
s2(t)...
sNt(t)
,x(t) =
x1(t)
x2(t)...
sNr(t)
and H(t) =
h11(t) h12(t)) · · · h1Nt(t))
h21(t) h22(t)) · · · h2Nt(t))...
... . . . ...
hNr1(t) hNr2(t) · · · hNrNt(t)
(2.7)
this results in
x(t) = H(t)s(t) (2.8)
Mathematically, a MIMO transmission can be seen as a set of equations (the
recordings on each Rx antenna) with a number of unknowns (the transmitted sig-
nals). If every equation represents a unique combination of the unknown variables
Chapter 2 Theory and Background
2.2 MIMO Single Carrier System and Channel Modeling 22
and the number of equations is equal to the number of unknowns, then there ex-
ists a unique solution to the problem. If the number of equations is larger than the
number of unknowns, a solution can be found by performing a projection using the
least squares method [31], also known as the Zero Forcing (ZF) method. For the
symmetric case (i.e., Nt = Nr), the ZF solution results in the unique solution.
Suppose the coefficients of the unknowns are gathered in the channel matrix
H(t) and the number of parallel transmitting signals (unknown variables) equals
the number of received signals (equations), i.e., Nt = Nr, then the equations are
solvable when H(t) is invertible. Under this condition, the solution of equation 2.8
can be found by multiplying both sides with the inverse of H(t):
H−1(t)x(t) = H−1(t)H(t)s(t) = INts(t) = s(t), (2.9)
where IN is the N ×N dimensional identity matrix. Thus, to estimate the transmit-
ted signals at the receiver, the vector x(t) must be multiplied by the inverse of the
channel matrix H(t). To that end, the channel matrix must be known to the receiver.
This can be done by, e.g., sending a training sequence, that is known to the receiver,
to train the channel. A system with four transmitting antennas (Nt = 4) and four
receiving antennas (Nr = 4), or briefly, a 4 × 4 system is considered. It will be
assumed that the receiver perfectly knows the channel. With this assumption, we
may write the four equations s1(t),s2(t), s3(t) and s4(t) as
s1(t) = w1(t)x(t),
s2(t) = w2(t)x(t),
s3(t) = w3(t)x(t),
s4(t) = w4(t)x(t),
(2.10)
where wi(t) denotes the weight vector that is applied at the receiver to estimate the
i -th transmitted signal and can be shown to be equal to the i -th row of H−1(t).
As multiple data streams are transmitted in parallel from different antennas there
is a linear increase in throughput with every pair of antennas added to the system.
MIMO is a more significant change to radio architecture than any changes made in
Chapter 2 Theory and Background
2.3 SISO Multi-Carrier System and Channel Modeling 23
radio history so far. It is really quite simple in principle: through one transmitter,
some data from point A to point B can be transmitted. Now with four transmitters
or four carriers the likelihood of the data getting there will be increased by close to
four times, but the transmission will take up four times as much bandwidth. MIMO
takes those four independent OFDM carriers, all independently modulated, and puts
them on top of each other. Using this process MIMO generates four separate trans-
missions, all sharing the same frequency. MIMO systems can carry up to 4 times as
much information in the same bandwidth as a single carrier. OFDM, as discussed
in detail in the next section, uses a large number of carriers spaced apart at slightly
different frequencies. Although Frequency Division Multiplexing (FDM) implies
multiple data streams, OFDM carries only one data stream broken up into multiple
signals. Hundreds or thousands of carriers, known as sub-carriers are used for a
single data channel.
2.3 SISO Multi-Carrier System and Channel Model-
ing
OFDM is a popular modulation scheme that is used in wireless LAN standards like
802.11a, g, High Performance Radio LAN (HIPERLAN/2) and in the Digital Video
Broadcasting Terrestrial (DVB-T) . It is also used in the Asymmetric Digital Sub-
scriber Line (ADSL) standard, where it is referred to as Discrete Multitone modula-
tion. OFDM modulation divides a broadband channel into many parallel subchan-
nels. This makes it a very efficient scheme for transmission in multipath wireless
channels. The use of the Fast Fourier Transform / Inverse Fast Fourier Transform
(FFT/IFFT) pair for modulation and demodulation makes it computationally effi-
cient as well. FDM is a technology that transmits multiple signals simultaneously
over a single transmission path, such as a cable or wireless system. Each signal
travels within its own unique frequency range (carrier), which is modulated by the
data (text, voice, video, etc.). OFDM’s spread spectrum technique distributes the
Chapter 2 Theory and Background
2.3 SISO Multi-Carrier System and Channel Modeling 24
data over a large number of carriers that are spaced apart at precise frequencies.
This spacing provides the “orthogonality” in this technique, which prevents the de-
modulators from seeing frequencies other than their own. The benefits of OFDM
are high spectral efficiency, resiliency to Radio Frequency (RF) interference, and
lower multipath distortion. This is useful because in a typical terrestrial broadcast-
ing scenario there are multipath channels (i.e. the transmitted signal arrives at the
receiver using various paths of different length). Since multiple versions of the
signal interfere with each other ISI, it becomes very hard to extract the original in-
formation. OFDM offers very good spectral efficiency and is quite tolerant of the
ever-present interference in the bands where it is used. The popularity of OFDM
lies in its high data transmission capabilities with a low rate per symbol. Due to
its high rate transmission capability with high bandwidth efficiency and its robust-
ness with regard to multipath fading and delay, it has been used in Digital Audio
Broadcasting (DAB) systems, Digital Video Broadcasting (DVB) systems, Digital
Subscriber Line (DSL) standards, wireless LAN and as the core technique for the
fourth-generation (4G) wireless mobile communications.
(a) Signal Spectrum as Transmitted (b) Received Over a Dispersive, Time-Invariant
Channel
Figure 2.4: OFDM Signal Plot
Fig. 2.4(a) shows the signal spectrum, which consists of spectra of many bits
of an OFDM signal and Fig. 2.4(b) shows the received signal over a time invariant
channel. The orthogonal part of OFDM indicates a precise mathematical relation-
ship between the frequencies of the carriers in the system. In a normal FDM system,
the many carriers are spaced apart to receive signals using conventional filters and
Chapter 2 Theory and Background
2.3 SISO Multi-Carrier System and Channel Modeling 25
demodulators. In such receivers, guard bands have to be introduced between the dif-
ferent carriers, and the introduction of these guard bands in the frequency domain
results in a lowering of the spectrum efficiency. In an OFDM signal the sidebands
of the individual carriers overlap and the signals can still be received without the
interference of an adjacent carrier. To achieve this, the carriers must be mathemat-
ically orthogonal. The receiver acts as a bank of demodulators, translating each
carrier down to Direct Current (DC), the resulting signal then being integrated over
a symbol period to recover the raw data. If all the carrier frequencies in the time
domain have a whole number of cycles in the symbol period (t), then the integra-
tion process results in zero contribution from all these carriers. Thus the carriers
are linearly independent (i.e. orthogonal) if the carrier spacing is a multiple of 1/t .
Now, suppose a set of signals Ψ, where Ψp is the p-th element in the set. The signal
will be orthogonal when
∫ a
b
Ψp(t)Ψ∗q(t)dt =
K for p = q
0 for p 6= q
(2.11)
where ∗ indicates the complex conjugate and interval [a, b] is a symbol period.
OFDM is essentially a discrete implementation of multi-carrier modulation,
which divides the transmitted bit stream into many different sub-streams and sends
them over many different sub-channels. Typically, the sub-channels are orthogonal
and the number of sub-channels are chosen such that each sub-channel has a band-
width much less than the coherence bandwidth 1of the channel. Thus, ISI on each
sub-channel is very small. For this reason, OFDM is widely used in many high data
rate wireless systems. MIMO-OFDM combines OFDM and MIMO techniques,
thereby achieving spectral efficiency and increased throughput. A MIMO-OFDM
system transmits independent OFDM modulated data from multiple antennas si-
multaneously. At the receiver, after OFDM demodulation, MIMO decoding on each
of the sub-channels extracts the data from all the transmitting antennas on all the
1Coherence Bandwidth is a statistical measurement of the range of frequencies over which the
channel can be considered “flat ”
Chapter 2 Theory and Background
2.3 SISO Multi-Carrier System and Channel Modeling 26
sub-channels.
(a) Single subchannel (b) Combination of Five Subchannels
Figure 2.5: OFDM spectrum
Fig. 2.5(a) depicts a spectrum of a single OFDM subchannel and Fig. 2.5(b)
shows the OFDM spectrum with multiple subchannels. OFDM transmits a large
number of closely spaced narrowband carriers in the frequency domain. Digital
signal processing techniques such as FFT are often implemented in current telecom-
munication systems. This allows avoidance of a large number of modulators, filters
at the transmitter and complementary filters and demodulators at the receiver. Each
carrier can be mathematically described as a complex wave:
SNt(t) = ANt(t)ej[ωNt t+φNt (t)] (2.12)
Here ANt(t) is the amplitude of the signal SNt(t) , ej[ωNt t+φNt (t)] is the phase of
the carrier and t is the symbol duration period. OFDM consists of the combination
of many single carrier Ss(t), which can be written as,
SsNt(t) =1
Nt
Nt−1∑n=0
An(t)ej[ωnt+φn(t)] (2.13)
where ωn = ω0 + n∆ω . By considering the waveform of each component
of the signal over one symbol period, the variables take on fixed values and that
Chapter 2 Theory and Background
2.3 SISO Multi-Carrier System and Channel Modeling 27
depends on the frequency of the particular carrier. This can be expressed as φn(t) ⇒φn, An(t) ⇒ An. The resulting signal using a sampling frequency of 1/T can be
represented by:
SsNt(kT ) =1
Nt
Nt−1∑n=0
Anej[(ω0+n∆ω)kT+φn] (2.14)
At this point, the received signal can be expressed as a digital signal of N sam-
ples. It is convenient to sample over the period of one data symbol. Therefore,
t = NtT By simplifying 2.14, the signal becomes:
SsNt(kT ) =1
Nt
Nt−1∑n=0
Anejφnej(n∆ω)kT (2.15)
Now the resulting 2.15 can be compared with the general form of the inverse
Fourier transform:
g(kT ) =1
Nt
Nt−1∑n=0
G[
nNtT
]ej2πrk/Nt (2.16)
2.15 and 2.16 are equivalent if ∆f = ∆ω2π
= 1NtT
= 1t. By maintaining orthog-
onality an OFDM signal can be defined by using the Fourier transform. The use of
Differential Phase-Shift Keying (DPSK) in OFDM systems avoids the need to track
a time varying channel. However, it limits the number of bits per symbol and re-
sults in a 3 dB loss in Signal-to-Noise Ratio (SNR) compare to coherent modulation.
Coherent modulation allows arbitrary signal constellations, but efficient channel es-
timation strategies are required for coherent detection and decoding. The basic idea
underlying OFDM systems is the division of the available frequency spectrum into
several sub-carriers. To obtain a high spectral efficiency, the frequency responses
of the sub-carriers are overlapping and orthogonal, hence the name OFDM. This
orthogonality can be completely maintained at the small price of a loss in SNR.
Chapter 2 Theory and Background
2.4 MIMO Multi Carrier System and Channel Modeling 28
2.4 MIMO Multi Carrier System and Channel Mod-
eling
MIMO-OFDM systems are mainly composed of channel coding, OFDM modula-
tion, OFDM demodulation and channel decoding. The exact channel estimation is
the key to make MIMO-OFDM reach the decoding performance, the channel esti-
mation needs two dimensional (2D) estimation both at time domain and frequency
domain. If the antenna array is Nt×Nr, then comparing with a single carrier system,
the complexity of channel estimation is the Nt×Nr times of it. OFDM modulation
may cause the difficulties and complexities of channel estimation to increase more
and more. The purpose of channel estimation is to obtain the pulse response of the
channel in frequency domain or time domain. In order to trace the changed chan-
nel, some sub-channel is selected to transmit guided frequency in a MIMO-OFDM
system. The guided frequency message is a set of sequences known at the receiver.
The fading of sub-channels can be obtained by comparing the received guider with
the known sequence at the receiver. The channel parameters for the entire frequency
band can be found by using the response at sub-channel level and a tracing algo-
rithm [32]. With increasing high data access requirement the key challenge faced by
future wireless communication systems is to maintain the QoS in addition to high
speed data requirement. MIMO wireless technology seems to meet these demands
by offering increased spectral efficiency through spatial multiplexing gain, and im-
proved link reliability, due to antenna diversity gain. At one end of the wireless
links multiple antennas have been used to perform interference cancelation and to
realize diversity and array gain through coherent combining. The use of multiple
antennas at both ends of the link offers spatial multiplexing gain, which results in
increased spectral efficiency. The number of transmitting and receiving antennas
spatial multiplexing yields a linear capacity increase, compared to systems with a
single antenna at one or both sides of the link, at no additional power or band-
width expenditure. The corresponding gain is available if the propagation channel
Chapter 2 Theory and Background
2.4 MIMO Multi Carrier System and Channel Modeling 29
exhibits rich scattering phenomenon. The gain can also be realized by the simulta-
neous transmission of independent data streams in the same frequency band. The
receiver exploits differences in the spatial signatures induced by the MIMO channel
onto the multiplexed data streams to separate the different signals, hence realizing
a capacity gain.
Diversity leads to improved link reliability by rendering the channel less fad-
ing and by increasing the robustness to co-channel interference. Diversity gain is
obtained by transmitting the data signal over multiple independently fading dimen-
sions in time, frequency and space. In addition, the gain also depends on the proper
combination of the above factors in the receiver end. Spatial (i.e., antenna) diver-
sity is particularly attractive when compared to time or frequency diversity, due to
the fact that this does not incur an expenditure in transmission time, or bandwidth,
respectively. Space-time coding [33] realizes spatial diversity gain in systems with
multiple transmitting antennas without requiring channel knowledge at the transmit-
ter. Array gain can be realized both at the transmitter and the receiver. It requires,
channel knowledge for coherent combining and results in an increase in average re-
ceive SNR and hence improved coverage. Multiple antennas at one or both sides of
the wireless link can be used to cancel or reduce co-channel interference, and hence
improve cellular system capacity [19, 34–36].
In 2.1 we have derived the impulse response which is assumed to be constant
over the duration of one OFDM symbol. For an OFDM system with N carriers,
the frequency response on sub carrier k, h[k], is found by calculating the Fourier
transform of 2.2 at normalized frequency f = fk = k/N [37].
h[k] =N−1∑
l=0
M∑n=1
αnej2πl kN
τn (2.17)
Consider a 4× 4 antenna system using OFDM, where an OFDM sysmbol con-
sists of N samples and the length of the Cyclic Prefix (CP) is set to L samples. The
system thus consists of sixteen different antenna-to-antenna channels, all treated as
independent and linear. The received signal at antenna j rj[k], where k = 1, ...., N ,
Chapter 2 Theory and Background
2.5 Channel Temporal Variation 30
is a linear combination of the transmitted signals. In the frequency domain this can
be expressed as [38]
rj[k] =
NT∑i=1
hij[k].si[k] + wj[k] (2.18)
Where NT is the number of transmit antennas, hij[k] the channel frequency
response of the k-th sub-channel between the i-th transmitter and j-th receiver an-
tenna, si[k]is the signal from the i-th transmitting antenna and wj[k] is the noise
on the j-th receiver branch, which here will be treated as complex white Gaussian
noise with zero mean and variance σ2n.
2.5 Channel Temporal Variation
Due to moving pedestrians in most indoor environments, the channel is non-stationary
in time; i.e., there is a significant change in the channel’s characteristics even with
Fixed transmitter and receiver. This is reflected in a time-varying filter model, how-
ever the analysis of this time varying filter model is very difficult [2]. Most prop-
agation measurements that utilize digital signal processing have therefore assumed
some form of stationarity while collecting the impulse response profiles.
Fig. 2.6 represents an example of CW temporal envelope fading [2]. In Fig. 2.6
(a) the immediate environment of the receiver was clear of motion for the first 20
seconds, while motion occurred after 20 seconds. Fig. 2.6(b) corresponds to a mea-
surement during which there was constant motion in the vicinity of the receiver
throughout the measurement period of 30 seconds. Examination of these figures
reveals great variations in the signal level, even though both antennas are stationary.
Deep fades of up to 20 dB below the mean value can be observed in these figures.
This is due to constructive and destructive combination of multipath.
The indoor channel temporal variation has been studied extensively by differ-
ent researchers [4, 16, 26]. To avoid distortions caused by the motion of people
and equipment, a number of indoor measurements have been carried out at night
Chapter 2 Theory and Background
2.5 Channel Temporal Variation 31
Figure 2.6: Temporal CW Envelope Fading for a Medium Size Office Building.
Carrier frequency is 915 MHz and both antennas were stationary during the mea-
surements. (a) Antenna separation 10 m; (b) antenna separation 20 m. (Measure-
ments and processing by David Tholl of TRLabs.)[2][nsec=second]
Chapter 2 Theory and Background
2.5 Channel Temporal Variation 32
or during the weekends. A major conclusion is that for office buildings, where
the environment is divided into separate rooms, fading normally occurs in “bursts”
lasting tens of seconds, with a dynamic range of about 30 dB [39]. Additionally,
for open office environments, however, fading is rather continuous, with a dynamic
range of 17 dB [39]. Extensive CW measurements around 1 GHz in five factory
environments [40, 41] and office buildings [39] have shown that even in the absence
of a direct LOS path between the transmitter and receiver, the temporal fading data
show a good fit to the Rician distribution. Another work reporting measurements at
60 GHz, however, indicates that with no LOS path the CW envelope distribution is
nearly Rayleigh. A measure of the channel’s temporal variation is the width of its
spectrum when a single sinusoid (constant envelop) is transmitted. This has been
estimated to be about 4 Hz [39] for an office building. A maximum value of 6.1 Hz
has also been reported [42, 43].
In this thesis, we defined three different time scales in analysing temporal vari-
ation of indoor channels. They are small time scale, medium time scale and large
time scale.
We loosely define that the period of the small time scale is approximately less
than 1 second and the time resolution should be less than 1/20 seconds. Effects of
fluorescent lights and fans can be considered within this small time scale. Radio
waves reflected from active fluorescent light tubes are modulated at twice the fre-
quency of the power network, which causes fast temporal variation of the channel
[44–46]. The magnitude of this fading is related to the ratio of the power of the
sum of signal components following paths reflected by the fluorescent light tubes to
the total received signal power, and is highly dependent on the local geometry and
exact location of the antennas [44].
Similarly, we loosely define the period of medium time scale as approximately
less than 5 minutes and the time resolution should be less than 1 second. Effects
of pedestrian and industry equipment movement can be considered within this time
scale. Several researchers have reported the effect of pedestrians in indoor environ-
Chapter 2 Theory and Background
2.6 MIMO-OFDM Channel Capacity 33
ment [4, 16, 26, 47, 48] whose details are reviewed in Chapter 3. This thesis focuses
on the medium time scale statistics.
The large time scale considers the temporal variation of the channel in the longer
term, such as diurnal or even seasonal. The channel statistics may be quite different
between the day time and the night time, due to the differences in human activities.
2.6 MIMO-OFDM Channel Capacity
The channel capacity is the tightest upper limit on the amount of information that
can be reliably transmitted over a communications channel. It also represents a
given channel’s limiting information rate which can be achieved with arbitrarily
small error probability. The channel capacity is a measure of channel availability
or goodness, the larger the capacity value the more information that can be sent
reliably by the system at a higher data rate.
The MIMO-OFDM channel capacity without the knowledge of the channel at
the transmitter is given by [15]
C =1
nf
nf∑
k=1
nt∑j=1
log2(1 +ργj(fk)
nt
), (2.19)
where C is the normalized capacity in bits/sec/Hz, nf is the number of OFDM
sub-carriers, nt is the number of Tx antennas, ρ is the average SNR and γj is the
eigenvalue of H(fk)H(fk)H . H(fk) is the normalized channel coefficient matrix
at sub-carrier fk and (.)H denotes Hermitian transpose. The normalization is per-
formed such that [49]
E(||H||2F
)= ntnr, (2.20)
where E(·) denotes the expected value, || · ||F denotes the Frobenius norm, and nr
is the number of Rx antennas.
Two different criteria are employed to evaluate the MIMO-OFDM channel ca-
pacity. The first assumes an interference limited system, where transmitting power
Chapter 2 Theory and Background
2.6 MIMO-OFDM Channel Capacity 34
can be adjusted without a limit, to provide a fixed average SNR at the receivers. The
averaging of SNR and normalization of channel coefficient matrix is performed over
all MIMO sub-channels and over all OFDM sub-carriers. This criterion is called
Fixed SNR capacity. It corresponds to the system where co-channel interference is
the limiting factor for the system capacity, and enough Tx power is reserved to cater
for every location within area of coverage . SNR=15 dB is used in the following
analysis. The second criterion assumes a power limited system where the transmit-
ting power is Fixed. In this case the averaging of SNR and normalization of the
channel coefficient matrix is performed over all MIMO sub-channels, OFDM sub-
carriers, measurement samples, and different numbers of pedestrians. This is called
Fixed Tx power capacity. It incorporates the effects of the reduction of power due
to body shadowing by the pedestrian. This criterion is more suitable for the analysis
of the WLAN system where the transmitting power is typically Fixed.
Fig. 2.7 shows an example of temporal variation of the measured Fixed SNR
4×4 MIMO-OFDM channel capacity when one pedestrian is crossing the direct
LOS path. Details of the experimental setting is given in Chapter 4. For Fixed SNR
due to the rise in transmission power, the capacity increases as a person cross the
direct LOS path in populated indoor environment.
In [16], capacity dynamic range has been defined as the difference between the
maximum and the minimum value of the MIMO channel capacity. In this thesis, we
define 90% capacity dynamic range, which is the difference between the top 95%
and the bottom 5% values, in order to remove extreme cases. 90% capacity dynamic
range can also be defined from the CDF of the MIMO-OFDM channel capacity as
shown in Fig. 2.8, which corresponds to the results presented in Fig. 2.7. It may be
expected that no temporal channel variation can be observed when no pedestrian is
inside the room. However, a sample measurement as described in Chapter 4 showed
that the measured 4×4 Fixed SNR capacity dynamic range without pedestrians was
0.16 bits/sec/Hz. The main cause of this slight variation in channel capacity is
considered to be due to disturbances outside the room at the time of measurements.
Chapter 2 Theory and Background
2.6 MIMO-OFDM Channel Capacity 35
0 20 40 60 80 10013
13.5
14
14.5
15
15.5
Sample index
MIM
O−
OF
DM
cha
nnel
cap
acity
(bi
ts/s
/Hz)
Figure 2.7: Example of Measured Temporal Variation of 4×4 MIMO-OFDM Fixed
SNR Channel Capacity with 1 Pedestrian (2 samples per second).
In this thesis disturbances outside the room have been ignored as they do not prove
to be significant when analyzing the experimental data.
The analysis of channel capacity dynamic range shows the severity of channel
temporal variation. The higher the severity of the channel temporal variation, it is
generally more difficult for a system to adapt. Understanding of the channel ca-
pacity dynamic range can significantly contribute to the system engineers who can
utilize and handle the channel capacity dynamic range information to improve the
indoor WLAN performance. To obtain a signified data range as well as due to the
quality enhancement an outage capacity of 5% has been introduced. Through this
process the extreme case scenarios have been removed, which can introduce poten-
tial variation in the channel estimation process. Besides, these high variations are
not a regular phenomenon in the transmission process, but happen to be introduced
now and then by the sudden movement of the existing structural movement or other
external noise such as (lighting, high voltage equipments, mobile devices in use).
Consideration of such disturbance in the real life environment has been eliminated
as much as possible to imitate the experimental scenarios to simulation environ-
ment. Either the system engineer will design a system that can adapt to a larger
Chapter 2 Theory and Background
2.7 Summary 36
12 13 14 15 160
20
40
60
80
100
MIMO−OFDM channel capacity (bits/s/Hz)
CD
F (
%)
5%
95%
Dynamic Range
Figure 2.8: Sample of Measured CDF of 4×4 MIMO-OFDM Fixed SNR Channel
Capacity with 1 Pedestrian.
dynamic range or will create a system which will reduce the dynamic range in order
to improve the quality of the communication channel.
2.7 Summary
Theoretical background and the fundamental mathematical concepts relating to the
MIMO-OFDM system have been presented in this chapter in a step by step man-
ner starting with SISO systems. This was followed by a detailed description of the
most significant principles of MIMO, OFDM and combined MIMO-OFDM sys-
tems. These significant basics are crucial for understanding MIMO-OFDM channel
characteristics and its modeling. An extensive literature review considering all the
discussed theory and a comprehensive review of other researchers’ work is pre-
sented in Chapter 3.
Chapter 2 Theory and Background
37
Chapter 3
Literature Review
This chapter presents an overview of significant theoretical concepts of MIMO-
OFDM systems and recently reported work on MIMO-OFDM channel characteri-
zation. It also describes previous research related to the pedestrian effects on indoor
MIMO wireless channels.
3.1 MIMO-OFDM System
In recent years, the requirement for high data-rate wireless access is growing in
many applications. Traditionally, more frequency bandwidth has been deployed to
achieve the required higher data-rate transmission. However, due to spectral limi-
tations, techniques that use more frequency bandwidth for increasing data rate are
often impractical and/or expensive. The prospect of many orders of magnitude im-
provement in wireless communication performance at no cost for the extra spectrum
is largely responsible for the success of MIMO. The MIMO approach can yield sig-
nificant gains for both link and network capacities, with no additional energy or
frequency bandwidth consumption when compared to conventional single-array di-
versity methods [15, 50, 51]. This has prompted progress in areas as diverse as
channel modeling, information theory and coding, signal processing, antenna de-
sign and multi-antenna-aware cellular design, Fixed or mobile. MIMO-OFDM is
Chapter 3 Literature Review
3.1 MIMO-OFDM System 38
currently considered to be one of the most spectrally efficient techniques [7, 9]. In
this section we have reviewed the general concept of the MIMO-OFDM system, its
background implementation, as well as the MIMO-OFDM system in practice.
3.1.1 MIMO-OFDM General Concept
Many researchers have conducted research on the MIMO system [52–54].
Different approaches to exploit the MIMO capacity strongly rely on adequate
coding and signal processing [7]. However, the actual attainable capacity of the
system depends on the channel conditions and antenna arrays used. The first chan-
nel model introduced was the independent, identically distributed (i.i.d) Rayleigh
model [7], which is an extension of the SISO model. This model represents a high
scattering environment, with many equal power signal paths arriving at the receiver
from all directions. The Rayleigh i.i.d. model results in an upper bound for the
capacity of a MIMO system [53]. Authors [53] also reported several measurement
campaigns carried out by other researchers to characterize the MIMO channel, in-
cluding the effects of temporal variation, compact arrays, antenna correlation, and
polarization diversity. The research work of several experts on the correlation anal-
ysis of MIMO channels has also been reviewed [55, 56]. MIMO communications
channels provide an interesting solution to the multipath challenge by requiring
multiple signal paths. In effect, MIMO systems use a combination of multiple an-
tennas and multiple signal paths to gain knowledge of the communications chan-
nel. By using the spatial dimension of a communications link, MIMO systems can
achieve significantly higher data rates than traditional SISO channels. Tutorials
on the overview of the work done in the area of MIMO systems and the opera-
tion of MIMO wireless communication systems have been presented [23, 24], as
well as illustrating how multiple antennas can lead to increased system capacity for
multipath communication channels. The authors focus on channel capacity com-
putation, channel measurement and modeling approaches, the impact of antenna
element properties and array configuration on system performance.
Chapter 3 Literature Review
3.1 MIMO-OFDM System 39
The presence of multipath greatly improves the achievable data rate if the ap-
propriate communication structure is employed [7, 8]. The signal transmitted from
the transmitter reaches the receivers via one or more main waves/paths. These main
waves consist of a LOS ray and several rays reflected or scattered by main struc-
tures such as outer walls, floors or ceilings. The LOS wave may be attenuated by
the intervening structure to an extent that makes it undetectable. The main waves
are random upon arrival in the local area of the receiver. They break up in the en-
vironment of the receiver due to scattering by local structure and furniture. The
resulting paths for each main wave arrive with very minor delays, experience about
the same attenuation, but have different phase values due to different path lengths.
The individual multipath components are added according to their relative arrival
times, amplitudes and phases, and their random envelop sum is observed by the re-
ceiver. The number of distinguishable paths recorded in a given measurement, and
at a given point in space, depends on the shape and structure of the building, and on
the resolution of the measurement setup [2].
While the multipath improves MIMO performance, the multipath also causes
the problem of ISI among transmitted symbols on wide-band wireless channels. An
OFDM system operating over a wireless communication channel effectively forms
a number of parallel frequency-nonselective fading channels, thereby obviating the
need for complex equalization and thus greatly simplifying equalization/decoding
[57].
Multiple antennas are useful in OFDM systems for providing transmit and re-
ceive diversity to overcome fading [58, 59]. The OFDM system introduces a guard
interval to avoid ISI [60]. However, an equalizer is required to compensate for fad-
ing distortion even if the delay of the multipath channel is shorter than the length of
guard interval. The OFDM gains computational efficiency using FFT in modulation
and demodulation.
In addition, multiple antenna system designs require considerable separation
between the antennas. Spatial correlation is introduced when antennas are not well
Chapter 3 Literature Review
3.1 MIMO-OFDM System 40
separated, and it often leads to performance degradation in a flat fading environ-
ment. However, in frequency selective fading channels with rich multipath diver-
sity, OFDM receivers can overcome this performance degradation due to antenna
correlation. This is due to transformation of a highly spatially correlated channel
impulse response to a less spatially correlated channel frequency response, inher-
ently by an OFDM system in the presence of rich multipath diversity. Moreover,
OFDM distributes the data over a large number of carriers that are spaced apart at
precise frequencies. This spacing provides the “orthogonality ”in this technique,
which prevents the demodulators from seeing frequencies other than their own. The
benefits of OFDM are high spectral efficiency, resiliency to RF interference, and
lower multipath distortion. This is useful because in a typical indoor environment
there are multipath-channels.
Many other researchers have investigated different forms of ISI solution incor-
porating OFDM, and agreed with the fact that MIMO-OFDM is a strong candidate
for the physical layer transmission scheme of next generation broadband wireless
communication systems [61–65].
For such improvement and quality assurance, researchers started focusing on
the MIMO-OFDM combination systems. The paper [51] is one of the earliest pa-
pers where authors reported the MIMO-OFDM as an emerging technology for up
coming 1-Gb/s wireless links.
Authors in [62] present a MIMO-OFDM physical prototype and have analyzed
the system performance and demonstrated high bandwidth efficiency in different
environmental and structural conditions typical of practical wireless networks.
This leads to further investigations of MIMO-OFDM in terms of quality assur-
ance for high data rate and improved modeling of the channels and sub channels
in MIMO-OFDM systems. In addition, researchers have conducted investigations
combining different modulation techniques [66, 67] such as Bit Interleaved Coded
Modulation (BICM) as well as channel codes [68] to reduce bit error probability us-
ing a MIMO-OFDM system. Investigation on performance analysis and design op-
Chapter 3 Literature Review
3.1 MIMO-OFDM System 41
timization of Low-Density Parity Check (LDPC) coded MIMO-OFDM systems for
high data rate wireless transmission have been conducted [69]. The tools of density
evolution with mixed Gaussian approximations are used to optimize irregular LDPC
codes and to compute minimum operational SNRs for ergodic MIMO-OFDM chan-
nels. In particular, the optimization is done for various MIMO-OFDM system con-
figurations, which include a different number of antennas, different channel mod-
els, and different demodulation schemes. The optimized performance is compared
with the corresponding channel capacity. These researches show that the proposed
MIMO-OFDM system have been investigated in many different areas, to point out
the fact of its reliability and effectiveness as most promising. One of the main
points of interest while conducting research relating MIMO-OFDM is to measure
the actual propagation channels and characterize them for different environments.
Channel estimation is critical in obtaining practical channel information. The
investigation of MIMO-OFDM channel characteristics is also highly important for
the system performance improvement. Analysis and modeling of practical MIMO-
OFDM channels for designing and optimizing a transmission scheme depends on
accurate channel estimation. Generally channel estimation is to estimate and char-
acterize a channel through a received signal, which can be affected by modulation,
propagation, environment, Time Of Arrival (TOA)s , Phase Noise (PN) , Angle Of
Arrival (AOA) , Angle Of Departure (AOD) , frequency offset. Investigation of a
channel estimation scheme based on TOAs has been carried out [38, 58]. The re-
searcher also conducted channel estimation investigation considering a combination
of effects due to frequency offset and phase noise, and a robust Mean-Square Error
(MSE) optimal training signal designs was developed and reported [70].
For proper channel estimation, channel measurement plays a vital role in terms
of accuracy and reliability of the predicted model. Research on MIMO-OFDM
channel measurements was performed in indoor environments based on extensive
measurement results (4 × 4 MIMO sub-channels, 117 OFDM sub-carriers, and
6400 receiving antenna array locations per local area) [17]. In this investigation,
Chapter 3 Literature Review
3.1 MIMO-OFDM System 42
the channel fading characteristic is analyzed both in space and in frequency. Area
plots reveal how the MIMO-OFDM channel capacity is distributed within a local
area in different environments. Results show that MIMO-OFDM channel capac-
ity is strongly dependent on the local scattering environment in the case of a LOS
situation, while it is less affected in the case of a NLOS situation. In addition,
MIMO-OFDM channel capacity in a local area is strongly dependent on the local
scattering environment in the case of a LOS situation, while it is less affected in the
case of NLOS situations.
In this thesis, the MIMO-OFDM channel estimation technique as reported in
[17] is used to obtained MIMO-OFDM channels in indoor environments.
3.1.2 MIMO-OFDM History
OFDM had been proposed in 1966 [71], with a principle of orthogonal multiplexing
for transmitting a number of data message simultaneously through a liner band lim-
ited transmission medium at a maximum data rate without interchannel and inter-
symble interference. As OFDM becomes a popular technique for transmitting sig-
nals over wireless channels, it has been adopted in several wireless standards such
as DAB , DVB-T , the IEEE 802.11a [72] LAN standard and the IEEE 802.16a [73]
Metropolitan Area Network (MAN) standard. Many researchers have emphasized
the OFDM as a popular method for high data rate wireless transmission [74, 75].
Combining OFDM with antenna arrays at the transmitter and receiver end can in-
troduce the diversity gain and/or enhance the system capacity on time-variant and
frequency-selective channels, resulting in a MIMO configuration [74].
The concept of spatial multiplexing using MIMO technology had been proposed
as early as 1994 [76]. A refined new approach was presented in 1996, which consid-
ers a configuration where multiple transmit antennas and the new communication
structure, termed the layered space-time architecture, targets application in future
generations of Fixed wireless systems, bringing high bit rates to the office and home
[77]. Although multipath improves MIMO performance, it also causes the problem
Chapter 3 Literature Review
3.1 MIMO-OFDM System 43
of ISI among transmitted symbols on wide-band wireless channels [57]. To im-
prove and avoid such complexity, as well as handle the multipath channel more
efficiently, OFDM modulation combines with the MIMO system. This is known as
MIMO-OFDM [57].
New communications standards are using MIMO-OFDM to maximize through-
put and coverage, while preserving bandwidth [72, 73]. The following section fo-
cuses on the MIMO-OFDM in practice, for real world communication solutions.
3.1.3 MIMO-OFDM in Practice
MIMO-OFDM systems are used in modern wireless standards, including in IEEE
802.11n [11, 78], Long Term Evolution (LTE) [74] and mobile Worldwide Inter-
operability for Microwave Access (WiMAX) systems [79]. Moreover, to fully sup-
port cellular environments, MIMO research consortia including Information Society
Technologies - Multiple Access Space Time Code (IST-MASCOT) propose to de-
velop advanced MIMO techniques, i.e., Multi-User MIMO (MU-MIMO) [80]. The
technique supports enhanced data throughput even under conditions of interference,
signal fading, and multipath, and is applicable to MIMO-OFDM.
In recent years MIMO-OFDM is attracting significant interest on implemen-
tation of services for WLAN. Authors in [78] presented a prototype focusing on
WLAN and demonstrated that very high data rates in excess of 200 Mb/s over wire-
less are feasible. According to [78], MIMO-OFDM, providing high data rates, can
simplify Quality of Service (QoS) handling, such as scheduled operation with two
priority classes.
Several reasons have been identified for WLAN data rates which were outpac-
ing the data rates available in Third Generation (3G) WANs. WLANs use wider
bandwidths (typically 20 MHz) that are available in unlicensed bands. Also, the
low mobility and limited range requirements as well as the need to combat smaller
delay spread simplify some aspects of system design. Furthermore, the design and
implementation of a high-performance next-generation WLAN had been discussed
Chapter 3 Literature Review
3.1 MIMO-OFDM System 44
and it was standarized as IEEE 802.11n [11].
In [81] authors presented a novel synchronization architecture for a 2×2 MIMO-
OFDM WLAN system. Carrier synchronization can be achieved in several ways
and most synchronization methods utilize a Digital Signal Processing (DSP) pro-
cessor, but these consume significant power. In this research, authors adopts the Co-
ordinate Rotation Digital Computer (CORDIC) algorithm to manage the timing and
carrier synchronization efficiently with, a precise digital oscillator and a re-modified
Booth multiplier [81]. In this thesis, the carrier synchronization is achieved by ac-
curate Rubidium frequency references employed at the transmitter and the receiver.
As the MIMO-OFDM system appears to be a promising solution for the phys-
ical layer of indoor multimedia transmission via WLANs, antenna selection is an
excellent way of reducing the hardware costs of MIMO-OFDM systems, while re-
taining high performance [82]. Authors in [82] address two major practical con-
cerns for the application of antenna selection. a) antenna selection training protocol
design and b) calibration to solve RF imbalance. Authors presented novel solu-
tions that are especially suitable for slowly time-varying environments, e.g., indoor
scenarios, sports stadiums, and shopping malls. Specifically, a low Doppler spread
associated with such environments enables us to train all antenna subsets by multi-
ple training packets transmitted in bursts. Both numerical and analytical approaches
are used to verify the effectiveness of the proposed solutions, which make antenna
selection more easily adaptable for high-throughput WLAN systems. The proposed
techniques move a step closer to the practical implementation of MIMO antenna se-
lection techniques in high throughput WLAN systems, especially for indoor multi-
media applications. In this thesis, the antenna selection is not considered. However,
the measurement and simulation results presented in this thesis can be utilized to
analyze antenna selection algorithms and their performance selecting antennas up
to four antennas (e.g. selecting two antennas from four antennas).
Recent research investigated the suitability of a range of MIMO-OFDM archi-
tectures for use in urban hotspots [83]. In [83], a ray-tracing propagation model was
Chapter 3 Literature Review
3.1 MIMO-OFDM System 45
used to produce realistic MIMO-OFDM channel data. This information was used
to determine the expected throughput and area coverage for various physical (PHY)
layer schemes. Site-specific throughput predictions were generated in a city centre
environment. Link Adaption (LA) was shown to play a key role in the choice of
spacetime algorithm, the use of adaptive modulation and coding, and the number
of antennas employed at both ends of the radio link. The combination of area-wide
link-level simulations using MIMO-OFDM channel data from an urban ray-tracing
propagation model provided unique insights into system performance.
[83] confirmed that 87.83% of locations were covered with a data rate of 108
Mbits/s or better, with an average SNR of 27dB and with frequency bandwidth of
40 MHz using 4× 4 transmitters and receivers.
In [79] performance was evaluated for the various MIMO-OFDM WiMAX sys-
tems. The per-tone signal-to-noise plus interference ratio of a MIMO-OFDM sys-
tem was derived in a multi user, multicell and multisector communication system.
The sector average spectral efficiency for a coded MIMO-OFDM system was evalu-
ated under a single frequency re-use deployment scenario. It was shown that the sec-
ond antenna at the subscriber station receiver provided significant gains over SISO
systems. Spatial multiplexing MIMO scheme on the downlink improved the sector
spectral efficiency by 10% over single transmit and two receive antenna (SIMO)
systems in a single frequency re-use deployment.
From the presented discussion, it has been clearly identified that the MIMO-
OFDM is continually being incorporated into the latest wireless communication
media. With the increasing demand of high date rate wireless connection in the
WLAN and WiMAX, researchers from diverse arenas are focusing on a solution
which can provide increased throughput with the existing frequency bandwidth limit
[3, 63, 84]. Despite a strong link improvement possibility through MIMO-OFDM
deployment in indoor environments, an important factor, the human body, has not
been considered for systematic investigation. A few researchers have contributed in
this sector ([4, 17, 22, 26], but to the best of author’s knowledge none has considered
Chapter 3 Literature Review
3.2 Pedestrians and the Indoor Channel 46
a measurement campaign with verification through customized simulation. While
considering indoor environments, for a MIMO-OFDM system, multipath propa-
gation plays a key role in reliable wireless communication [85]. Due to the fact
that MIMO-OFDM can offer significant bandwidth efficiency in broadband wire-
less applications, an increasing interest in the study of MIMO systems in multipath
environments has been observed in the last few years [15, 34, 86]. The following
section provides a review of the multipath effects in indoor environments, followed
by discussion on pedestrian effects.
3.2 Pedestrians and the Indoor Channel
Multipath propagation is one of the basic requirements of the MIMO wireless sys-
tems operation [25] and pedestrian movement introduces significant effect on the
multipath propagation conditions in indoor environments [16]. Temporal channel
variations can occur as a result of personnel, industrial machinery, vehicles and
different equipment moving within the indoor environment.
In [4, 16, 26, 27], researcher have modeled pedestrian traffic effects on MIMO
channels for indoor environment.
The time varying effects on the propagation channel within populated indoor
environments depends on different pedestrian traffic conditions, as well as type of
environment considered [26]. Pedestrian movement are important phenomena at
microwave frequencies as antennabody interaction and scattering caused temporal
variation in the channel capacity. Investigations have been carried out through mea-
surements and statistical analysis of the indoor narrowband propagation channel at
2.45 GHz and 5.2 GHz [4, 16, 26, 27]. From the investigation it was observed that
in fixed link the statistical distribution of the received envelope was dependent on
the number of pedestrians present. However, fading was slower than expected, with
an average fade duration of more than 100 ms for a Doppler frequency of 8.67 Hz
[16].
Chapter 3 Literature Review
3.2 Pedestrians and the Indoor Channel 47
In [4, 27], a new channel modeling technique, which offers an efficient solution
to the performance evaluation of MIMO wireless systems in populated indoor envi-
ronment, was presented. The presented model was based on geometrical optics and
a detailed Radar Cross Section (RCS) representation of the human body and was
capable of estimating the temporal response and the capacity behavior of MIMO
channels in the presence of pedestrians traffic. Investigators used FDTD modeling
of Tx and Rx arrays at 2.45 GHz. Initial results indicated that an increase in the
value of the dynamic channel capacity occured when pedestrians blocked the direct
LOS path in a single room environment. In addition, for a single room environment,
the new channel model predicted an increase in capacity from 19.1 bits/sec/Hz to
31.4 bits/sec/Hz solely caused by the movement of pedestrians [4, 27]. Fig. 3.1
shows the top view of simulated scenarios for the pedestrian trajectories, which was
implied during the conducted investigation [4].
Further analysis of the effect of pedestrian movement on channel capacity for
an otherwise LOS MIMO link in a single room has been presented in [16]. Pre-
sented model generated a temporal profile for the complex transfer function of each
antenna combination in the MIMO system in the presence of specified pedestrian
movement. Simulation based investigations were performed in a 42m2 single room,
using a 2.45 GHz narrowband 8 × 8 MIMO array with 0.4 λ element spacing.
Predicted model shows a significant increases in the peak channel capacity due to
pedestrian movement, as well as show mean capacity values was more modest. For
the static empty room case, the channel capacity was 10.9 bits/sec/Hz, while the
mean capacity under dynamic conditions was 12.3 bits/sec/Hz for four pedestrians,
while pedestrian were moving with similar walking speed. The empirical charac-
terization of the narrowband MIMO channels has been established that the capacity
in indoor environment can be enhanced by pedestrian movement.
Chapter 3 Literature Review
3.2 Pedestrians and the Indoor Channel 48
Figure 3.1: Simulated Scenarios (Top View) for Pedestrians Trajectories [4]
Chapter 3 Literature Review
3.3 MIMO-OFDM Testbeds 49
3.3 MIMO-OFDM Testbeds
As discussed above, studies relating pedestrians effect on narrowband MIMO chan-
nels have been conducted [4, 16, 26, 27]. However, so far in best of author’s knowl-
edge, no research has been conducted considering pedestrians effects in an indoor
MIMO-OFDM system environment. In order to accurately assess the effects of
human movement on the performance of emerging MIMO-OFDM systems, it is
important to consider frequency correlation and analyze MIMO-OFDM channels,
which cannot be obtained from the analysis of single carrier MIMO channels. More-
over to design a realistic indoor wireless model, it is important to consider the multi-
path effect in conjunction with the human body shadowing implication in an indoor
environment.
In order to conduct the real life measurement in indoor environments, a testbed
that can transmit and receive actual wireless packets is needed. In the following,
a review of different MIMO or MIMO-OFDM testbeds developed and utilized by
various institutions is given.
Researchers in [18] investigated experimental measurement platform capable of
providing the narrowband channel transfer matrix for wireless communications sce-
narios. The system was used to directly measure key MIMO parameters in an indoor
environment at 2.45 GHz. Linear antenna arrays of different sizes and construction
with up to 10 elements at transmit and receive were utilized in the measurement
campaign. This data was analyzed to reveal channel properties such as transfer
matrix element statistical distributions and temporal and spatial correlation.
In [87], authors reported a testbed which used Altera Stratix II Field Program-
able Gate Array (FPGA) with 4 Analog to Digital (AD) and 4 Digital to Analog
(DA) converters. Each of these devices can handle signals up to 50 MHz in band-
width. The baseband signal produced by the hardware was a Quadrature Phase-Shift
Keying (QPSK) signal. It was encoded using 2x2 Alamouti Space Time Code.
One of the channel sounders, that has been widely used in MIMO channel mea-
surements, is Medav RUSK BRI vector sounder. Authors in [49, 88], have pre-
Chapter 3 Literature Review
3.3 MIMO-OFDM Testbeds 50
sented the details of the Medav RUSK BRI vector sounder and conducted MIMO
channel measurements using the vector sounder. The Medav RUSK BRI channel
sounder consists of eight-element uniform linear arrays at both the transmit and
receive sides. The transmit elements were omnidirectional and can transmit up to
27 dBm to the receiver. The distance between two neighboring antenna elements
was 0.5λ for both arrays. This employs a periodic multi-tone signal with a max-
imum bandwidth of 120 MHz, centred at 5.2 GHz. There was feedback from the
receiver to the transmitter by a cable in order to synchronize the transmitter and
receiver. Authors presented a partial MIMO testbed design and implementation in
[89], which can be expanded from a simple 1× 1 SISO to a complete 8× 8 MIMO
setup. The development and implementation reported was based on 5.25 GHz with
a bandwidth of 25 MHz. Initial results in [89] using a 2x2 MIMO system showed
that uncoded data rates of up to 140 Mbits/s was feasible in a rich scattering indoor
wireless environment.
In [90], authors reported an implemented MIMO-OFDM prototype system with
FPGA. The prototype used by the researchers, targeted over 200Mbps of the max-
imum physical data rate in 40 MHz bandwidth and compatibility with the legacy
IEEE 802.11a. For a cost-effective implementation and improved Packet Error Rate
(PER) performance, 2 transmit and 3 receive antennas and 40 MHz channel were
used to achieve the desired PER and throughput performance in the 5 GHz band.
Information relating to several testbeds from different corners of the world have
been reviewed as well. In [91] authors reported the performance evaluation of a 8×8
multiuser MIMO-OFDM testbed in an actual indoor environment. The presented
testbed can deal with MIMO-OFDM transmission based on the IEEE802.11a signal
format. The testbed transmits RF signal at 4.85 GHz with a maximum operational
bandwidth of 100 MHz. A 2 × 2 MIMO-OFDM channel measurement conducted
by NTT corporation at 5.2 GHz using 10 MHz bandwidth is reported in [92], where
the channels were measured every 5mm along measurement routes in an anechoic
chamber and four different indoor environments. Preliminary 2× 2 MIMO-OFDM
Chapter 3 Literature Review
3.3 MIMO-OFDM Testbeds 51
channel measurement results at 2.4 GHz with 16 MHz bandwidth are reported in
[93]. Graphs of frequency responses within short time scale (200 milliseconds)
measured in NLOS office environment were presented. In [94], 2 × 2 MIMO-
OFDM channels were measured at 5.25 GHz with a bandwidth of 25 MHz for
200 locations along traveling paths indoors, with steps larger than a wavelength in
order to obtain independent channel realizations. Within a laboratory environment,
the LOS and NLOS channels did not show significant differences in terms of the
condition number, which is the ratio of the smallest and largest singular values of
MIMO channel matrix.
In [86, 95] authors reported a hardware implementation of high spectral effi-
ciency MIMO-OFDM, without the knowledge of the channel at the transmitter.
Previously commercial products utilizing MIMO-OFDM with two transmitters and
three receivers, denoted as 2 × 3, were available for WLAN, achieving up to 300
Mbit/s physical layer (PHY) data rate with 7.5 bits/sec/Hz spectral efficiency. The
proposed implementation was based on the draft IEEE 802.11n standard with the
optional LDPC codes at 5.2 GHz using four transmitters and four receivers, and
achieved 600 Mbit/s data rate and 15 bit/s/Hz spectral efficiency. Researchers also
considered two different spatial multiplexing systems, one using ZF another using
a List Sphere Decoder (LSD) . The simpler ZF achieved packet success probability
of 73%, while the more complex LSD achieved packet success probability of 83%.
In both the cases, the average measured SNR was 26 dB.
In addition, using the reported testbed, authors in [17], reported an extensive
measurement campaign considering 4 × 4 MIMO sub-channels, 117 OFDM sub-
carriers, and 6400 receiving antenna array locations per local area. The system was
operated at 5.25 GHz and an operational bandwidth of up to 40 MHz. In [17],
authors provided several area plots, which reveal how the MIMO-OFDM channel
capacity was distributed within an indoor environment. The MIMO-OFDM channel
capacity was found to be strongly dependent on the local scattering environment in
the case of LOS situation, while it was less affected in the case of NLOS situation.
Chapter 3 Literature Review
3.3 MIMO-OFDM Testbeds 52
In relation to this PhD thesis, the CSIRO developed testbed [17, 86, 95] was used
for our investigation. It’s portability and number of accessible antenna elements are
a perfect match for the thesis. A detailed description of the testbed used in this
project can be found in Chapter 4.
An extensive detail and elaborated description of different channel modeling
techniques had been presented in [96]. Authors in [96], reviewed several MIMO-
OFDM channel modeling techniques. In light with their description the following
distribution of channel modeling technique has been established.
• Physical models
• Analytical models
• Standardized models
Under the physical model section there are three more types namely:
1. Deterministic physical models
2. Geometry-based stochastic models
3. Non-geometrical stochastic models
On the other hand the analytical models are
1. Correlation based analytical models
2. Propagation-motivated analytical models
Finally, with standardized models there are a few more subsection in the modeling.
They are
1. Calibration models
2. Simulation models
3. Winner channel models
Chapter 3 Literature Review
3.3 MIMO-OFDM Testbeds 53
Researchers in [85], used a stochastic channel model to analyze and characterize
MIMO radio channels with 4× 4 and 2× 4 at a SNR of 20 dB. They have also con-
ducted experimental validation , which used the correlation matrices at the Mobile
Station (MS) and Base Station (BS) . The model was simplified to the narrowband
channels. The validation of the model is based upon data collected in both picocell
and microcell environments. The stochastic model had also been used to investigate
the capacity of MIMO radio channels with two different antenna topologies, 4 × 4
and 2× 4. Authors in [85], presented general description of MIMO channel model
and validation technique, which is vital for the investigation process of this thesis.
Experimental investigation of the multi path propagation in indoor MIMO chan-
nels was carried out [97]. Authors presented a physically based statistical multipath
propagation model to match capacity statistics and pairwise magnitude and phase
distributions of measured 4×4 and 10×10 narrow band MIMO at 2.4 GHz. Authors
in [97], also considered the normalization to specifying the average receiver SNR
when transmit streams were uncorrelated. The normalization constant computed
for over all matrices at a single location.
Many researchers have confirmed that the performance of MIMO system in a
realistic WLAN environment with OFDM can improved the data transmission [63–
65, 98]. More reviews have been conducted considering OFDM and channel esti-
mation in MIMO-OFDM modeling.
Authors in [15] present the results of MIMO-OFDM channel measurements.
The measurements were performed in indoor environments using four transmitters
and four receivers with 40 MHz bandwidth at 5.25 GHz. From the correspond-
ing measurements it has been observed that in the LOS case, the MIMO-OFDM
channel capacity is found to be strongly dependent on the local scattering environ-
ment and much less dependent in the NLOS case. Also, MIMO channel capacity is
found to be largely uncorrelated over 20 MHz in NLOS, while a strong correlation
is found over 40 MHz in some LOS environments. The validity of the conventional
Kronecker correlation channel model is tested, along with a recently proposed joint
Chapter 3 Literature Review
3.4 Conclusions 54
correlation model. The effects of varying antenna element spacing are also investi-
gated, taking into account such effects as mutual coupling, radiation efficiency, and
radiation pattern.
Empirical analysis is also a very well adopted method, for establishing the chan-
nel characteristics in a simple and standard manner. Researchers have been using
this process for many years to establish channel characteristics [16, 99, 100]. In
[16] authors presented a empirical characteristics of the MIMO channel under the
pedestrian movement considering moving pedestrian with different speeds.
There are several types of MIMO-OFDM channel modeling available. In this
thesis, we have considered deterministic model and empirical characterization of the
MIMO-OFDM channel. Different approaches have been considered by researchers
to characterize MIMO-OFDM channel but human body shadowing effect was never
been considered in those models. For perfectly design and estimate the indoor chan-
nel, consideration of pedestrian effect is highly crucial and important.
3.4 Conclusions
This thesis aims to design an improved model for channel capacity and character-
ize the time varying MIMO-OFDM channels in presence of pedestrian. From the
study it has been confirmed that the MIMO-OFDM system is the next generation
effective wireless system solution for physical layer transmission. With the grow-
ing demand for QoS in the indoor environment, MIMO-OFDM can bring ultimate
satisfaction. Evidence of a significant temporal variation, due to the multipath prop-
agation, human body as well as other industrial equipment. But so far no systematic
analysis relating pedestrian has been conducted. For future design and development
of WLAN, understand and model the channel capacity of the MIMO-OFDM system
with systematic study campaign is highly essential.
Chapter 3 Literature Review
55
Chapter 4
Measurement Equipment and
Scenarios
In this chapter a detailed description of the measurement equipment and procedures
is presented. The aim of this chapter is to describe the equipment used for the
measurements. Additionally, location and measurement scenarios are also depicted.
In this thesis, a 4 × 4 MIMO-OFDM with 40MHz bandwidth has been considered
for conducting measurements. Interestingly, there is a complete lack of currently
available measurement sets for indoor MIMO-OFDM channel in presence of human
bodies.
In this thesis, a systematic measurement campaign on the spatial characteristics
of the MIMO-OFDM channels in a number of indoor local areas has been car-
ried out. The change in channel as a function of number of antenna and number of
pedestrians present in the indoor environment has been explored and analyzed. Two
different measurement scenarios have been considered for MIMO-OFDM channel
data collection, namely deterministic scenarios and random scenarios. Under de-
terministic scenarios, controlled pedestrians ranging from one to three have been
moved between the Txs and Rxs, following a given trajectory. On the other hand,
for random scenarios, up to ten arbitrarily moving pedestrians have been placed
between Txs and Rxs, within a given area.
Chapter 4 Measurement Equipment and Scenarios
4.1 Measurement Equipment 56
4.1 Measurement Equipment
4.1.1 General Description
All the measurements for this thesis were performed using the MIMO-OFDM chan-
nel sounder developed by CSIRO ICT Centre currently equipped with 4 transmit-
ters and 4 receivers. A detailed description of the equipment can be found in
[15, 101, 102]. The channel sounder operates at a carrier frequency of 5.24 GHz and
has an operational bandwidth of 40 MHz. The channel sounder has 4 transmitters
with maximum power of 23 dBm per channel and 4 receivers with 3 dB noise figure
over the 40 MHz bandwidth. Commercially available omnidirectional loop anten-
nas (Sky-Cross SMA-5250-UA) were used both for transmitter and receiver arrays.
The antenna elements are placed in a square array fashion, with a spacing of three
wavelengths for the transmitter emulating an access point and two wavelengths for
the receiver emulating a PC client. The hardware was designed and built in-house
end to end, including full multi-channel radio Tx and Rx, and digital hardware
that supports multiple high processing power FPGA with a flexible configuration.
The effects of antenna spacing on the performance of MIMO capacity have been
investigated by several researchers [103–105]. However, the complex interaction
of mutual coupling between the antenna elements and changes in radiation pattern
make an analytical approach difficult. Authors [15] reported the direct measure
of MIMO-OFDM channels, while varying antenna element spacing of the uniform
square array from 0.5 wavelengths to 2 wavelengths with 0.5-wavelength. When
operating the channel sounder, users can generate, via software, signals which are
simultaneously sent from the transmitters, and captured as multiple signal streams
at the receivers. For all the experiments conducted, the sampling rate was approxi-
mately 2 samples/sec.
Fig. 4.1(a) and Fig. 4.1(b) show the CSIRO ICT centre in-house developed
channel sounder. The CSIRO developed signal processing hardware demonstra-
tor is suitable for flexible prototyping of new wireless signalling proposals, using a
Chapter 4 Measurement Equipment and Scenarios
4.1 Measurement Equipment 57
(a) Transmitter
(b) Receiver
Figure 4.1: MIMO-OFDM Channel Sounder
Chapter 4 Measurement Equipment and Scenarios
4.1 Measurement Equipment 58
MIMO communication link. There are only a few demonstrators of this capability
currently available worldwide [62, 78, 106]. Significant features of the demonstra-
tor include the use of high quality radio components and flexible digital hardware.
The use of high quality radio components allows accurate wideband MIMO chan-
nel measurements [15] as well as testing of advanced transmission schemes such as
256 Quadrature Amplitude Modulation (QAM) .
4.1.2 Technical Specifications
For MIMO-OFDM channel sounding purposes, typically a packet consists of a
preamble (for performing packet detection,frame synchronization, and frequency
offset correction [50]) and a channel training sequence. The channel training se-
quence is designed to estimate the frequency response over 114 OFDM sub-carriers
in a 40 MHz bandwidth with the subcarrier spacing of 312.5 kHz. The choice of
OFDM sub-carriers is consistent with [65], except that the three middle null car-
riers are also used. To avoid the interference of signals transmitted from different
transmitting antennas, the channel training sequence is sent from each transmitting
antenna at different times [50]. In order to reduce the effect of noise, the chan-
nel training sequence is sent ten times at each location, while the estimation of the
channel is performed ten times and the averaged results are used for the analysis.
A detailed calibration of the system was performed prior to the measurement, by
directly connecting each of the Txs to each of the Rxs via cables and an attenuator,
and measuring the frequency response of each pair of Tx and Rx. The frequency
response of the system is subtracted from the measured over-the-air MIMO-OFDM
channels. This removes any effects of RF front-end filters in Tx and Rx devices.
The transmitting power used during the measurement was varied from 0 dBm to 10
dBm per transmitting antenna, depending on the environment and the distance be-
tween Tx and Rx. Observed SNR was better than 26 dB [95], when averaged over
frequency. A set of MIMO-OFDM channels in a local area consists of the channel
coefficients of 16 MIMO sub-channels at 114 OFDM sub-carriers at several loca-
Chapter 4 Measurement Equipment and Scenarios
4.1 Measurement Equipment 59
tions, which amounts to millions of channels per local area measurement.
A multi-channel radio Tx, a multi-channel radio Rx, and two MIMO digital
hardware platforms have been included in the MIMO hardware design. A PC con-
nection can be established using USB to provide debugging and display of various
outputs. The radio Tx and Rx translates 5.24 GHz RF signals from and to 140 MHz
digital Intermediate Frequency (IF) signals. During experiments four Txs and four
Rxs have been included. An upgrade path to eight Txs and eight Rxs are also
available in the system. The DA and AD operate at IF (digital IF). Common IQ
mismatch problems, such as DC offset, amplitude mismatch, and phase mismatch,
are avoided by the use of the digital IF. The channel sounder uses burst mode to
transfer data from Txs to Rxs. In burst mode condition the transmitters send data
repeatedly without waiting for input from the receivers or waiting for an internal
process to terminate before continuing the transfer of data. Further details of the
MIMO-OFDM demonstrator can be acquired from [78].
Fig. 4.3 shows a schematic diagram of the MIMO-OFDM testbed. The hardware
supports (4×4) MIMO at 5.2 GHz with up to 56 MHz bandwidth digital IF channels
or 112 MHz baseband channels. The testbed is designed with two PCs, one with
four Digital to Analog Converter (DAC)s and the other with four Analog to Digital
Converter (ADC)s , each using 12 bit resolution sampling at 112 Mega samples per
second1. Both DACs and ADCs are designed to process signals at IF which are
converted to and from the RF signals by the multi-channel Tx and Rx. DACs can
load signals and ADCs can capture signals via the network, which allows a user with
a remote PC to generate a signal sequence (for example by using the MATLAB2
programming environment), load it on to the DACs, transmit it via air, capture it
on the ADCs, and process it on the remote PC. This flexibility allows a user to
perform not only the channel sounding, but also testing of different modulation and
1The use of 112 Mega samples per second DAC and ADC to generate and sample IF signals at
140 MHz is performed by making use of high order Nyquist zones.2MATLAB version 7.8.0.347(R2009a) 32-bit(win32) developed and distributed by The Math
Works Inc.
Chapter 4 Measurement Equipment and Scenarios
4.1 Measurement Equipment 60
(a) Enlarged Transmitter Panel
(b) Enlarged Receiver Panel
Figure 4.2: Details Front Panel View of Transmitter and Receiver
Chapter 4 Measurement Equipment and Scenarios
4.2 Measurement Locations 61
Figure 4.3: A Schematic Diagram of the MIMO-OFDM Testbed
coding schemes of MIMO-OFDM transmission. A detail front view photograph of
the equipment is shown in Fig. 4.2.
4.2 Measurement Locations
The experiment location was CSIRO ICT centre, Marsfield, Sydney. All the ex-
periments for this thesis have been conducted in the ground floor of the building.
Fig. 4.4 shows the entire floor plan for the CSIRO ICT Centre. Here the black
marked box highlights the specific rooms for experiments. LOS deterministic burst
mode experiments were carried out in Room 386 and LOS random burst mode ex-
periments were carried out in Room 52C. Fig. 4.5(a) and Fig. 4.5(b) depicts the
individual floor plans for each experiment site, Room 386 and Room 52C respec-
tively.
In all adjacent locations, the walls were constructed from painted concrete blocks
Chapter 4 Measurement Equipment and Scenarios
4.2 Measurement Locations 62
Figure 4.4: Floor Plan of CSIRO ICT Centre. Measurement Sites, Rooms 386 and
52C, are Highlighted.
Chapter 4 Measurement Equipment and Scenarios
4.2 Measurement Locations 63
(a) Deterministic Measurements Site, Room 386
(b) Random Measurements Site, Room 52C
Figure 4.5: Experimental Floor Plans
Chapter 4 Measurement Equipment and Scenarios
4.2 Measurement Locations 64
and plywood and the floor was cement based. During the experiments, both loca-
tions were cleared of furniture and obstructions, to allow the free movement of
pedestrians. All the doors and windows of the experiment rooms were kept shut at
the time of the experiments. The following Sections provide a detailed description
of each experimental setup.
4.2.1 LOS Deterministic Burst Mode: Room 386
For the LOS deterministic burst mode experiments Room 386 was used; this room
is also known as the Showcase Room. Within the 60m2 room, both Tx and Rx were
separated by 10m. Both the Tx and Rx were kept approximately at the same height
of approximately 1m. The room was completely furniture free. The given trajectory
for the pedestrian was 6m, as shown in Fig. 4.5(a). The indicated trajectory was
diagonally aligned between the LOS of the Tx and Rx. The ceiling was suspended
at a height of 5 m and was composed of mineral tiles and fluorescent lights. Along
the 6m wall near the Tx antenna location there was an empty shelf attached to
the wall by metal fixing rods. Fig. 4.6 shows a pictorial view of the entire room,
including transmitter and receiver locations.
4.2.2 LOS Random Burst Mode: Room 52C
Channel measurements under random conditions were carried out in Room 52C,
also known as the Schottky Room. During these random trajectory experiments,
both Tx and Rx, separated by 6.5 m, were located inside the same 42m2 room.
All the wooden tables were lined up against the inside walls around the room. In-
side the room a 30m2 space was used by pedestrians following random trajectories.
Fig. 4.7(a), 4.7(b), 4.7(c) show the arrangements of the Schottky room. Room 52C
has similar structural material (such as mineral tiles) as Room 386, with the ceil-
ing at a height of 5m and 6m×7m walls. In the room there were a few permanent
fixtures, including two plasma screens and a projector screen against the 6m walls.
Chapter 4 Measurement Equipment and Scenarios
4.3 Measurement Procedure 65
Figure 4.6: Experimental Setup at Room 386
4.3 Measurement Procedure
In this section, a detailed description of the measurement scenarios is presented.
As discussed in Section 4.2, two different locations were considered for the experi-
ments and several sets of data were collected on each location.
Complex channel coefficients for each of 16 MIMO subchannels at 114 OFDM
subcarriers were collected at 100 time samples while pedestrians were walking in a
given deterministic trajectory. Due to the hardware limitation, the sampling rate of
the measurement was limited to approximately two samples/s, and hence, pedestri-
ans moved slower than usual (less than 1 km/h). We note that the capacity dynamic
range, as defined in Chapter 2 Section 2.6, does not depend on the speed of the
pedestrian or on the sampling rate as long as enough measurement points are col-
lected. The measurements were performed once for each scenario. Results from
this analysis reported in [47].
Due to the presence of a strong multipath environment NLOS scenarios have
Chapter 4 Measurement Equipment and Scenarios
4.3 Measurement Procedure 66
(a) (a) Room 52C (b) (b) Room 52C
(c) (c) Room 52C
Figure 4.7: Schottky Room (52C), used for Random Trajectory Experiments
Chapter 4 Measurement Equipment and Scenarios
4.3 Measurement Procedure 67
been proved to provide excellent conditions for the improved performance of MIMO
systems. In that regards, this thesis will focus on the analysis of LOS scenarios
exclusively, where a better understanding of how human body can affect and con-
tribute to an increase multipath condition and therefore an increased channel capac-
ity for MIMO systems.
4.3.1 LOS Deterministic Burst Mode Measurement Procedure
Data have been collected under controlled pedestrian traffic conditions in Room
386. Pedestrian trajectories for LOS experiments are shown in Fig. 4.5(a). Four dif-
ferent scenarios were considered: vacant, one, two and three persons walking along
the indicated trajectories. Complex channel coefficients were collected as pedestri-
ans walked within the given 6m trajectory crossing the direct line of sight of Tx and
Rx within the room. Additionally, as the performance of the MIMO-OFDM sys-
tem can dramatically change due to a small shift of the antenna array, two data sets
have been collected for each scenario by placing the Rx antenna in two different
locations 4λ (approximately 25 cm) apart. Wide band relative power was collected
for the 4x4 antenna-to-antenna channels. For each scenario 100 samples were col-
lected. Each sample has 16 antenna-to-antenna channels and each of these is made
up of 114 OFDM sub carrier samples. During the LOS experiments Tx and Rx
were placed within the same laboratory. The distance between the Rx and Tx was
10 meters. For the first data set, the Rx was placed within the laboratory, as shown
in Fig. 4.5(a). Received power was recorded for the four different pedestrian traffic
scenarios: vacant, one, two, and three pedestrians walking along the trajectory. The
second data set was collected after moving the Rx antenna approximately 4λ apart.
4.3.2 LOS Random Burst Mode Measurement Procedure
At the time of the random trajectory experiments, uncontrolled pedestrian traffic
is considered for data collection. Single antenna array location is considered for
randomly moving pedestrians numbering 1-5, 7, 10 within Room 52C. Fig. 4.5(b)
Chapter 4 Measurement Equipment and Scenarios
4.4 Data Processing 68
shows the antenna locations and empty pedestrian moving space in the Room 52C
perspective. The distance between the Rx and Tx was 6.5 m. For individual scenar-
ios, at least 180 samples and approximately 200 samples were collected. Received
power was recorded for the experimental scenarios. Wide-band relative power was
collected for 4x4 antenna element array. Complex channel coefficients for each
of 16 MIMO sub-channels and 114 OFDM sub-carriers were collected at approxi-
mately 2 samples per second as pedestrians walked randomly within the room. We
note that the average channel capacity does not depend on the speed of the pedes-
trian or on the sampling rate, as long as enough measurement points are collected.
The measurements were performed during the day in normal office hours.
Fig. 4.8(a) and Fig. 4.8(b) show the Schottky Room (Room 52C) arrangement
showing transmitter and receiver locations. In addition, a picture of randomly mov-
ing people has also been included to show an example of the random experimental
scenario, see Fig. 4.9. Several data sets have been collected during working hours
and after hours. This thesis has only considered the working hour measurements, as
that imposes more realistic indoor environment conditions. Additionally, analysis
shows a negligible difference between working and after hours data.
4.4 Data Processing
In this thesis, during investigation a large amount of measurement and simulation
data has been handled and processed. This section aims to deliver the overall picture
of data collected during deterministic and random measurement scenarios. The
section is organized as follows: The first two sections discuss data collection for
• Deterministic Measurement Scenarios, and
• Random Measurement Scenarios,
The third section describes the amount of collected data and the PC configuration
that has been utilized for data processing.
Chapter 4 Measurement Equipment and Scenarios
4.4 Data Processing 69
(a) Room 52C Setup (Transmitter End)
(b) Room 52C Setup (Receiver End)
Figure 4.8: Schottky Room (52C) Arrangement Showing Tx and Rx Location
Chapter 4 Measurement Equipment and Scenarios
4.4 Data Processing 70
Figure 4.9: Randomly Moving People between Tx and Rx in Room 52C
Please note that the computational resource requirement statistics presented in
this section excludes data samples for preliminary testing, which will increase the
size of the collected data and amount of storage disk space.
4.4.1 Deterministic Measurement Scenarios
Deterministic measurement was the first sets of data collected in the systematic
measurement campaign. During this time data was collected considering 4× 4 sys-
tem with 114 sub-carriers. Two sets of data have been collected using two different
antenna locations. In addition, data was also collected considering four different
scenarios (vacant, one, two and three persons walking). A total of 4(scenario) ×2(dataset) × 100(timesample) = 800 MIMO-OFDM channels was collected,
which is made up of 4(scenario) × 2(dataset) × 100(timesample) × 4(Tx) ×4(Rx)× 114(subcarriers) = approximately 1.5 million data samples of SISO sin-
gle carrier channel. The collected and processed data occupied around 0.5 GB of
Chapter 4 Measurement Equipment and Scenarios
4.4 Data Processing 71
disk space. In this thesis, out of two only one position was taken into account for
analysis purpose. 3
4.4.2 Random Measurement Scenarios
For random scenarios, more realistic human bodies were involved in measurement
campaign. This time 8 different scenarios (vacant, one to five, seven and ten people
moving randomly) were considered. For random scenarios 200 measurement sam-
ples were collected in 200 individual files, from which were generated 200 chan-
nel data for individual scenario, resulted 8(scenario) × 400(numberoffiles) ×2(dataset) = 6400 files. With 200 time samples and 2 sets of data these 6400
files hold 8(scenario) × 2(dataset) × 200(timesample) = 3200 MIMO-OFDM
channels and 8(scenario)× 2(dataset)× 200(timesample)× 4(Tx)× 4(Rx)×114(subcarrier) = approximately 6 million data samples of SISO single carrier
channel have been collected. The entire data required 5 GB of disk space to store.
Two sets of data considering working and non working hours have been collected
and analyzed. Only data relating to working hours was included in this thesis as
that represents a more realistic indoor environment.
4.4.3 Total Measured Data
Type Scenarios Time Samples MO Samples SO Samples Disk Space
Det Mes 4 100 800 1.5 Million 0.5 GB
Ran Mes 8 200 3200 6 Million 5 GB
Grand Total 4000 7.5 Million 5.5 GB
Table 4.1: Statistical Facts of the Project (Det:Deterministic, Ran:Random,
Mes:Measurement, MO:MIMO-OFDM channel, SO: SISO Single Carrier channel)
3The second dataset, though collected, was subsequently found not usable due to measurement
error.
Chapter 4 Measurement Equipment and Scenarios
4.5 Conclusions 72
Table 4.1 shows the statistical summary of the entire measurement data col-
lected for this thesis. All together approximately 4000 MIMO-OFDM channel and
7.5 million SISO data samples have been collected. Around 5.5 GB disk space has
been used to store the entire measurement data set. The entire data analysis for mea-
surement scenarios has been handled by a simple desktop PC. A detail configuration
of the desktop PC is as follows:
• Desktop PC:
System: Microsoft Windows XP Professional, Version 2002, Service Pack 3
Computer: Intel(R) Core(TM)2 Duo CPU, [email protected], 4GB RAM
4.5 Conclusions
This chapter has presented the measurement setup and procedures for LOS deter-
ministic and random burst mode experiments. Detailed layouts of Rooms 386 and
52C have been included, showing the transmitter and receiver antenna location.
The floor layout of both rooms shows detail of the room structure. In addition, the
pedestrian trajectories for both deterministic and random experiments have been
described. In this thesis several experiments have been conducted, considering dif-
ferent antenna array combinations, as well as different environmental conditions. In
all the experiment sets an identical configuration of the same channel sounder has
been used. For LOS deterministic burst mode 0-3 people and LOS random burst
mode 0-5, 7 and 10 people have been considered. Complex channel coefficients for
each 16 MIMO sub-channels and 114 sub-carriers were measured for 4× 4 antenna
element array. Data has been extracted from 4 × 4 results to analyze 3 × 3 and
2× 2 antenna arrays. Results from the deterministic experiments will be discussed
in Chapter 6, while Chapter 7 will present a detailed analysis of the random tra-
jectory measurements. The next chapter describes the simulation technique used to
replicate all the measurement scenarios.
Chapter 4 Measurement Equipment and Scenarios
73
Chapter 5
Simulation Software and Scenarios
This chapter focuses on the details of the customized simulation software and sce-
narios developed for the thesis. The main aim of this chapter is to provide a clear
understanding of the conducted simulations that replicate all the measurement loca-
tions and scenarios considered in this research.
5.1 Simulation Software
In this thesis, a MATLAB 1 based ray tracing simulation has been implemented for
all experimental scenarios described in Chapter 4.
The ray tracing simulation tool uses Frustum Ray Tracing Technique (FRTT)
for the prediction of channel characteristic maps in a complex indoor environment
[107], has been implemented. The ray tracing technique has been widely utilized to
predict the static narrow-band/wide-band characteristics of indoor radio channels.
This GO based technique is strictly valid only as long as the dimension of the ob-
jects in the modeled structure is large enough compared to the wavelength. The
accuracy of the predicted results for an indoor environment has been verified by
other prediction techniques, such as the exact integration of Kirchoff integral [108]
1MATLAB version 7.8.0.347(R2009a) 32-bit(win32) developed and distributed by The Math-
Works Inc.
Chapter 5 Simulation Software and Scenarios
5.1 Simulation Software 74
and the exact modal solution [109].
Although there have been several algorithms proposed to enhance the calcula-
tion efficiency of ray tracing prediction, they often introduce additional errors that
are not inherited from the GO approximation. The technique utilized very closely
follows the GO solution and is capable of accommodating a large number of re-
ceiving points involved in a channel characteristic map. It allows the prediction
of hundreds of thousands of receiving points to be calculated on a typical personal
computer efficiently and accurately. Unlike other ray tracing techniques, which
trace rays or ray tubes, the adopted method traces pyramids or frustums, and thus is
named as FRTT. The main advantages of FRTT are its accuracy as it predicts the GO
path exactly and its capability of guaranteeing that no receiving point or modeled
object is missed. In addition, the CPU time consumed by FRTT is highly acceptable
and nominal, hence a simple desktop computer can be utilized for simulation. In
general the time simplicity can be approximately expressed as mr where m is the
average number of building objects inside a frustum [107]. When the number of r
is small, the FRTT is more efficient than the conventional ray tracing techniques,
because the number of frustums that FRTT creates is much smaller than the number
of rays or ray tubes generated by any conventional ray tracing technique. Since it
is a full 3-D technique, FRTT can easily be used, even for predictions involving
multiple floors.
In the customized section of the software, we have implemented several modules
to replicate the measurement scenarios which were dynamically simulated consid-
ering permeability and conductivity of materials in the environments. This modified
section of the software provides the extra feature of a simplified human body, which
can be located at different positions in either a deterministic or a random fashion.
Several replicated investigations have been carried out to confirm the reflection or-
der for the simulation. Here, the reflection order is the number of reflections that
have been considered for the analysis of the predicted scenarios. After analyzing the
data, four reflections have been considered for all the MATLAB based simulations.
Chapter 5 Simulation Software and Scenarios
5.1 Simulation Software 75
The FRTT has the advantage of accurately modeling radio propagation (spec-
ular reflection) in a three dimensional environment. The value of the transmission
coefficient is largely dependent on the value of conductivity, while the value of per-
mittivity has a major influence on the reflection order [110]. If the transmission and
reflection coefficients of the target building material are known, permittivity and
conductivity of the material can be estimated by inverse calculation of the multi-
layer dielectric slab model [110]. For the simulations, walls are modeled as sin-
gle slabs whose permittivity and permeability are determined from penetration loss
measurement of the actual material [111].
A very simple model of the human body was employed (a rectangular block with
a dimension of 0.62 m depth, 0.31 m width and 1.70 m height with the permittivity
and conductivity characteristics of a real human body [16]). Diffraction and scatter-
ing were not included in the analysis. One simulation was performed for different
receiver antenna array locations defined on a grid within an area of two wavelengths
times two wavelengths with 0.1 wavelength resolution resulting in 400 locations, in
order to observe the variation of the capacity dynamic range as a function of small
scale displacement of the antennas. While small variations in the capacity dynamic
range was observed, depending on the exact location of the receiver antenna array,
the dynamic range results are averaged over 400 receiver antenna array locations,
to obtain the trend as a function of the number of antennas and of pedestrians.
At the initial stage of the simulations, we have conducted a systematic reflec-
tion order analysis by conducting simultaneous simulations considering 4 and 5
reflections in the ray tracing algorithm. Using these tests an appropriate reflection
order of 4 has been chosen for the entire analysis. Fig. 5.1 shows an example of
the simulated capacity dynamic range for the Fixed SNR for a reflection order of
4 and 5 and Fig. 5.2 shows for the simulated capacity dynamic range for the Fixed
Tx power scenarios. In both cases, a minimum difference is observed in terms of
Dynamic Range. A similar trend is noted, for reflections coefficient of 4 and 5,
in the conducted simulated results. Although it could be argued that a reflection
Chapter 5 Simulation Software and Scenarios
5.1 Simulation Software 76
order of 5 will generate more accurate results, it will also claim a higher compu-
tational time and efficiency. To minimize the computational calculation time, we
have considered a reflection order of 4 for entire simulation analysis. By using a
smaller reflection order of 4 a significant reduction in the computational time has
been achieved. Therefore, a conventional desktop PC could handle the simulation
if required. In addition, Fig. 5.3 shows the repetitive simulation results of randomly
selected number of human body. An exact result has been obtained when the simu-
lations were reiterated.
1ppl 2ppl 3ppl 4ppl 5ppl 6ppl 7ppl
2
4
6
(a)
Simulation Results in Fixed SNR Ref 4
2x2 3x3 4x4
1ppl 2ppl 3ppl 4ppl 5ppl 6ppl 7ppl1
2
3
4
5
D
ynam
ic R
ange
of
Med
ian
Cap
acity
(b)
Simulation Results in Fixed SNR Ref 5
2x2 3x3 4x4
Figure 5.1: Simulated Comparison for Reflection order Analysis (Fixed SNR)
To maintain consistency and to assure accurate results, all simulation scenarios
have been conducted more than once. A negligible variation has been noticed be-
tween repetitions. Fig. 5.1 and Fig 5.2 show the comparison between repetitions of
Chapter 5 Simulation Software and Scenarios
5.1 Simulation Software 77
1ppl 2ppl 3ppl 4ppl 5ppl 6ppl 7ppl
5
10
15
(a)
Simulation Results in Fixed Tx Ref 4
2x2 3x3 4x4
1ppl 2ppl 3ppl 4ppl 5ppl 6ppl 7ppl4
6
8
10
12
14
D
ynam
ic R
ange
of
Med
ian
Cap
acity
(b)
Simulation Results in Fixed Tx Ref 5
2x2 3x3 4x4
Figure 5.2: Simulated Comparison for Reflection Order Analysis (Fixed Tx)
Chapter 5 Simulation Software and Scenarios
5.1 Simulation Software 78
the random scenarios for the same reflection order.
1ppl 2ppl 3ppl 3pplRep 4ppl 5ppl 5pplRep 6ppl 7ppl 7pplRep1
2
3
4
5
6
(a)
Simulation Results in Fixed SNR Ref 4 with Random Repeat
2x2 3x3 4x4
1ppl 2ppl 3ppl 3pplRep 4ppl 5ppl 5pplRep 6ppl 7ppl 7pplRep
5
10
15
D
yn
amic
Ran
ge
of
Med
ian
Cap
acit
y
(b)
Simulation Results in Fixed Tx Ref 4 with Random Repeat
2x2 3x3 4x4
Figure 5.3: Repeated Simulation Comparison Analysis
The aim of the simulation is to capture the variation trend on the capacity dy-
namic range, rather than predicting the exact MIMO-OFDM channel capacity at the
time of measurement.
The algorithms were implemented on MATLAB with double-precision floating-
point values. The OFDM parameters used in the simulations are identical to those
used for the measurements. Using the implemented simulation all the variation
trend of the capacity dynamic range due to the human body shadowing effect has
been captured.
Chapter 5 Simulation Software and Scenarios
5.1 Simulation Software 79
Figure 5.4: Deterministic Model Room and Pedestrian Block
Figure 5.5: Random Model Room and Pedestrian Block
Chapter 5 Simulation Software and Scenarios
5.2 Simulated Locations 80
5.2 Simulated Locations
The FRTT starts by enclosing the entire model space with a rectangular box called
the environment bounding box. In the present case, the environment bounding box
is formed by the four walls, a floor, and a ceiling of the room. Then the modeled
space is split into six pyramids, each of whose apex coincides with the location of
the source antenna.Tx, and whose base face coincides with one of the six faces of
the building bounding box. The pyramid used in the modelled space is called a ray
pyramid. A ray pyramid consists of a view point, a base face, and three or more side
faces. A ray frustum is created when a view face is defined between the view point
and the base face of a ray pyramid. For each ray pyramid, the FRT is performed.
Here, the term frustum is used to denote both a pyramid and a frustum.
Simulation was performed for different receiver antenna array locations defined
on a grid within an area of two wavelengths times two wavelengths, with 0.1 wave-
length resolution resulting in 400 locations, in order to observe the variation of the
capacity dynamic range as a function of small scale displacement of the antennas.
While a small variation in the capacity dynamic range was observed, depending on
the exact location of the receiver antenna array, the dynamic range results are aver-
aged over 400 receiver antenna array locations, to obtain the trend as a function of
the number of antennas and of pedestrians.
5.2.1 LOS Deterministic Simulation: Room 386
Five different building blocks have been placed together to form the room shape
similar to Room 386 (See Fig. 4.6. For this section of the simulation, the human
body block was placed on the given trajectory. The defined trajectory is placed in
between the Tx and Rx antenna array. Fig. 5.4 shows a sample of deterministic
Model Room with pedestrian block, antenna elements distribution and determin-
istic trajectory. Here the diagonal lines indicate the defined trajectory that will be
followed by different numbers of people.
Chapter 5 Simulation Software and Scenarios
5.3 Simulation Scenarios 81
5.2.2 LOS Random Simulation: Room 52C
LOS Random Burst Mode has been replicated in this section of the simulation.
The replicated room block is similar to the Room 52C. We have placed several
numbers of human bodies in the room environment. Fig. 5.5 presents the Random
Model Room with randomly placed human blocks. Here the blocks show the diverse
direction to replicate the random movement of the human in an indoor environment.
5.3 Simulation Scenarios
Simulations are designed to replicate the measured experimental scenarios. An ex-
tensive number of simulated scenarios was conducted before establishing a hypoth-
esis for the measurement scenarios. The main simulated scenarios were
1. Deterministic LOS Burst Mode
2. Random LOS Burst Mode
In both cases, the custom build environment box was utilized considering all the
existing material characteristics.
5.3.1 Deterministic LOS Burst Mode
LOS Deterministic Simulation has been conducted through replication of the LOS
Deterministic Burst Mode measurement. In this simulation we have created a simi-
lar room block, a near match of Room 386. With a given trajectory we have placed
1-3 human blocks from one end of the trajectory to the other. In more human block
cases, such as 2 and 3, the blocks have been placed perpendicular to each other.
Also we have made sure all the blocks are moving together while posing as 2 or 3
persons walking together. We have kept the Tx location Fixed and moved the Rx
location in 400 places. Finally we have averaged channel capacity over 400 loca-
tions to capture the best possible results. Data has been acquired placing the block
Chapter 5 Simulation Software and Scenarios
5.4 Data Processing 82
in the formed room environment. The body model was then removed and from its
previous location. Now following the trajectory a new block has been placed and
similar process has been carried out to collect the data for rest of the locations. This
is the simple process of replicating the real time scenario, such as people walking
within a given trajectory between the Tx and Rx locations. We note that the capacity
dynamic range, as defined in Chapter 2 section 2.6, does not depend on the speed
of the pedestrian or on the sampling rate, as long as enough measurement points are
collected. The measurements were performed once for each scenario.
5.3.2 Random LOS Burst Mode
Within the given replicated room environment, we have randomly placed one to sev-
eral human blocks and collected the data for channel measurement. Once data have
been collected, new sets of the same number of human blocks have been placed and
data have been collected for several receiver antenna locations. We have randomly
created the different directions of the human body, to generate realistic movement.
Similar to the deterministic simulation, we have kept the Tx Fixed and placed the
Rx in 400 different locations.
5.4 Data Processing
For this thesis, a large amount of simulated data have been gathered and processed.
This section summarizes the required computational resource and total amount of
data collected during simulations. The first two sections discuss data collection for
• Deterministic Simulation Scenarios, and
• Random Simulation Scenarios
The third section describes the high performance computing configuration that has
been utilized for data processing.
Chapter 5 Simulation Software and Scenarios
5.4 Data Processing 83
The computational resource requirement statistics presented in this section ex-
cludes data samples for preliminary testing and reflection depth testing, as explained
in 5.1, which will increase the size of the collected data and amount of storage disk
space.
5.4.1 Deterministic Simulation Scenarios
During deterministic simulation, a few step by step procedures were followed, start-
ing with generation of 4 × 4 impulse response files for the 4 × 4 MIMO-OFDM
system with 4 different scenarios (vacant, one to three) and using 27 time samples.
Then a frequency response file was generated from individual files. This resulted in
4(scenario) × 4(Tx) × 4(Rx) × 27(timesample) = 1728 impulse response and
1728 frequency response files (In total 3456 files). An average of 27 time samples
have been considered and 4(scenario) × 1(dataset) × 27(timesample) = 108
MIMO-OFDM channels in 400 receiving antenna locations resulted 4(scenario)×27(timesample)×400(receiverlocation)×4(Tx)×4(Rx)×114(subcarrier) =
approximately 79 million data samples of SISO single carrier channel. It is noted
that the capacity dynamic range, as defined in Chapter 2 section 2.6, does not de-
pend on the speed of the pedestrian or on the sampling rate, as long as enough
measurement points are collected. In total, the simulated data for deterministic sce-
narios occupied around 50 GB of disk space.
5.4.2 Random Simulation Scenarios
For random simulation, we have considered 10 different scenarios, including vacant,
one to seven, ten, fifteen and twenty people moving randomly. On an average 60
time samples for the each random scenario were also considered. To achieve con-
sistency, a collection was started with 4 × 4 impulse response files for the MIMO-
OFDM system, then a frequency response file from individual file was generated.
This resulted in 10(scenario) × 4(Tx) × 4(Rx) × 60(timesample) = 9600 im-
pulse response files and 9600 frequency response files (In total 19200 files). An av-
Chapter 5 Simulation Software and Scenarios
5.4 Data Processing 84
erage of 60 time samples have been considered, and 10(scenario)× 2(dataset)×60(timesample) = 1200 MIMO-OFDM channels in 400 receiving antenna loca-
tions resulted 10(scenario)×60(timesample)×400(receiverlocation)×4(Tx)×4(Rx)× 114(subcarrier) = approximately 438 million data samples of SISO sin-
gle carrier channel. A total of 120 GB disk space was used for storing the collected
and processed data. In this thesis, data relating vacant, one to five, seven and ten
people randomly moving in an indoor environment have been presented.
5.4.3 Total Simulated Data
Type Scenarios Time Samples MO Samples SO Samples Disk Space
Det Sim 4 27 108 79 Million 50 GB
Ran Sim 10 60 1200 438 Million 120 GB
Grand Total 1279 517 Million 170 GB
Table 5.1: Statistical Facts of the Project (Det:Deterministic, Ran:Random,
Sim:Simulation, MO:MIMO-OFDM channel, SO: SISO Single Carrier channel)
Table 5.1 shows the statistical summary of the entire simulated data collected
for this thesis. This table does not include data from several other test cases such
as reflection depth analysis from one to five, lights on/off and NLOS scenarios. All
together approximately 1308 MIMO-OFDM channel and 517 million SISO data
samples have been collected. Around 170 GB disk space has been used to store
the entire data set. A high performance PC has been utilized to quickly process the
simulated data. A detailed configuration of the high performance PC, that has been
used for data collection and analysis is as follows:
• High Performance Computer:
The SGI Altix XE Cluster is part of QUT’s High Performance Computing
(HPC) resources and was commissioned in late 2007, with an upgrade per-
formed early in 2009. The machine is a cluster with a total of two-hundred
Chapter 5 Simulation Software and Scenarios
5.5 Conclusions 85
(200) cores. The specifications are:
System:
SUSE Linux Operating System
Computer:
120 x [email protected] 64bit Intel Xeon processor cores
80 x [email protected] 64bit Intel Xeon processor cores
Configured as 24 compute nodes of quad core, dual processors (8 cores per
node)
384 GBytes of main memory (24 x 16GBytes)
5.5 Conclusions
This investigation replicated the experiments conducted in CSIRO ICT Centre. For
LOS deterministic burst mode 0-4 people and LOS random burst mode, 0-5, 7 and
10 people have been considered for the experiment. The frustum ray tracing tech-
nique has the advantage of accurately modeling radio propagation (specular reflec-
tion) in a three dimensional environment efficiently. For the simulations, walls are
modeled as single slabs whose permittivity and permeability are determined from
penetration loss measurement of the actual material [111]. In previous studies us-
ing the frustum ray tracing algorithms, [107], it was found that with a maximum of
4 reflections, accurate channel prediction was achieved in an indoor LOS environ-
ment. Hence we have opted to use up to 4 consecutive reflections for the simulated
scenarios. Complex channel coefficients for each 16 MIMO sub-channels and 114
sub-carriers were considered for 4x4 antenna element array. Despite the fact that
the simulations computing were very time consuming, by using a reflection order
of 4, the process can be handled by a conventional desktop PC.
Chapter 5 Simulation Software and Scenarios
5.5 Conclusions 86
Chapter 5 Simulation Software and Scenarios
87
Chapter 6
Analysis of Results for Deterministic
Scenarios
In this chapter, a detailed analysis of the measured and simulated results using de-
terministically determined pedestrian movement are presented. The chapter also
presents a comprehensive comparison between measurements and simulations re-
sults obtained for the deterministic scenarios. Deterministic scenarios, where pedes-
trians are in motion following a given trajectory between the Rxs and Txs, have been
considered for up to three pedestrians. The perpendicularly placed pedestrians are
following a predetermined walking trail, which allows them to block the direct LOS
path between Txs and Rxs. Although, these controlled human movements are not
expected to be the regular real life scenarios, it is very crucial for better under-
standing of human body shadowing effect in an indoor environment. This is also
the platform for the next investigation, namely randomly moving people, which is
detailed in the next Chapter 7. Deterministically moving pedestrians allow us to
identify the precise location and samples, when LOS path between Txs and Rxs is
blocked and freed by the human body. Hence, the immediate effect on the MIMO-
OFDM channel capacity and MIMO-OFDM channel capacity dynamic range, due
to the controlled human movement, can be observed and modeled. All the results
in this chapter have been reported in [5, 47, 112–114].
Chapter 6 Analysis of Results for Deterministic Scenarios
6.1 Introduction 88
6.1 Introduction
Temporal variation of MIMO-OFDM channel characteristics due to human move-
ment within a given trajectory is presented in this section. As discussed in Chap-
ter 4, dominant LOS path was present for much of measurement time, and as such
the variation of MIMO-OFDM channel capacity was small for most of the time.
A significant variation has been observed when pedestrians are blocking the direct
LOS path between TXs and Rxs. Previously, many researchers have characterized
various LOS & NLOS effects on indoor radio channels [42, 43], MIMO channels
[115, 116] and MIMO-OFDM channels [17, 66]. In these investigations, authors
reported their measurement campaigns and different channel characterization con-
sidering various indoor environments. But, none of these investigations have ana-
lyzed the MIMO-OFDM channel characteristics in populated indoor environment.
Simulated analysis of the human body shadowing effects, on narrowband indoor
MIMO channels have been analyzed and reported in [4, 16, 26]. So far, to the best
of our knowledge, no research has characterize the indoor MIMO-OFDM channel
considering real human bodies in a systematic manner.
As human body presence creates a significant variation in indoor channels, it is
crucial to take human body shadowing effect into account while designing indoor
WLAN. This thesis focuses on the experimental MIMO-OFDM channel measure-
ments, considering pedestrian effects in indoor environments. It also captures the
trend of MIMO-OFDM channel capacity and MIMO-OFDM channel capacity dy-
namic range change with increasing number of antenna elements and pedestrians.
All measurement scenarios are verified through replicated simulations implemented
during this investigation. Congregated deterministic (for both Fixed SNR and Fixed
Tx) results show a general trend of increment in MIMO-OFDM channel capacity
and channel capacity dynamic range, with the number of people present, as well
as with the number of antenna elements combinations. Measurement and anal-
ysis of MIMO-OFDM channel capacity have been carried out considering up to
three pedestrians in an indoor environment. Similar MIMO-OFDM channel char-
Chapter 6 Analysis of Results for Deterministic Scenarios
6.1 Introduction 89
acteristics in populated indoor environment have also been captured in replicated
simulation results.
Fig. 6.1 portrays the entire deterministic measurement site with predetermined
6m trajectory, antenna array locations, and pedestrians. Further details can be ac-
quired, regrading measurement scenarios and procedures, from Chapter 4 and sim-
ulations from Chapter 5.
Figure 6.1: The 6m Preset Trajectory for Deterministic Measurement Scenarios
In this chapter, measurement and simulation results are grouped under the fol-
lowing sections:
1. MIMO-OFDM average channel capacity
2. MIMO-OFDM channel capacity cumulative distribution function
3. MIMO-OFDM capacity dynamic range
The results then followed by the Discussion section, where the detail comparison
and analysis of the measured and simulated findings have been presented.
Chapter 6 Analysis of Results for Deterministic Scenarios
6.2 MIMO-OFDM Channel Measurements 90
6.2 MIMO-OFDM Channel Measurements
Following the work of Ziri-Castro et al. [4, 26] on the variations on channel capacity
for a MIMO system due to the presence of pedestrians, here MIMO-OFDM channel
capacity and MIMO-OFDM channel capacity dynamic range for 2 × 2, 3 × 3, and
4×4 antenna configurations in LOS environments have been analyzed, using Fixed
SNR and Fixed Tx power. It has been found that, for Fixed SNR the average MIMO-
OFDM channel capacity as a function of pedestrians number increases and for Fixed
Tx power it decreases.
Time variation characteristics due to the pedestrians movement are analyzed,
based on measured MIMO-OFDM channels at 5.24 GHz band with 40 MHz band-
width. Pedestrians walking and crossing the direct LOS path between Tx and Rx
have been analyzed. The number of pedestrians ranges from zero to three. In
every case, 100 time samples are collected with 114 OFDM sub-carriers and 16
MIMO sub-channels. Preliminary analysis shows, the mean channel capacity and
the dynamic range of the received power increases with the number of pedestrians
present within the indoor environment. During measurements approximately 2 sam-
ples/sec have been collected and pedestrians speed has been limited to a maximum
of 0.5 meter/sec. In this chapter, reported results are in terms of channel capacity
dynamic range as a function of antenna combinations and number of pedestrians.
Since this thesis focuses on the analysis of the channel capacity which does not de-
pend on the time resolution of the system, the pedestrian speed has no impact on the
analyzed results. Moreover, substantial amount of channel data has been collected
for establishing a reliable model.
Fig. 6.2 shows a sample of the total relative received power for 4× 4 LOS mea-
surements in Fixed Tx scenario, with pedestrians ranging from none to three. A
significant decrease in the received power is more apparent, when pedestrians are
blocking the direct LOS. This is due to body-shadowing effects on MIMO-OFDM
channels, and can easily be separated from the 0 or vacant scenario. With increasing
numbers of human body in the indoor environment, higher variation in relative re-
Chapter 6 Analysis of Results for Deterministic Scenarios
6.2 MIMO-OFDM Channel Measurements 91
0 20 40 60 80 100−11
−10.5
−10
−9.5
−9
−8.5
−8
−7.5
−7
Sample index
Rel
ativ
e po
wer
(dB
)
0123
Figure 6.2: A Sample of 4x4 Relative Received Power for Fixed Tx Scenario [5]
ceived power is evident. An increase in dynamic range is also found in conjunction
with an increase in the number of pedestrians present in the measurement locations
for deterministic LOS and Fixed SNR scenario. In Fixed Tx power scenario, re-
duction in receivable power causes the relative power to drop with more number
of people, while they are blocking the direct LOS path. Received power dynamic
range increased approximately 3 dB from the vacant scenario compared to the three
pedestrians scenario. Note that 3 dB reduction of received power is over 40 MHz
bandwidth in the highly multipath environment.
A sample of MIMO-OFDM channel relative received power is shown in Fig. 6.3
considering 16 subchannels. Here x axis is the frequency ( MHz) and y axis is the
relative power (dB). From the graph, several frequency selective fading for each
subchannel is observed as pedestrians block the LOS path. Individual subplot is
showing a different combination of 4 × 4 MIMO-OFDM system. Among the sub-
plots in Fig. 6.3, a deep fading close to 20 dB for Tx4Rx1 has been observed. In
addition, a comparatively flat variation in relative power as a function of frequency
Chapter 6 Analysis of Results for Deterministic Scenarios
6.2 MIMO-OFDM Channel Measurements 92
Figure 6.3: A Sample of the 4x4 MIMO-OFDM Sub-Channels when pedestrian is
blocking LOS path[5]
Chapter 6 Analysis of Results for Deterministic Scenarios
6.2 MIMO-OFDM Channel Measurements 93
has been observed for Tx4Rx3. This exemplifies a kind of variation expected in
indoor environment for different MIMO channels. Note that, a similar frequency
selective fading was observed when pedestrian was not blocking the LOS path, due
to highly multipath environment.
6.2.1 Average Channel Capacity
Approximately 50 plots of average channel capacity have been acquired, from which
only six graphs have been presented in this section, to show the change in MIMO-
OFDM average channel capacity due to pedestrians effect in indoor environments.
Individual MIMO-OFDM average channel capacity for Fixed SNR and Fixed Tx
has been obtained averaging over frequency, over number of samples and over an-
tenna combinations (such as, for 4 × 4 system there will be 4 combinations, for
3 × 3 system there will be 16 combinations, for 2 × 2 system there will be 36
combinations).
In general for Fixed SNR, a sudden plunge in MIMO-OFDM channel capacity
has been observed, as soon as the pedestrians block the direct LOS path followed by
an immediate increase. This is due to the reduction in receivable power when human
body blocks the direct LOS path, which causes immediate lessening in MIMO-
OFDM channel capacity. In Fixed SNR criteria to compensate this reduction of
receivable power, the transmitter increases the transmitting power, which shows the
immediate uplifting of the channel capacity. On the other hand, for Fixed Tx the
lessening of channel capacity is more noticeable with more number of people. This
reduction lasts as long as the pedestrians are blocking the LOS path and return to
normal, when human body go out of the direct blocking region.
Fig. 6.4(a) shows the deterministic MIMO-OFDM channel capacity for Fixed
SNR scenario, assuming a Fixed SNR of 15dB for a 4 × 4 system. Additionally,
Fig. 6.4(b) shows the deterministic MIMO-OFDM channel capacity for Fixed Tx
scenario for a 4× 4 system. Here x axis represents the time sample in seconds and
y axis represents the MIMO-OFDM average channel capacity in bits/sec/Hz. A sig-
Chapter 6 Analysis of Results for Deterministic Scenarios
6.2 MIMO-OFDM Channel Measurements 94
0 20 40 60 80 10012
12.5
13
13.5
14
14.5
15
15.5
16
16.5
Sample index (4x4)
Fixe
d SN
R A
vera
ge c
apac
ity (
bps/
Hz)
0
1
2
3
(a) Capacity Analysis for Deterministic Fixed SNR scenarios
0 20 40 60 80 10012
12.5
13
13.5
14
14.5
15
15.5
16
16.5
Sample index (4x4)
Fixe
d T
x A
vera
ge c
apac
ity (
bps/
Hz)
0
1
2
3
(b) Capacity Analysis for Deterministic Fixed Tx scenarios
Figure 6.4: Capacity Analysis for Deterministic Scenarios (4× 4)
Chapter 6 Analysis of Results for Deterministic Scenarios
6.2 MIMO-OFDM Channel Measurements 95
0 20 40 60 80 100
10.8
11
11.2
11.4
11.6
11.8
12
12.2
Sample index (3x3)
Fixe
d SN
R A
vera
ge c
apac
ity (
bps/
Hz)
0
1
2
3
(a) Capacity Analysis for Deterministic Fixed SNR scenarios
0 20 40 60 80 1009
9.5
10
10.5
11
11.5
12
12.5
Sample index (3x3)
Fixe
d T
x A
vera
ge c
apac
ity (
bps/
Hz)
0
1
2
3
(b) Capacity Analysis for Deterministic Fixed Tx scenarios
Figure 6.5: Capacity Analysis for Deterministic Scenarios (3× 3)
Chapter 6 Analysis of Results for Deterministic Scenarios
6.2 MIMO-OFDM Channel Measurements 96
0 20 40 60 80 1007.7
7.8
7.9
8
8.1
8.2
8.3
8.4
8.5
Sample index (2x2)
Fixe
d SN
R A
vera
ge c
apac
ity (
bps/
Hz)
0
1
2
3
(a) Capacity Analysis for Deterministic Fixed SNR scenarios
0 20 40 60 80 1006.5
7
7.5
8
8.5
9
Sample index (2x2)
Fixe
d T
x A
vera
ge c
apac
ity (
bps/
Hz)
0
1
2
3
(b) Capacity Analysis for Deterministic Fixed Tx scenarios
Figure 6.6: Capacity Analysis for Deterministic Scenarios (2× 2)
Chapter 6 Analysis of Results for Deterministic Scenarios
6.2 MIMO-OFDM Channel Measurements 97
nificant variation in channel capacity with the number of pedestrians present in the
environment has been observed. Variations in channel capacity are more noticeable
at sample index 60-80, when pedestrians were directly obstructing the LOS path
between Txs and Rxs. For Fixed SNR with three pedestrians using a 4 × 4 array
an increase of 2 bits/sec/Hz in channel capacity is observed relative to the vacant
scenario, due to the increase in multipath conditions caused by body-shadowing ef-
fects. This shows that, the use of MIMO-OFDM is effective in compensating for
the presence of pedestrians. On the other hand, there is a decrease of 2 bits/sec/Hz
in channel capacity when considering Fixed Tx scenarios. This decrease is due to
Fixed Tx power, which limits the receivable power in populated indoor environ-
ments.
Further analysis on MIMO-OFDM channel capacity as a function of number of
pedestrians with 2× 2 (Fig 6.6) and 3× 3 (Fig. 6.5) antenna combination also show
the trend of increasing capacity in Fixed SNR and trend of decreasing capacity in
Fixed Tx power. In both cases, x axis represents the time sample in seconds and
y axis represents the MIMO-OFDM average channel capacity in bits/sec/Hz. For
2× 2 (Fig 6.6) antenna combination, an increase up to 1 bits/sec/Hz for Fixed SNR
and a decrease up to 2 bits/sec/Hz for Fixed Tx has been recorded. In addition, for
3 × 3 (Fig. 6.5) antenna combination, an increase up to 0.25 bits/sec/Hz for Fixed
SNR and a decrease up to 1.75 bits/sec/Hz for Fixed Tx has been noted.
In all Fixed SNR and Fixed Tx cases, an increase of MIMO-OFDM average
channel capacity has been observed with more number of antenna combinations
compared to less number of antenna combinations. The increase in average chan-
nel capacity with the numbers of antenna combination is due to the increase of
receivable power and the increase in multipath components at the receiver array.
A maximum increase in average channel capacity of approximately 6 bits/sec/Hz
has been recorded, when using a 2 × 2 comparing to a 4 × 4 array, for both Fixed
SNR and Fixed Tx. Moreover, in all Fixed SNR scenarios, the preliminary drop
suggests the reduction in receivable power when people block the LOS path, which
Chapter 6 Analysis of Results for Deterministic Scenarios
6.2 MIMO-OFDM Channel Measurements 98
compensates immediately after to keep the SNR fixed.
Antenna combination 1ppl 2ppl 3ppl
2× 2 (FSNR) 8.08 8.05 8.09
3× 3 (FSNR) 11.28 11.29 11.32
4× 4 (FSNR) 14.21 14.34 14.25
2× 2 (FTX) 8.08 8.09 8.22
3× 3 (FTX) 11.37 11.38 11.57
4× 4 (FTX) 14.32 14.40 14.55
Table 6.1: Measured MIMO-OFDM Channel Capacity for Deterministic Fixed
SNR and Fixed Tx Power
This investigation has captured the average capacity. Results are summarized in
Table 6.1. Presented values in Table 6.1 are averaged over frequency, over number
of samples, over number of antenna combinations. Table 6.1 shows the increasing
trend of the average MIMO-OFDM channel capacity with the number of people,
as well as with the number of antenna combinations. From the Table 6.1 with up
to three people, a linear increase of approximately 3 bits/sec/Hz in MIMO-OFDM
channel capacity has been observed for each additional antenna element, from 2 to 3
antenna elements, and for 3 to 4 antenna elements, for both Fixed SNR and Fixed Tx
criteria. As a result, an increment of approximately 77% has been observed, while 4
antenna elements are used, compared with 2 antenna elements. Moreover, there is a
negligible increment of 0.04 bits/sec/Hz (this could be considered as a experimental
noise) when the number of pedestrians increases from 1 to 3 in Fixed SNR and a
maximum increment of 0.25 bits/sec/Hz for Fixed Tx power. This variation in both
Fixed SNR and Fixed Tx scenarios are due to averaging factors, as large number of
sample values are reflected on dominant LOS path.
Chapter 6 Analysis of Results for Deterministic Scenarios
6.2 MIMO-OFDM Channel Measurements 99
12 13 14 15 16 170
10
20
30
40
50
60
70
80
90
100
Fixed SNR MIMO−OFDM capacity (bits/s/Hz)
CD
F (
%)
0
1
2
3
(a) CDF Analysis for Deterministic Fixed SNR
12 13 14 15 16 170
10
20
30
40
50
60
70
80
90
100
Fixed Tx MIMO−OFDM capacity (bits/s/Hz)
CD
F (
%)
0
1
2
3
(b) CDF Analysis for Deterministic Fixed Tx
Figure 6.7: Measured CDF Analysis for Deterministic Fixed SNR and Fixed Tx
Chapter 6 Analysis of Results for Deterministic Scenarios
6.2 MIMO-OFDM Channel Measurements 100
6.2.2 Channel Capacity Cumulative Distribution Function
To accurately model indoor MIMO-OFDM channels, it is important to thoroughly
understand the behavior of the channel capacity. To gain insight in this matter,
cumulative probability distribution functions (CDFs) of measured MIMO-OFDM
average channel capacity are presented in this section. Fig. 6.7 shows CDFs of
MIMO-OFDM capacity for the LOS scenario using the Fixed SNR (Fig. 6.7(a)) and
Fixed Tx power criteria (Fig. 6.7(b)). Here, x axis represents the MIMO-OFDM ca-
pacity in bits/sec/Hz and y axis represents CDF in %. As expected, the variation
of the capacity for the vacant case is minimal, while the introduction of even one
pedestrian changes the CDF dramatically. The incremental effect of having more
than one pedestrian seems less significant, having similar CDFs for one, two, or
three pedestrians. The presence of the pedestrian tends to increase the capacity for
Fixed SNR (this can be observed by the fact that the capacity with the pedestrian
always surpasses that for the vacant scenario above 20% CDF), while it appears
to increase with the number of people at above average, while decreasing with the
number of people at below average for Fixed Tx power. This is due to the blocking
of the direct LOS path by the pedestrians which causes the reduction of receiv-
able power. Comparing the spread of CDFs for vacant case and one, two, or three
pedestrians cases, the pedestrians are found to cause some temporal variation when
they are crossing the LOS path. In addition, an increase in spread of CDFs with
increasing number of people has been noted.
6.2.3 Channel Capacity Dynamic Range
In this section, measured channel capacity dynamic range is analyzed. The analysis
is performed based on the 90% capacity dynamic range which is ,as defined in
Chapter 2, Section 2.6, the difference between the top 95% and the bottom 5%
values, in order to remove extreme cases.
The measured 90% average capacity dynamic range values are summarized in
Table 6.2 for both measured Fixed SNR and Fixed Tx Power scenarios. In general,
Chapter 6 Analysis of Results for Deterministic Scenarios
6.2 MIMO-OFDM Channel Measurements 101
Antenna combination 1ppl 2ppl 3ppl
2× 2 (FSNR) 0.94 0.99 1.14
3× 3 (FSNR) 1.10 1.10 1.15
4× 4 (FSNR) 1.42 1.16 1.58
2× 2 (FTX) 1.71 1.83 2.37
3× 3 (FTX) 1.91 2.02 2.53
4× 4 (FTX) 2.12 2.34 2.58
Table 6.2: Measured MIMO-OFDM Channel Capacity Dynamic Range for Deter-
ministic Fixed SNR and Fixed Tx Power(90%)
it has been observed that dynamic range increases with the number of pedestrians in
indoor environments. This can be explained by larger number of pedestrian further
attenuating the LOS path.
Moreover, with an increasing number of antenna elements, the dynamic range
also increases. This may be surprising, as typically with a larger number of antennas
more diversity is expected, hence more robust channel can be found. However,
the dynamic range increase with the number of antenna elements because the base
channel capacity increases with the number of antenna elements.
For both Fixed SNR and Fixed Tx the table shows an increment of approxi-
mately 51% for Fixed SNR and 23% for Fixed Tx while 4 antenna elements are
used, compared with 2 antenna elements. Additionally, there is a maximum incre-
ment of 0.2 bits/sec/Hz when the number of pedestrians increases from 1 to 3 in
Fixed SNR, and a maximum increment of 0.66 bits/sec/Hz for Fixed Tx power. The
increment in dynamic range as a function of number of people is due to the fact of
variation in channel capacity, as more number of pedestrians crossing direct LOS
path between Txs and Rxs. Analysis of MIMO-OFDM channel capacity dynamic
range has projected an incremental trend with an increasing number of people, as
well as with an increasing number of growing antenna elements.
Chapter 6 Analysis of Results for Deterministic Scenarios
6.3 MIMO-OFDM Channel Simulations 102
6.3 MIMO-OFDM Channel Simulations
For decades, simulation has been utilized to predict and validate the measured data
for wireless communication [16, 107, 117, 118]. In this thesis, extensive simula-
tion using GO based Frustum Ray Tracing Technique (FRTT) has been conducted.
Using simulation, capacity dynamic range for 2× 2, 3× 3, and 4× 4 antenna con-
figurations in LOS environments using Fixed SNR and Fixed Tx power has been
derived. Simulations capture the similar increasing trend of MIMO-OFDM channel
capacity as a function of antenna combination for both Fixed SNR and Fixed Tx.
Due to the higher number of simulation samples with dominant LOS path, a very
small variation has been observed in MIMO-OFDM channel capacity as a function
of number of people. In addition, because of the simplicity of the replicated body
model a similar impression emerged, while walking together in a given trajectory
and perpendicular fashion to each other.
The simulations, which replicate the measurement scenarios, present time vari-
ation characteristics due to pedestrian movement, considering 5.24 GHz band with
40 MHz bandwidth. Placing a computer generated pedestrian model between Txs
and Rxs MIMO-OFDM channels have been analyzed. In total, 0-3 pedestrians have
been considered for deterministic simulation with 400 receiver antenna locations,
114 OFDM sub-carriers and 16 MIMO sub-channels. A total of approximately 79
million SISO channels are obtained. Preliminary analysis show agreement between
measurements and simulations. Simulated results show the channel capacity dy-
namic range increases with the number of pedestrians and antenna elements, within
the populated indoor environment as found in measurements.
6.3.1 Average Channel Capacity
The simulation environment replicates the physical characteristics of the entire room
environment, where measurements were conducted, considering all different build-
ing blocks as well as human bodies. Permeability and conductivity of the building
Chapter 6 Analysis of Results for Deterministic Scenarios
6.3 MIMO-OFDM Channel Simulations 103
1ppl 2ppl 3ppl5
6
7
8
9
10
Ave
rage
Cha
nnel
Cap
acity
(bi
ts/s
/Hz)
(a) Simulated FSNR
4x4 3x3 2x2
1ppl 2ppl 3ppl5
6
7
8
9
10
Ave
rage
Cha
nnel
Cap
acity
(bi
ts/s
/Hz)
(b) Simulated FTx
4x4 3x3 2x2
Figure 6.8: Simulated Average Capacity with Different Number of Pedestrians and
Antennas.
Chapter 6 Analysis of Results for Deterministic Scenarios
6.3 MIMO-OFDM Channel Simulations 104
Antenna combination 1ppl 2ppl 3ppl
2× 2 (FSNR) 5.99 5.97 5.94
3× 3 (FSNR) 7.73 7.78 7.73
4× 4 (FSNR) 9.37 9.36 9.44
2× 2 (FTX) 5.55 5.56 5.53
3× 3 (FTX) 6.38 6.42 6.42
4× 4 (FTX) 8.48 8.52 8.58
Table 6.3: Simulated MIMO-OFDM Channel Capacity Dynamic Range for Deter-
ministic Fixed SNR and Fixed Tx Power
materials and different human tissues classification have been taken into account.
Detailed tissue dielectric parameters are obtained from [16, 99]. As expected, a very
minor difference in predicted average channel capacity has been observed compar-
ing the results of one and three pedestrians. As described earlier, this is considered
to be due to the distribution of the human body model and the presences of a domi-
nant LOS path during most of the simulations, which play a key role for obtaining
such near flat line capacity values. However, increasing number of antenna arrays
show prominent increase in average channel capacity. Fig. 6.8 shows the average
MIMO-OFDM channel capacity with different numbers of pedestrians and antenna
combinations in different subplots. In these figures, regular (100%) values of the
average channel capacity have been depicted. The flat response is considered to be
due to the fact that MIMO-OFDM channel capacity is further averaged over time,
where most of the results are LOS scenarios in which the LOS is not blocked by the
pedestrian for most of the time. Hence, the average MIMO-OFDM channel capac-
ity does not vary with different number of pedestrian. The simulated scenarios also
capture higher channel capacity values for Fixed SNR than Fixed Tx. This is due to
the increase in receivable power as more pedestrians are blocking the LOS path. A
rise of approximately 3 bits/sec/Hz in average channel capacity has been recorded,
where more antenna elements are deployed.
Chapter 6 Analysis of Results for Deterministic Scenarios
6.3 MIMO-OFDM Channel Simulations 105
Table 6.3 shows the average MIMO-OFDM channel capacity with the number
of antenna combinations and the number of pedestrians. An increment in MIMO-
OFDM average channel capacity of approximately 55% for Fixed SNR and an in-
crement of approximately 53% for Fixed Tx has been found, while 4 antenna ele-
ments are used, compared with 2 antenna elements. From the presented numerical
results a maximum variation of 0.08 bits/sec/Hz for Fixed SNR and 0.10 bits/sec/Hz
for Fixed Tx has been recorded, while considering increasing number of people.
The findings reflect that, longer presence of a dominating LOS path and noise free
conditions have a strong effect on the variation of MIMO-OFDM channel capacity
in the populated indoor environment.
6.3.2 Channel Capacity Cumulative Distribution Function
The simulated channel capacity CDFs capture the trend of the measured results, by
preserving broader CDF spread, when introducing more people in the indoor envi-
ronment. Fig. 6.9 shows the CDF plots for the simulated MIMO-OFDM average
channel capacity using the Fixed SNR (Fig. 6.9(a)) and Fixed Tx power (Fig. 6.9(b))
criteria. The variation of the capacity for the Fixed SNR criteria preserves consis-
tency with a higher spread around the mean value for more number of pedestrians
present in the environment. In addition, Fixed Tx also shows a increasing spread
around mean average capacity, when more number of people are introduced in the
environment. However, the effects of having more pedestrians seem less significant,
having similar CDFs for different numbers of pedestrians, for Fixed SNR criteria. In
this case, the presence of the pedestrian tends to increase average channel capacity
less significantly (up to 1 bits/sec/Hz from 0 to 3 pedestrians). While in Fixed Tx,
the MIMO-OFDM channel capacity always surpasses that for the vacant scenario
above approximately 30% with every pedestrian. In addition, when comparing the
spread of channel capacity CDFs for vacant, one, two, or three pedestrians scenar-
ios. Since the pedestrians cause temporal variations when they are obstructing the
LOS path. Hence, for both Fixed SNR and Fixed Tx a broader spread with increas-
Chapter 6 Analysis of Results for Deterministic Scenarios
6.3 MIMO-OFDM Channel Simulations 106
7 7.5 8 8.5 9 9.5 10 10.5 11 11.5 120
10
20
30
40
50
60
70
80
90
100
Fixed SNR MIMO−OFDM channel capacity (bits/s/Hz)
CD
F (
%)
0 ppl1ppl2ppl3ppl
(a) Average MIMO-OFDM Channel Capacity CDF using Fixed SNR Criteria
4 5 6 7 8 9 10 11 12 13 14 150
10
20
30
40
50
60
70
80
90
100
Fixed Tx MIMO−OFDM channel capacity (bits/s/Hz)
CD
F (
%)
0 ppl1ppl2ppl3ppl
(b) Average MIMO-OFDM Channel Capacity CDF using Fixed Tx Criteria
Figure 6.9: Channel Capacity CDF Plots for Simulated Deterministic Fixed SNR
and Fixed Tx Scenarios
Chapter 6 Analysis of Results for Deterministic Scenarios
6.3 MIMO-OFDM Channel Simulations 107
Antenna combination 1ppl 2ppl 3ppl
2× 2 (FSNR) 0.82 1.04 1.16
3× 3 (FSNR) 0.82 1.10 1.33
4× 4 (FSNR) 0.90 1.43 1.83
2× 2 (FTX) 1.76 2.43 3.03
3× 3 (FTX) 1.89 2.88 3.79
4× 4 (FTX) 1.92 3.28 4.55
Table 6.4: Simulated MIMO-OFDM Channel Capacity Dynamic Range for Deter-
ministic Fixed SNR and Fixed Tx Power (90%)
ing number of people is evident from the CDF plots.
6.3.3 Channel Capacity Dynamic Range
Simulated MIMO-OFDM channel capacity dynamic range results have been in-
cluded in Table 6.4. The presented results indicate an incremental trend in channel
capacity dynamic range with growing number of people in accordance to measured
results. Similarly, results show a constant increase in channel capacity dynamic
range with increasing number of antenna elements. Table 6.4 shows the MIMO-
OFDM channel capacity dynamic range for simulated Fixed SNR and Fixed Tx
Power scenarios. Here it has been observed that dynamic range increases with the
number of pedestrians in indoor environments. Moreover, with increasing number
of antenna elements, the dynamic range also increases. The table shows a max-
imum increment of 57% for Fixed SNR and a maximum increment of 50% for
Fixed Tx while 4 antenna elements are used, compared with 2 antenna elements.
Furthermore, there is a maximum increment of 0.9 bits/sec/Hz when the number of
pedestrians increases from 1 to 3 in Fixed SNR and a maximum increment of 2.63
bits/sec/Hz for Fixed Tx power.
Chapter 6 Analysis of Results for Deterministic Scenarios
6.4 Measurements Vs. Simulations 108
6.4 Measurements Vs. Simulations
In this section, a comparison between measured and simulated findings is presented.
The comparison has been distributed in different sections namely, MIMO-OFDM
Channel Capacity, MIMO-OFDM Channel Capacity Dynamic Range and finally
an empirical analysis under the section Capacity Dynamic Range Vs Number of
Pedestrians. In the presented empirical analysis, both number of pedestrians and an-
tenna element combination have been considered to present the trend of the MIMO-
OFDM channel capacity dynamic range.
6.4.1 MIMO-OFDM Channel Capacity
MIMO-OFDM average channel capacity simulated results clearly capture the trend
found in the measurement conducted in the indoor populated environments. Due to
the lack of modeling diffuse scattering and small objects in the simulation, both of
which would increase the number of propagation paths from the transmitter to the
receiver and thus would increase the channel capacity, the simulated average chan-
nel capacity results show a little lower value than the conducted measured average
capacity.
Fig. 6.10(a) shows the average channel capacity comparison for measured and
simulated Fixed SNR scenarios, while Fig. 6.10(b) shows the Fixed Tx scenarios.
Here, x axis being the number of people and y axis being the median MIMO-OFDM
average channel capacity. In both cases, there is a similar increasing trend, when
more numbers of antenna elements are introduced. Moreover, the results from the
measurements closely match the increasing trend of the simulation results. In gen-
eral, the measured Fixed Tx MIMO-OFDM average channel capacity values found
to be larger than the measured Fixed SNR.
Table 6.5 for simulated and measured MIMO-OFDM channel capacity also
shows the trend of very minor fluctuations when increasing pedestrians in the in-
door environment as well as a significant increase in channel capacity when more
Chapter 6 Analysis of Results for Deterministic Scenarios
6.4 Measurements Vs. Simulations 109
1ppl 2ppl 3ppl5
10
15M
edia
n C
apac
ity
(a) Simulation Result in Fixed SNR
2x2 3x3 4x4
1ppl 2ppl 3ppl5
10
15
Med
ian
Cap
acity
(b) Measurement Result in Fixed SNR
(a) Average Channel Capacity Comparison for Fixed SNR
1ppl 2ppl 3ppl5
10
15
Med
ian
Cap
acity
(a) Simulation Result in Fixed Tx
2x2 3x3 4x4
1ppl 2ppl 3ppl5
10
15
Med
ian
Cap
acity
(b) Measurement Result in Fixed Tx
(b) Average Channel Capacity Comparison for Fixed Tx
Figure 6.10: Average Channel Capacity Comparison for Measured and Simulated
Fixed SNR and Fixed Tx
Chapter 6 Analysis of Results for Deterministic Scenarios
6.4 Measurements Vs. Simulations 110
antenna elements have been deployed. Interestingly, in both simulated and mea-
sured cases for MIMO-OFDM average channel capacity, variations in the order of
decimal values have been observed, when numbers of people increased in the indoor
environment.
Measured Simulated
Antenna combination 1ppl 2ppl 3ppl 1ppl 2ppl 3ppl
2× 2 (FSNR) 8.08 8.05 8.09 6.00 5.98 5.94
3× 3 (FSNR) 11.28 11.29 11.32 7.73 7.78 7.73
4× 4 (FSNR) 14.21 14.34 14.25 9.38 9.37 9.44
2× 2 (FTX) 8.08 8.09 8.22 5.55 5.56 5.53
3× 3 (FTX) 11.37 11.38 11.57 6.38 6.43 6.42
4× 4 (FTX) 14.32 14.40 14.55 8.48 8.52 8.59
Table 6.5: Measured and Simulated MIMO-OFDM Channel Capacity for Deter-
ministic Fixed SNR and Fixed Tx Power
6.4.2 MIMO-OFDM Channel Capacity Dynamic Range
Fig. 6.11 shows the variation in the measured and simulated capacity dynamic range
as a function of the number of pedestrians and antennas for Fixed SNR and Fixed
Tx power capacity. In general, it has been observed that the measured and simu-
lated 90% capacity dynamic range is larger for Fixed Tx power criteria than for the
Fixed SNR. This is due to human body shadowing effects being much more notice-
able than the expected increase in capacity due to the decorrelation of the channel
caused by the obstruction of the direct LOS path in Fixed Tx power criteria. In
Fig. 6.11, the increasing trend of measured capacity dynamic range with the num-
ber of pedestrians is also captured by the simulated results. With a growing number
of pedestrians, a larger reduction of the LOS power is introduced, and hence a larger
dynamic range has resulted for the Fixed Tx power capacity. For the Fixed SNR ca-
pacity, the blocking of the LOS path by a larger number of pedestrians introduces
Chapter 6 Analysis of Results for Deterministic Scenarios
6.4 Measurements Vs. Simulations 111
SM SS TM TS0
1
2
3
4
5
Dyn
amic
ran
ge (
bits
/s/H
z)
2 × 2
3 p2 p1 p
SM SS TM TS0
1
2
3
4
53 × 3
SM SS TM TS0
1
2
3
4
54 × 4
Figure 6.11: Measured and Simulated Dynamic Range Variation with Different
Numbers of Pedestrians and Antennas. SM: Fixed SNR, measurement. SS: Fixed
SNR, simulation. TM: Fixed Tx power, measurement. TS: Fixed Tx power, simu-
lation.
Chapter 6 Analysis of Results for Deterministic Scenarios
6.4 Measurements Vs. Simulations 112
SM SS TM TS0
10
20
30
40
50
60N
orm
aliz
ed d
ynam
ic r
ange
(%
)2 x 2
SM SS TM TS0
10
20
30
40
50
603 x 3
SM SS TM TS0
10
20
30
40
50
604 x 4
3 p2 p1 p
Figure 6.12: Percentage Dynamic Range Variation with Different Number of Pedes-
trian and Antennas. SM: Fixed SNR, measurement. SS: Fixed SNR, simulation.
TM: Fixed Tx power, measurement. TS: Fixed Tx power, simulation.
a further decorrelation of the channel, and the Fixed SNR capacity dynamic range
also increases with the number of pedestrians. This has been confirmed by both
measurements and simulations for 0 to 3 pedestrians.
MIMO-OFDM channels corresponding to 2×2 and 3×3 are extracted from 4×4
results both for the measurement and simulation, using all possible antenna combi-
nations. Note that the results include different antenna spacing by using adjacent
or diagonal antenna elements. Since the antenna spacing is large, the exact antenna
spacing is considered to have small effects on the results [15]. The dynamic range
results for 2×2 and 3×3 are also averaged over different combinations to provide
representative values.
It has also been observed in both measurements and simulations, the MIMO-
OFDM capacity dynamic range slightly increases with the number of antennas used.
Chapter 6 Analysis of Results for Deterministic Scenarios
6.4 Measurements Vs. Simulations 113
The increase of the dynamic range as a function of the number of antennas is con-
sidered to be due to the increase in the MIMO-OFDM capacity with the number
of antennas. To verify this point, the normalized dynamic range, which is the ra-
tio of the dynamic range value with respect to the median capacity, is plotted in
Fig. 6.12. While the trend of increasing dynamic range with the number of pedes-
trians is maintained, the relationship between the normalized dynamic range and the
number of antennas is reversed. Here it has been observed that, when the dynamic
range is scaled by the median capacity, a larger number of antennas tend to provide
a smaller variation in the normalized dynamic range. This is considered to be due
to increased path diversity by a larger number of MIMO channels with the larger
number of antennas. While it may be desired to obtain more stable (less absolute
dynamic range) MIMO-OFDM channel performance with increase in the number
of antennas, both measurements and simulations show that the absolute capacity
dynamic range slightly increases with the number of antenna used. The system
designer needs to consider how one might mitigate the absolute dynamic range to
provide stable performance in the presence of moving objects with increasing num-
ber of antennas.
A large deviation of the simulation results from the measured results is observed
for Fixed Tx power criterion. This is considered to be due to the simplicity of the
models of moving human bodies and of the environment employed in the simula-
tions.
6.4.3 Capacity Dynamic Range vs Number of Pedestrians
An empirical analysis has been used to estimate variations of MIMO-OFDM chan-
nel capacity in the presence of pedestrians. Empirical methods have been utilized by
researchers (e.g.[99, 100]) to estimate and model channel propagation for MIMO-
OFDM systems. Specifically, in this thesis linear and quadratic regression analysis
of MIMO-OFDM channel capacity dynamic range against number of pedestrians
are conducted. In both cases, the first order derivative gives the gradient or the
Chapter 6 Analysis of Results for Deterministic Scenarios
6.4 Measurements Vs. Simulations 114
FSNR FTX
2× 2 (SimLin) 0.17× x + 0.67 0.64× x + 1.14
2× 2 (SimQua) −0.045× x2 + 0.35× x + 0.53 −0.03× x2 + 0.76× x + 1.03
2× 2 (MesLin) 0.10× x + 0.82 0.33× x + 1.32
2× 2 (MesQua) 0.047× x2 − 0.89× x + 0.98 0.21× x2 − 0.51× x + 2.02
3× 3 (SimLin) 0.26× x + 0.57 0.95× x + 0.96
3× 3 (SimQua) −0.023× x2 + 0.35× x + 0.49 −0.04× x2 + 1.10× x + 0.83
3× 3 (MesLin) 0.03× x + 1.07 0.31× x + 1.54
3× 3 (MesQua) 0.03× x2 − .08× x + 1.16 0.20× x2 − 0.49× x + 2.21
4× 4 (SimLin) 0.47× x + 0.46 1.31× x + 0.62
4× 4 (SimQua) −0.07× x2 + .74× x + .23 −0.04× x2 + 1.47× x + 0.49
4× 4 (MesLin) 0.08× x + 1.23 0.23× x + 1.89
4× 4 (MesQua) 0.34× x2 − 1.3× x + 2.38 0.01× x2 + 0.20× x + 1.92
Table 6.6: Linear and Quadratic Regression for Different deterministic Measured
and Simulated Scenarios (Sim: Simulation, Mes: Measurement, Lin: Linear Re-
gression, Qua: Quadratic Regression, FSNR: Fixed SNR, FTX: Fixed Tx)
greatest rate of change of the contingent variable (dynamic range), depending on
the known variable (number of pedestrians [0-3]).
To establish an empirical and general model for linear regression for all Fixed
SNR scenarios, the mean first order coefficient has been calculated, over all sim-
ulated and measured Fixed SNR data, in addition to different numbers of antenna
combinations (Simulated and Measured 2×2, 3×3, 4×4). Similar calculation has
also been conducted for the Fixed Tx power criteria. In this, the mean first order
coefficient, averaging over all possible Fixed Tx power criteria in addition to the
number of antenna combinations has been derived. Table 6.6 shows all the ana-
lyzed linear and quadratic regression equations for different antenna combinations
and Table 6.7 shows the average linear and quadratic equations over all possible
antenna combinations.
Chapter 6 Analysis of Results for Deterministic Scenarios
6.4 Measurements Vs. Simulations 115
Linear Regression (FSNR) 0.18× x + 0.63
Quadratic Regression (FSNR) 0.05× x2 − 0.003× x + 0.96
Linear Regression (FTX) 0.63× x + 1.01
Quadratic Regression (FTX) 0.05× x2 + 0.42× x + 1.42
Table 6.7: Average Linear and Quadratic Regression for Deterministic Measured
and Simulated Scenarios (FSNR: Fixed SNR, FTX: Fixed Tx)
Fig. 6.13 shows the linear regression plot for deterministic Fixed SNR scenar-
ios. Here x axis is the number of people (up to three people have been considered
for deterministic scenarios) while y axis is the MIMO-OFDM channel capacity dy-
namic range in bit/s/Hz. All possible antenna combinations have been plotted in
the graph. As expected, the linear regression shows an increasing trend of MIMO-
OFDM channel capacity dynamic range with the number of people. Here, the mean
first order linear coefficient is 0.184 over all possible antenna combinations. The
resulted linear regression equation for the deterministic Fixed SNR is
y = 0.184× x + 0.63. (6.1)
For all Fixed SNR scenarios with up to three pedestrians, a positive gradient of
0.184 has been found. This relates to the fact that in deterministic Fixed SNR
scenarios, an expected average increase of 0.184 bit/sec/Hz in channel capacity
dynamic range per additional pedestrian can be estimated, while MIMO-OFDM
system is deployed in a populated indoor environment.
Quadratic regression results are shown in Fig. 6.14. Similar procedures have
been followed for this analysis. An increasing trend is observed in the Fixed SNR
scenarios, with a slope of 0.05. The resulted quadratic regression equation for the
deterministic Fixed SNR is
y = 0.05× x2 − 0.0036× x + 0.96. (6.2)
The first derivative of 6.2 gives a linear gradient of approximately [2×.047×x] (the
second coefficient can be safely ignored due to its negligible value), which shows
Chapter 6 Analysis of Results for Deterministic Scenarios
6.4 Measurements Vs. Simulations 116
1ppl 2ppl 3ppl0.8
1
1.2
1.4
1.6
1.8
2
Dyn
amic
ran
ge (
bits
/s/H
z)
(a) Linear Regression (FSNR)
Simulation / Measurement Result for Deterministic Fixed SNR
2x2sim2x2mes3x3sim linear3x3mes4x4sim4x4mes
Figure 6.13: Linear Regression for Deterministic Fixed SNR
a positive increment of MIMO-OFDM channel capacity dynamic range when the
number of pedestrians ranges from one to three.
In addition, Fig. 6.15 and Fig. 6.16 show linear and quadratic regression results
for Fixed Tx power scenarios. As expected, an incremental trend in capacity dy-
namic range with the number of people has been found. For linear regression in
deterministic Fixed Tx the first order coefficient is 0.628. In addition, for quadratic
regression, the second order coefficient is 0.051.
The resulted linear regression equation for the deterministic Fixed Tx is
y = 0.628× x + 1.1. (6.3)
For all Fixed Tx scenarios with up to three pedestrians, a positive linear regres-
sion gradient of 0.628 has been found. In consequence, for deterministic Fixed
Tx scenarios, an expected average increase of 0.628 bit/sec/Hz in channel capac-
ity dynamic range per additional pedestrian can be predicted, when MIMO-OFDM
systems are deployed in a populated indoor environment.
Chapter 6 Analysis of Results for Deterministic Scenarios
6.4 Measurements Vs. Simulations 117
1ppl 2ppl 3ppl0.8
1
1.2
1.4
1.6
1.8
2D
ynam
ic r
ange
(bi
ts/s
/Hz)
(b) Quadratic Regression (FSNR)
Simulation / Measurement Result for Deterministic Fixed SNR
2x2sim2x2mes quadratic3x3sim3x3mes4x4sim4x4mes
Figure 6.14: Quadratic Regression for Deterministic Fixed SNR
1ppl 2ppl 3ppl1.5
2
2.5
3
3.5
4
4.5
5
Dyn
amic
ran
ge (
bits
/s/H
z)
(a) Linear Regression (FTx)
Simulation / Measurement Result for Deterministic Fixed TX
2x2sim linear2x2mes3x3sim3x3mes4x4sim4x4mes
Figure 6.15: Linear Regression for Deterministic Fixed Tx
Chapter 6 Analysis of Results for Deterministic Scenarios
6.5 Conclusions 118
1ppl 2ppl 3ppl1.5
2
2.5
3
3.5
4
4.5
5D
ynam
ic r
ange
(bi
ts/s
/Hz)
(b) Quadratic Regression (FTx)
Simulation / Measurement Result for Deterministic Fixed TX
2x2sim2x2mes3x3sim3x3mes4x4sim4x4mes quadratic
Figure 6.16: Quadratic Regression for Deterministic Fixed Tx
The resulted quadratic regression equation for the deterministic Fixed Tx is
y = 0.051× x2 + 0.423× x + 1.4. (6.4)
Through first derivative calculation of (6.4) a linear gradient of approximately [2×.051 × x + 0.423] is obtained. That shows a positive increment in MIMO-OFDM
channel capacity dynamic range, when the number of pedestrians ranges from one
to three.
6.5 Conclusions
Measured data has been analyzed for average MIMO-OFDM channel capacity and
capacity dynamic range. Similar analysis has been carried out using GO based
FRTT simulations. Both results show a similar trend of MIMO-OFDM average
channel capacity and channel capacity dynamic range increment, with the number
of people in an indoor environment. Additionally, channel capacity CDF analy-
Chapter 6 Analysis of Results for Deterministic Scenarios
6.5 Conclusions 119
sis of the measured data shows a larger spread for an increasing number of people
(up to 3) over vacant scenarios. A linear regression curve for the average channel
capacity dynamic range against the number of pedestrians shows a positive gradi-
ent of 0.184 bits/sec/Hz in dynamic range for Fixed SNR per additional pedestrian.
Additionally, a positive gradient of 0.63 bits/sec/Hz in channel capacity dynamic
range for Fixed Tx scenarios against the number of pedestrians has been observed.
Moreover, quadratic analysis of deterministic scenarios also show a positive gradi-
ent with increasing number of people. Chapter 7 will discuss the random scenarios
and compare measured and simulated results.
Chapter 6 Analysis of Results for Deterministic Scenarios
6.5 Conclusions 120
Chapter 6 Analysis of Results for Deterministic Scenarios
121
Chapter 7
Analysis of Results for Random
Scenarios
This chapter presents a detailed analysis of the effects of random pedestrian move-
ment in an indoor environment. The chapter also presents a comprehensive compar-
ison between measurement and simulation results for different random scenarios. In
random scenarios, pedestrians are in random motion within a given space between
the Rxs and Txs. An extensive analysis using simulation has been carried out in-
cluding up to 10 human bodies to replicate measurement scenarios.
During measurements, pedestrians have been instructed to move randomly, within
a given square fashion room block without any speed restriction. This random
movement introduces blocking of the direct LOS path at different time, when pedes-
trian intersects the LOS path between Txs and Rxs. Use of such random human
body movement in the experiment and simulation will allow to comprehend the in-
door MIMO-OFDM channel behavior in a more realistic way. From the analysis
of the previous Chapter on deterministic scenarios, a general idea of human body
shadowing effects on MIMO-OFDM system has been presented. Such knowledge
will be very useful for the analysis of the random scenarios, as it turns into a much
complicated process to recollect the precise point of LOS blocking with pedestrians,
when more and more people start moving randomly in between Txs and Rxs.
Chapter 7 Analysis of Results for Random Scenarios
7.1 Introduction 122
To keep the consistency this chapter follows a similar structure as previous deter-
ministic results analysis chapter, starting with measured analysis, verification with
simulated results and finally a comparison between measured and simulated find-
ings.
7.1 Introduction
Pedestrians moving along different trajectories in a random fashion represents a
realistic scenario for indoor environments. General observations from previous
chapter indicate that, even introducing one pedestrian in the indoor environment
can establish a significant variation on MIMO-OFDM channel capacity comparing
with a vacant scenario [119, 120]. For proper characterization of the MIMO-OFDM
channel, consideration of human body movement effects in an indoor environment,
is highly important. In Chapter 6 we have detailed the deterministic measurement
campaign, which focuses on indoor MIMO-OFDM channel measurements and sim-
ulations up to 3 pedestrians walking in a given trajectory. A systematic measure-
ment campaign of MIMO-OFDM channel involving randomly moving pedestrians
in indoor environments, has never been investigated before. In this thesis, temporal
variations due to randomly placing none to ten human bodies have been taken into
account.
The main focal point of this thesis, is to analyze and characterize the MIMO-
OFDM channel considering pedestrians effects in indoor environments. A custom
build GO ray tracing simulation tool has been prepared for verification of all mea-
surement results. An introductory analysis of the random scenarios, shows a smaller
temporal variation of MIMO-OFDM channel capacity for Fixed SNR than that for
Fixed Tx. Also for both cases, Fixed SNR and Fixed Tx, a rise in MIMO-OFDM
channel capacity has been observed while deploying more antenna elements. In ad-
dition, a general increment in MIMO-OFDM channel capacity dynamic range has
been recorded for both scenarios, with increasing numbers of pedestrians or antenna
Chapter 7 Analysis of Results for Random Scenarios
7.1 Introduction 123
combinations.
During the channel measurements, pedestrians were asked to randomly move in
different directions between the Txs and Rxs. The blue area in Fig. 7.1 represents
the area for random human movement. The Fig. 7.1 also shows the antenna array
locations (Tx and Rx).
Figure 7.1: Area for Random Human Movement in the Measurements Site Room
52C
Further details relating to setup for measurement of random scenarios is showed
in Chapter 4 and simulation details can be acquired from Chapter 5.
Chapter 7 Analysis of Results for Random Scenarios
7.2 MIMO-OFDM Channel Measurements 124
7.2 MIMO-OFDM Channel Measurements
As human bodies can cause significant variations in the MIMO-OFDM channel
capacity and channel capacity dynamic range, there is a need of a comprehensive
characterization of the indoor MIMO-OFDM channel considering real life scenar-
ios. To capture the many different scenarios, a systematic measurement have been
conducted for 0-5, 7 and 10 randomly moving people in a given area between Txs
and Rxs. Results are based on measured MIMO-OFDM channels at 5.24 GHz band
with 40 MHz bandwidth, 114 OFDM sub-carriers and 16 MIMO sub-channels, a
total of approximately 6 million SISO channels are obtained. For each measured
scenario, approximately 200 time samples have been collected at the rate of 2 sam-
ples/sec. For clarity we have presented several plotted graphs with different num-
bers of people in comparison with vacant condition.
7.2.1 Average Channel Capacity
In this section, MIMO-OFDM average channel capacity due to randomly moving
pedestrians in between the direct LOS path of Txs and Rsx are presented, based
on measured and simulated results with different numbers of actual human body.
The number of pedestrians considered in the measurement ranges from one to five,
seven and ten. Millions of MIMO channels have been obtained with 114 OFDM
sub-carriers and 16 MIMO sub-channels. The detailed statistics of collected data
samples can be found in Chapter 1 Section 5.4.3.
Two separate sets of plots (one for Fixed SNR and other for Fixed Tx) are pre-
sented. Fig. 7.2 and Fig. 7.3 show the MIMO-OFDM channel capacity for different
numbers of pedestrians in an indoor environment, considering Fixed SNR and Fixed
Tx scenarios. Three individual graphs in each figure show the MIMO-OFDM ca-
pacity variation for 0, 1, 5 and 10 people. Here, x axis represents the time scale in
seconds and y axis represents the MIMO-OFDM channel capacity in bits/sec/Hz. In
all the scenarios, not very significant variations have been observed for the vacant
Chapter 7 Analysis of Results for Random Scenarios
7.2 MIMO-OFDM Channel Measurements 125
0 20 40 60 80 100 120 14013.5
14
14.5
15
15.5
16
16.5
17
17.5
Time (sec)
MIM
O−
OF
DM
cha
nnel
cap
acity
(bi
ts/s
/Hz)
Fix
ed S
NR
0
1
(a) Channel Capacity for Random FSNR(0 & 1 ppl)
0 20 40 60 80 100 120 14013.5
14
14.5
15
15.5
16
16.5
17
17.5
Time (sec)
MIM
O−
OF
DM
cha
nnel
cap
acity
(bi
ts/s
/Hz)
Fix
ed S
NR
0
5
(b) Channel Capacity for Random FSNR(0 & 5 ppl)
0 20 40 60 80 100 120 14013.5
14
14.5
15
15.5
16
16.5
17
17.5
Time (sec)
MIM
O−
OF
DM
cha
nnel
cap
acity
(bi
ts/s
/Hz)
Fix
ed S
NR
0
10
(c) Channel Capacity for Random FSNR(0 & 10 ppl)
Figure 7.2: Measured MIMO-OFDM Channel Capacity for Random Scenarios Vs
Numbers of People using Fixed SNR criteria
Chapter 7 Analysis of Results for Random Scenarios
7.2 MIMO-OFDM Channel Measurements 126
0 20 40 60 80 100 120 14020
20.5
21
21.5
22
22.5
23
23.5
Time (sec)MIM
O−
OF
DM
cha
nnel
cap
acity
(bi
ts/s
/Hz)
Fix
ed T
x P
ower
0
1
(a) Channel Capacity for Random FTx(0 & 1 ppl)
0 20 40 60 80 100 120 14020
20.5
21
21.5
22
22.5
23
23.5
24
24.5
Time (sec)MIM
O−
OF
DM
cha
nnel
cap
acity
(bi
ts/s
/Hz)
Fix
ed T
x P
ower
0
5
(b) Channel Capacity for Random FTx(0 & 5 ppl)
0 20 40 60 80 100 120 14019
20
21
22
23
24
25
Time (sec)MIM
O−
OF
DM
cha
nnel
cap
acity
(bi
ts/s
/Hz)
Fix
ed T
x P
ower
0
10
(c) Channel Capacity for Random FTx(0 & 10 ppl)
Figure 7.3: Measured MIMO-OFDM Channel Capacity for Random Scenarios Vs
Numbers of People using Fixed Tx criteria
Chapter 7 Analysis of Results for Random Scenarios
7.2 MIMO-OFDM Channel Measurements 127
scenarios, while even introducing a single person created significant variation of up
to 2 bits/sec/Hz in the channel capacity. In Fixed SNR scenarios, MIMO-OFDM
average channel capacity maintain a fairly steady change with increasing numbers
of people, while in Fixed Tx condition a declining value for MIMO-OFDM aver-
age channel capacity has been recorded. Due to the constant random human body
movement, a regular frequency selective fading was observed at all the time, with a
higher variation when more pedestrian were introduced in the indoor environment.
In addition, for Fixed Tx scenarios MIMO-OFDM channel capacity reduces, as re-
ceivable power decreases, due to the constant blocking of LOS path with increasing
number of randomly moving human bodies.
Temporal variations of MIMO-OFDM channel capacity has also been consid-
ered in terms of different numbers of antenna elements. Fig. 7.4 shows the variation
in average channel capacity as a function of the number of people (0-5,7,10), mov-
ing randomly within the room with different antenna array (2× 2, 3× 3 and 4× 4)
combination in random Fixed SNR scenarios. In the graph, constant variation due
to human body movement limits the ability to identify the effects caused by increas-
ing number of people, but it surely captures the trend of increment in capacity with
the number of antenna elements.
Antenna Combination 1ppl 2ppl 3ppl 4ppl 5ppl 7ppl 10ppl
2× 2 (FSNR) 7.37 7.40 7.76 7.58 7.41 7.42 7.31
3× 3 (FSNR) 9.77 9.82 10.55 10.31 9.98 9.98 9.69
4× 4 (FSNR) 12.63 12.67 12.53 12.59 12.59 12.61 12.56
2× 2 (FTX) 6.64 6.40 5.82 6.93 6.52 6.03 5.73
3× 3 (FTX) 9.81 9.45 8.64 10.28 9.65 8.88 8.44
4× 4 (FTX) 13.52 12.33 12.20 13.21 12.58 12.45 11.45
Table 7.1: Average Measured MIMO-OFDM Channel Capacity for Random Sce-
narios in Fixed SNR and Fixed Tx Power
Table 7.1 shows the increasing trend of the average MIMO-OFDM channel ca-
Chapter 7 Analysis of Results for Random Scenarios
7.2 MIMO-OFDM Channel Measurements 128
0 20 40 60 80 100 120 140 160 18011.5
12
12.5
13
13.5
14
Sample index
Ave
rage
cap
acity
(bp
s/H
z)
Random 4x4 Fixed SNR
0 1 2 3 4 5 7 10
(a) 4x4 Fixed SNR varying between 11.5 to 14
0 20 40 60 80 100 120 140 160 1809.5
10
10.5
11
Sample index
Ave
rage
cap
acity
(bp
s/H
z)
Random 3x3 Fixed SNR
0 1 2 3 4 5 7 10
(b) 3x3 Fixed SNR varying between 9.5 to 11
0 20 40 60 80 100 120 140 160 1807
7.1
7.2
7.3
7.4
7.5
7.6
7.7
7.8
7.9
8
Sample index
Ave
rage
cap
acity
(bp
s/H
z)
Random 2x2 Fixed SNR
0 1 2 3 4 5 7 10
(c) 2x2 Fixed SNR varying between 7.2 to 7.8
Figure 7.4: Measured Average Channel Capacity for Random Scenarios
Chapter 7 Analysis of Results for Random Scenarios
7.2 MIMO-OFDM Channel Measurements 129
pacity with the number of antenna combinations. A maximum of approximately
68% increment in the 4 × 4 configuration compared to the 2 × 2 for Fixed SNR
has been observed. While a maximum of 103% increment using 4 × 4 array com-
pared to the 2 × 2 for Fixed Tx in MIMO-OFDM average channel capacity has
been recorded. Additionally, the table shows a steady variation for Fixed SNR and
a decremented trend for Fixed Tx power, when number of pedestrian goes from 1 to
10. For Fixed SNR due to the compensating transmission power feature, the average
MIMO-OFDM channel capacity remains steady at most of the times. On the other
hand, for Fixed Tx scenarios, a decrement in MIMO-OFDM channel capacity has
been observed, as random uncontrolled movement of the human body introduces
the arbitrary blocking of the LOS site path, hence reducing the receivable power in
the signal transmission process.
To further study the relationship between channel capacity and number of pedes-
trian a CDF analysis has been carried out, which is presented in the next section.
7.2.2 Channel Capacity Cumulative Distribution Function
Two different sets of CDFs were presented, one set shows capacity CDFs for 0-3
people and the other one shows capacity CDFs for 0,3,5 and 10 people. Here, x axis
represents the MIMO-OFDM capacity in bits/sec/Hz and y axis represents CDF in
% . Fig. 7.5 shows CDFs of MIMO-OFDM capacity for 0 to 3 randomly moving
people using the Fixed SNR (Fig. 7.5(a)) and Fixed Tx power (Fig. 7.5(b)) crite-
ria. As expected, the variation of the capacity for vacant case is minimal, while the
introduction of even one pedestrian changes the CDF significantly. The effects of
having more pedestrians seem less significant, having similar CDFs for one, two,
or three pedestrians. The presence of the increasing pedestrians tends to maintain
a similar pattern for Fixed SNR as in most cases the plots overlap each other or
follow closely. In addition, the capacity with the pedestrian surpasses that for the
vacant scenario approximately between 70% to 90% CDF for Fixed SNR. Com-
paring the spread of CDFs for vacant case and one, two, or three pedestrian case,
Chapter 7 Analysis of Results for Random Scenarios
7.2 MIMO-OFDM Channel Measurements 130
10 10.5 11 11.5 12 12.5 13 13.5 140
10
20
30
40
50
60
70
80
90
100
Fixed SNR MIMO−OFDM capacity (bits/s/Hz)
CD
F (
%)
0
1
2
3
(a) Measured CDF for Random Scenarios in Fixed SNR (0-3 ppl)
10 10.5 11 11.5 12 12.5 13 13.5 140
10
20
30
40
50
60
70
80
90
100
Fixed Tx MIMO−OFDM capacity (bits/s/Hz)
CD
F (
%)
0
1
2
3
(b) Measured CDF for Random Scenarios in Fixed Tx (0-3 ppl)
Figure 7.5: Measured CDF Analysis for Random Scenarios in Fixed SNR and Fixed
Tx(0-3 ppl)
Chapter 7 Analysis of Results for Random Scenarios
7.2 MIMO-OFDM Channel Measurements 131
10 11 12 13 14 150
10
20
30
40
50
60
70
80
90
100
Fixed SNR MIMO−OFDM capacity (bits/s/Hz)
CD
F (
%)
0
3
5
10
(a) Measured CDF for Random Scenarios in Fixed SNR (0,3,5,10 ppl)
9 10 11 12 13 14 150
10
20
30
40
50
60
70
80
90
100
Fixed Tx MIMO−OFDM capacity (bits/s/Hz)
CD
F (
%)
0
3
5
10
(b) Measured CDF for Random Scenarios in Fixed Tx (0,3,5,10 ppl)
Figure 7.6: Measured CDF Analysis for Random Scenarios in Fixed SNR and Fixed
Tx (0,3,5,10 ppl)
Chapter 7 Analysis of Results for Random Scenarios
7.2 MIMO-OFDM Channel Measurements 132
the pedestrians are found to cause significant variations when they are randomly
moving between Txs and Rxs. In Fixed Tx power scenarios, the capacity appears
to decrease with increasing number of pedestrians. This can be attributed to the
frequent blocking of the direct LOS path, and the consequent absorbtion by the
pedestrians, which causes the reduction of receivable power. It has also been noted
that, for both Fixed SNR and Fixed Tx cases the spread of the CDF increases with
more number of people presents in the indoor environment.
Fig. 7.6 shows CDFs of MIMO-OFDM capacity for the randomly moving peo-
ple using the Fixed SNR (Fig. 7.6(a)) and Fixed Tx power (Fig. 7.6(b)) criteria with
the number of people ranging from 0 to 3,5 and 10. Here, with a larger number of
people, we observe similar results as found in Fig. 7.5. This time the capacity for 3,
5 and 10 pedestrians surpasses that for the vacant scenario between 65% to 75% for
Fixed SNR scenarios, due to the more people blocking the direct LOS path, causing
a reduced received power. Moreover, in Fixed Tx scenarios, a higher reduction in
MIMO-OFDM channel capacity has been observed with increasing number of peo-
ple due to the fixed transmission power and consequent reduction in received power
when pedestrians block the LOS.
7.2.3 Channel Capacity Dynamic Range
Antenna combination 1ppl 2ppl 3ppl 4ppl 5ppl 7ppl 10ppl
2× 2 (FSNR) 0.69 0.92 1.07 1.13 1.28 1.34 1.49
3× 3 (FSNR) 0.86 1.09 1.27 1.29 1.38 1.51 1.64
4× 4 (FSNR) 1.03 1.37 1.54 1.45 1.54 1.84 1.86
2× 2 (FTX) 0.86 1.07 1.29 1.50 1.59 1.71 1.92
3× 3 (FTX) 0.91 1.04 1.36 1.49 1.61 1.77 2.10
4× 4 (FTX) 0.89 0.98 1.51 1.50 1.58 1.69 2.41
Table 7.2: Measurement Average Channel Capacity Dynamic Range for Random
Fixed SNR and Fixed Tx with Different Numbers of People
Chapter 7 Analysis of Results for Random Scenarios
7.3 MIMO-OFDM Channel Simulation 133
Table 7.2 shows the MIMO-OFDM channel capacity dynamic range for mea-
sured Fixed SNR and Fixed Tx Power scenarios. Here, an increment in MIMO-
OFDM capacity dynamic range with increasing number of people and antenna array
combination has been observed for Fixed SNR conditions. For Fixed SNR the ta-
ble shows an maximum increment of approximately 49% while 4 antenna elements
are used, compared with 2 antenna elements. In addition, a maximum increment
of 115% has been recorded when the number of pedestrians increases from 1 to
10 in Fixed SNR criteria. A maximum increment of approximately 25% has been
observed for Fixed Tx while 4 antenna elements are used, compared with 2 antenna
elements. The table also shows a maximum increment of 170% when the number
of pedestrians increases from 1 to 10 in Fixed Tx criteria. For Fixed Tx criteria,
table shows a general incremental trend with few minor fluctuations. That is due to
external noise or interference.
Additionally, there is a maximum increment of 0.8 bits/sec/Hz when the number
of pedestrians increases from 1 to 10 in Fixed SNR, and a maximum increment of
1.52 bits/sec/Hz for Fixed Tx power. The increment in dynamic range is due to the
fact that, the variation in channel capacity increases as more number of pedestrians
crossing direct LOS path between Txs and Rxs.
7.3 MIMO-OFDM Channel Simulation
In this thesis simulations of random scenarios have been utilized to predict and
validate the data gathered through arbitrary human body movement in the real life
indoor environment. MATLAB 1 based simulations have been utilized to conduct
all the simulations, which replicates the measurement scenarios. Using these simu-
lations, we have analyzed the MIMO-OFDM channel capacity and MIMO-OFDM
channel capacity dynamic range for 2× 2, 3× 3, and 4× 4 antenna configurations
1MATLAB version 7.8.0.347(R2009a) 32-bit(win32) developed and distributed by The Math-
Works Inc.
Chapter 7 Analysis of Results for Random Scenarios
7.3 MIMO-OFDM Channel Simulation 134
in LOS environments, using Fixed SNR and Fixed Tx power criteria. Simulations
allow us to validate the measurement results as well as to provide an accurate predic-
tion tool. Real life channel measurements are time and cost consuming process. The
simulation implemented in this thesis can carry out measurement predictions using
a simple desktop PC. Conducted simulation results show similar trend in MIMO-
OFDM channel capacity and channel capacity dynamic range for Fixed SNR and
Fixed Tx power, as the one found in measurement for an indoor environment in the
presence of pedestrians and different antenna combinations.
Time variation characteristics due to pedestrian movement have been presented
through simulations. A similar experimental setup (5.24 GHz band with 40 MHz
bandwidth) has been considered for the simulations. The human model used in the
analysis of the deterministic scenarios, has been randomly placed between Txs and
Rxs for simulation of the MIMO-OFDM channel. Up to 10 pedestrians have been
considered for random simulation with 400 receiver antenna locations, 114 OFDM
sub-carriers, 16 MIMO sub-channels and on average 60 samples. A total of ap-
proximately 438 million SISO channels were obtained by conducting simulation.
Preliminary analysis indicates similarity with conducted measurements. In general,
simulated results show the mean channel capacity for Fixed SNR maintains a mini-
mum variation and Fixed Tx results show a decreasing trend with number of pedes-
trians. In addition, the channel capacity dynamic range increases, with the number
of pedestrians and number of antenna elements within the indoor environment.
7.3.1 Average Channel Capacity
Simulations were performed for different receiver antenna array locations defined
on a grid within an area of two wavelengths times two wavelengths, with 0.1 wave-
length resolution, resulting in 400 locations, in order to observe the variation of the
capacity dynamic range as a function of small scale displacement of the antennas.
While a small variation in the capacity dynamic range was observed, depending on
the exact location of the receiver antenna array, the dynamic range results are aver-
Chapter 7 Analysis of Results for Random Scenarios
7.3 MIMO-OFDM Channel Simulation 135
1ppl 2ppl 3ppl 4ppl 5ppl 7ppl 10ppl4
6
8
10
12
(a)
Simulation Result in Fixed SNR
2x2 3x3 4x4
1ppl 2ppl 3ppl 4ppl 5ppl 7ppl 10ppl4
6
8
10
12
Ave
rage
Cha
nnel
Cap
acity
(bi
t/s/H
z)
(b)
Simulation Result in Fixed Tx
2x2 3x3 4x4
Figure 7.7: Simulated Average Capacity for Random Scenarios with Different
Numbers of Pedestrians and Antenna Combinations
aged over 400 receiver antenna array locations to obtain the trend as a function of
the number of antennas and of pedestrians. In addition, 60 samples ( 60 simulations
results using different pedestrian distribution) have been considered to capture the
variation trend more precisely.
Fig. 7.7 (a) and (b) show the average channel capacity plotted for different num-
ber of randomly moving people in an indoor environment. Here x axis is the num-
ber of people and y axis is average channel capacity in bits/sec/Hz. For both Fixed
SNR and Fixed Tx scenarios, there is a clear increase in MIMO-OFDM average
channel capacity with the number of antenna element arrays. It has been noted
Chapter 7 Analysis of Results for Random Scenarios
7.3 MIMO-OFDM Channel Simulation 136
that, the MIMO-OFDM channel capacity does not vary significantly with the in-
creasing number of people for Fixed SNR. This is due to the transmission power
compensation factor, which triggers extra power from the transmitting antennas to
increase receivable power, as direct LOS gets blocked by the human bodies reduc-
ing the receivable power. This process takes place again and again as human bodies
block the LOS path. Therefore, the calculated average channel capacity shows a
very minimal fluctuation with the increasing number of people. In this analysis, the
MIMO-OFDM average channel capacity has been averaged over frequency, over
different antenna combinations, over time samples and over receiver locations.
In Fixed Tx scenarios, a decrease has been observed with the increasing number
of people. This decrease is due to a reduction in receivable power as more and more
human body block the direct LOS path, hence the observed declining trend in the
MIMO-OFDM channel capacity with increasing number of people.
Antenna combination 1ppl 2ppl 3ppl 4ppl 5ppl 7ppl 10ppl
2× 2 (FSNR) 5.56 5.60 5.57 5.55 5.49 5.34 5.52
3× 3 (FSNR) 7.43 7.49 7.29 7.51 7.53 7.48 7.47
4× 4 (FSNR) 9.26 9.30 9.13 9.20 9.42 9.31 9.43
2× 2 (FTX) 7.00 7.08 6.39 6.32 6.07 5.86 4.88
3× 3 (FTX) 7.38 7.21 6.67 6.59 6.32 6.07 5.15
4× 4 (FTX) 9.74 9.54 9.11 8.81 8.63 7.79 6.88
Table 7.3: Average Simulated Channel Capacity for Random Scenarios with Fixed
SNR and Fixed Tx Power (using middle 90 percent samples)
Table 7.3 shows the Simulated MIMO-OFDM average channel capacity for both
Fixed SNR and Fixed Tx Power criteria, while pedestrians are randomly moving
in the indoor environment. On average, a maximum increment of 80% has been
observed, comparing 4× 4 antenna elements with 2× 2 antenna elements for Fixed
SNR scenarios and a maximum increment of 50% has been recorded for Fixed Tx
conditions.
Chapter 7 Analysis of Results for Random Scenarios
7.3 MIMO-OFDM Channel Simulation 137
7.3.2 Channel Capacity Cumulative Distribution Function
Fig. 7.8 shows CDFs of MIMO-OFDM capacity for the randomly moving pedes-
trians scenario using the Fixed SNR (Fig. 7.8(a)) and Fixed Tx power (Fig. 7.8(b))
criteria. Dramatic changes have been observed even with one moving person, com-
pared with vacant. The presented CDFs for Fixed SNR and Fixed Tx, capture the
trend of the measured findings, by preserving the broader CDF spread for more
number of people present in the indoor environment. The incremental change in
measured MIMO-OFDM average channel capacity for Fixed SNR is well captured
by the simulation results as higher capacity values are found for higher number of
pedestrians. With similar CDFs for different number of people, the effects of having
more than one pedestrian seem less significant. For Fixed SNR, the increasing trend
in the MIMO-OFDM average channel capacity maintains a constant growth from
below the mean up to 100% probability. There is a greater CDF spread for higher
number of pedestrians present in the indoor environment. While in Fixed Tx, the
MIMO-OFDM average channel capacity with the pedestrians always surpasses that
for the vacant condition with 20% to 40% probability. Here, blocking of the direct
LOS path by the pedestrians contributed to the reduction of receivable power. Com-
paring the spread of CDFs for vacant case and one, two, or three pedestrian case,
the pedestrians are found to cause observable variations in channel capacity, in the
order of 0.5 bits/sec/Hz for mean channel capacity values.
Fig. 7.9 shows CDFs of MIMO-OFDM capacity for the randomly moving peo-
ple using the Fixed SNR (Fig. 7.9(a)) and Fixed Tx power (Fig. 7.9(b)) criteria with
the number of people ranging from 0 to 3,5 and 10. Here, with a larger number of
people, we observe similar results as found in Fig. 7.8. This time the capacity for 3,
5 and 10 pedestrians surpasses that for the vacant scenario between 5% to 25% for
Fixed SNR scenarios, due to the more people blocking the direct LOS path, causing
a reduced received power. Moreover, in Fixed Tx scenarios, a higher reduction in
MIMO-OFDM channel capacity has been observed with increasing number of peo-
ple due to the fixed transmission power and consequent reduction in received power
Chapter 7 Analysis of Results for Random Scenarios
7.3 MIMO-OFDM Channel Simulation 138
7 7.5 8 8.5 9 9.5 10 10.5 11 11.5 120
10
20
30
40
50
60
70
80
90
100
Fixed SNR MIMO−OFDM channel capacity (bits/s/Hz)
CD
F (
%)
0 ppl1ppl2ppl3ppl
(a) Simulated CDF for Random Scenarios in Fixed SNR
4 5 6 7 8 9 10 11 12 13 14 150
10
20
30
40
50
60
70
80
90
100
Fixed Tx MIMO−OFDM channel capacity (bits/s/Hz)
CD
F (
%)
0 ppl1ppl2ppl3ppl
(b) Simulated CDF for Random Scenarios in Fixed Tx
Figure 7.8: Simulated CDF for Random Scenarios in Fixed SNR and Fixed Tx with
0 to 3 pedestrians)
Chapter 7 Analysis of Results for Random Scenarios
7.3 MIMO-OFDM Channel Simulation 139
5 6 7 8 9 10 11 12 13 140
10
20
30
40
50
60
70
80
90
100
Random Fixed SNR MIMO−OFDM capacity (bits/s/Hz)
CD
F (
%)
0 ppl3ppl5ppl10ppl
(a) Simulated CDF for Random Scenarios in Fixed SNR
2 4 6 8 10 12 14 16 180
10
20
30
40
50
60
70
80
90
100
Random Fixed Tx MIMO−OFDM capacity (bits/s/Hz)
CD
F (
%)
0 ppl3ppl5ppl10ppl
(b) Simulated CDF for Random Scenarios in Fixed Tx
Figure 7.9: Simulated CDF for Random Scenarios in Fixed SNR and Fixed Tx with
0, 3, 5 and 10 pedestrians)
Chapter 7 Analysis of Results for Random Scenarios
7.3 MIMO-OFDM Channel Simulation 140
when pedestrians block the LOS.
7.3.3 Channel Capacity Dynamic Range Analysis
Antenna combination 1ppl 2ppl 3ppl 4ppl 5ppl 7ppl 10ppl
2× 2 (FSNR) 1.14 1.32 1.48 1.54 1.54 1.57 1.56
3× 3 (FSNR) 1.28 1.52 1.84 2.18 1.97 2.02 2.26
4× 4 (FSNR) 1.49 1.71 2.18 2.89 2.53 2.57 2.96
2× 2 (FTX) 2.95 3.68 4.04 4.52 5.54 5.55 5.99
3× 3 (FTX) 3.67 4.52 5.01 5.89 7.44 7.28 7.99
4× 4 (FTX) 4.40 5.42 6.09 7.40 9.10 8.86 9.94
Table 7.4: Simulated Average Channel Capacity Dynamic Range for Random Sce-
narios with Fixed SNR and Fixed Tx
Analysis of MIMO-OFDM channel capacity dynamic range shows an increas-
ing trend with higher number of people as well as with increasing number of antenna
elements. The projected results also capture the trend found in measured scenarios
presented earlier in this chapter. Table 7.4 shows the average MIMO-OFDM chan-
nel capacity dynamic range for measured Fixed SNR and Fixed Tx Power criteria.
Here it has been observed that, MIMO-OFDM channel capacity dynamic range
increases with the number of pedestrians as well as with the number of antenna ele-
ments in an indoor environment. For both Fixed SNR and Fixed Tx, the table shows
an increment of approximately 100% for Fixed SNR and 125% for Fixed Tx, while
4 antenna elements are used compared with 2 antenna elements. Besides, there is a
maximum increment of 1.5 bits/sec/Hz when number of pedestrians increases from
1 to 10 in Fixed SNR and a maximum increment of 5.5 bits/sec/Hz for Fixed Tx
power.
Chapter 7 Analysis of Results for Random Scenarios
7.4 Measurement Vs. Simulation 141
7.4 Measurement Vs. Simulation
Simulations show a similar trend of increasing MIMO-OFDM average channel ca-
pacity for Fixed SNR, and decreasing MIMO-OFDM channel capacity for Fixed
Tx power as found in experiments for indoor environment with increasing num-
ber of pedestrians. The conducted simulations also capture the measured change in
MIMO-OFDM channel capacity dynamic range observed in the real life scenarios.
The simulations replicated the time variation characteristics due to pedestrian
movement. A simulated pedestrian model, placed between Txs and Rxs, has been
used for the analysis of 0-5, 7 and 10 pedestrians moving randomly. A total of 400
receiver antenna locations, 114 OFDM sub-carriers and 16 MIMO sub-channels
have been simulated for each scenarios. Preliminary analysis for Fixed SNR cri-
teria, indicates a steady mean MIMO-OFDM channel capacity and an incremental
trend MIMO-OFDM dynamic range with the number of pedestrians. On the other
hand, analysis for Fixed Tx criteria shows a decreasing trend in MIMO-OFDM
channel capacity and an incremental trend in MIMO-OFDM channel capacity dy-
namic range with the number of pedestrians.
7.4.1 MIMO-OFDM Channel Capacity
In this section, both measured and simulated MIMO-OFDM average channel ca-
pacity for Fixed SNR and Fixed Tx criterion have been plotted against the numbers
of people and the number of antenna elements. Fig. 7.10 and Fig. 7.11 show the
MIMO-OFDM average channel capacity, assuming a Fixed SNR of 15dB for Fixed
SNR criteria. Here, x axis depicts the number of people and y axis depicts the
average channel capacity in bits/sec/Hz. This has been confirmed by both mea-
surements and simulations ranging from 1 to 5, 7 and 10 pedestrians. The rise in
multipath conditions caused by a higher number of pedestrians originated a general
increase in average channel capacity for all the array sizes, 2× 2, 3× 3 and 4× 4.
The highest measured average capacity, 13.52 bits/sec/Hz, corresponds to the 4× 4
Chapter 7 Analysis of Results for Random Scenarios
7.4 Measurement Vs. Simulation 142
1ppl 2ppl 3ppl 4ppl 5ppl 7ppl 10ppl
8
10
12
14
(a)
Measured Result in Fixed SNR
2x2 3x3 4x4
1ppl 2ppl 3ppl 4ppl 5ppl 7ppl 10ppl5
10
15
Ave
rage
Cha
nnel
Cap
acity
(bi
t/s/H
z)
(b)
Simulated Result in Fixed SNR
2x2 3x3 4x4
Figure 7.10: Measured and Simulated Average Channel Capacity for Random Sce-
narios in Fixed SNR
Chapter 7 Analysis of Results for Random Scenarios
7.4 Measurement Vs. Simulation 143
1ppl 2ppl 3ppl 4ppl 5ppl 7ppl 10ppl5
10
15
(a)
Measured Result in Fixed Tx
2x2 3x3 4x4
1ppl 2ppl 3ppl 4ppl 5ppl 7ppl 10ppl4
6
8
10
12
Ave
rage
Cha
nnel
Cap
acity
(bi
t/s/H
z)
(b)
Simulation Result in Fixed Tx
2x2 3x3 4x4
Figure 7.11: Measured and Simulated Average Channel Capacity for Random Sce-
narios in Fixed Tx
Chapter 7 Analysis of Results for Random Scenarios
7.4 Measurement Vs. Simulation 144
array with 1 pedestrians moving randomly. The 4× 4 array shows the highest aver-
age measured channel capacity, compared to the 2× 2 and 3× 3 arrays as a higher
number of parallel channels are created in a MIMO system with larger arrays. In
Fixed SNR scenarios (Fig. 7.10), MIMO-OFDM average channel capacity shows
very little variation, while still preserving higher capacity values for higher number
of antenna elements. This is due to the increase in transmission power to keep the
SNR fixed. On the other hand, as expected and in accordance with measurements
for Fixed Tx scenarios, Fig. 7.11 shows a decreasing MIMO-OFDM channel ca-
pacity with an increasing number of people, while also preserving higher capacity
value for higher number of antenna elements. The reduction in channel capacity
with increasing number of people is due to the Fixed Tx power constraint. In addi-
tion, the constant random movement of the pedestrians prevent dominant LOS path
in most cases; hence the observed reduction in receivable power with increasing
number of pedestrians.
Due to the noise free environment and other external factors, simulation results
tend to slightly under estimated channel capacity when compared to experimental
results. In general, simulation results closely match the trend of the measurement
results. Similar to the channel capacity dynamic range, the Fixed Tx power criterion
is larger than the Fixed SNR. In addition and in accordance with measured results,
the Fixed Tx MIMO-OFDM average channel capacity was found to be larger than
the Fixed SNR. Tables (Table 7.1 and Table 7.3 for simulated and measured MIMO-
OFDM channel capacity show very minor fluctuations for increasing pedestrians as
well as for more antenna elements in the indoor environment. Interestingly, in both
simulated and measured MIMO-OFDM average channel capacity, minor variations
in the order of one decimal places have been observed, with additional pedestrian
introduced in the indoor environment.
Chapter 7 Analysis of Results for Random Scenarios
7.4 Measurement Vs. Simulation 145
1ppl 2ppl 3ppl 4ppl 5ppl 7ppl 10ppl1
1.5
2
2.5
3
3.5
Dyn
amic
ran
ge (
bits
/s/H
z)
(a)
Simulation Result in Fixed SNR
2x2 3x3 4x4
1ppl 2ppl 3ppl 4ppl 5ppl 7ppl 10ppl0.5
1
1.5
2
2.5
Dyn
amic
ran
ge (
bits
/s/H
z)
(b)
Measurement Result in Fixed SNR
2x2 3x3 4x4
Figure 7.12: MIMO-OFDM Channel Capacity Dynamic Range Variation with Dif-
ferent Number of Pedestrians and Antennas for Random Scenarios in Fixed SNR.
(a)Simulation (b)Measurement
Chapter 7 Analysis of Results for Random Scenarios
7.4 Measurement Vs. Simulation 146
1ppl 2ppl 3ppl 4ppl 5ppl 7ppl 10ppl2
4
6
8
10
12
Dyn
amic
Ran
ge (
bits
/s/H
z)
(a)
Simulation Result in Fixed Tx
2x2 3x3 4x4
1ppl 2ppl 3ppl 4ppl 5ppl 7ppl 10ppl
1
1.5
2
2.5
Dyn
amic
Ran
ge (
bits
/s/H
z)
(b)
Measurement Result in Fixed Tx
2x2 3x3 4x4
Figure 7.13: MIMO-OFDM Channel Capacity Dynamic Range Variation with Dif-
ferent Number of Pedestrians and Antennas for Random Scenarios in Fixed Tx.
(a)Simulation (b)Measurement
Chapter 7 Analysis of Results for Random Scenarios
7.4 Measurement Vs. Simulation 147
7.4.2 MIMO-OFDM Channel Capacity Dynamic Range
To provide a clear comparison between the analyzed measurement and simulated
MIMO-OFDM channel capacity dynamic range, plots detailing Fixed SNR and
Fixed TX scenarios with increasing number of pedestrian and antenna elements
have been generated. Fig. 7.12 and 7.13 show the variation in measured and sim-
ulated dynamic range as a function of increasing numbers of people (1-5,7 and 10)
for Fixed SNR an Fixed Tx criteria. In general, we observe that for both simula-
tions and measurements the capacity dynamic range for the Fixed Tx scenarios is
larger than the Fixed SNR scenarios. With a growing number of randomly moving
pedestrians, a larger reduction of the LOS power is introduced, and hence a larger
dynamic range results for the Fixed Tx power capacity. For the Fixed SNR capacity,
the blocking of the LOS path by a larger number of pedestrians introduces further
decorrelation of the channel, and hence the capacity dynamic range for Fixed SNR
increases with the number of pedestrians. Also, the reduction in channel capacity
due to human body shadowing effects is much more noticeable than the expected in-
crease in capacity due to the decorrelation of the channel caused by the obstruction
of the direct LOS path for the Fixed Tx power criteria. Simulated results capture
the increasing trend in measured MIMO-OFDM capacity dynamic range with the
number of pedestrians. For the Fixed Tx criteria (Fig. 7.13), the differential incre-
ment in capacity dynamic range is approximately linear with the increasing number
of people within the indoor environment. A larger reduction of LOS power due to
the increased number of pedestrians moving randomly in the indoor environment,
causes larger dynamic range results for the Fixed Tx power capacity. For the Fixed
SNR criteria (Fig. 7.12), the differential increment of capacity dynamic range with
the increasing number of people in the room is less evident than in the Fixed Tx
criteria. For the Fixed SNR case, the blocking of the LOS path by a larger number
of pedestrians introduces an increasing de-correlation of the channel. However, the
Fixed SNR capacity dynamic range increases with the number of pedestrians at a
slower rate than the one exhibited by the Fixed Tx case. The system designer needs
Chapter 7 Analysis of Results for Random Scenarios
7.4 Measurement Vs. Simulation 148
to consider how one might mitigate the absolute dynamic range to provide stable
performance in the presence of moving objects with increasing number of anten-
nas. The spread between the highest and lowest value of the the dynamic range
is larger for Fixed Tx, predicted 5.5 bits/sec/Hz and measured 1.5 bits/sec/Hz, in
comparison with Fixed SNR criteria, predicted 1.5 bits/sec/Hz and measured 0.7
bits/sec/Hz.
MIMO-OFDM channels corresponding to 2×2 and 3×3 are extracted from 4×4
results for both measurement and simulation results, using all possible antenna com-
binations. Fig. 7.14 shows the MIMO-OFDM channel capacity dynamic range with
different number of antenna combinations. Note that the results include different
antenna spacing by using adjacent or diagonal antenna elements. The large antenna
spacing, set to 3 wavelengths at Tx and 2 wavelengths at Rx, is considered to have
small effects on the results, as found in [15]. The dynamic range results for 2×2 and
3×3 are also averaged over different combinations to provide representative values.
Both measurements and simulations results show that the MIMO-OFDM chan-
nel capacity dynamic range slightly increase with the number of antennas used. The
increase on MIMO-OFDM capacity dynamic range as a function of the number of
antennas is considered to be due to the increase in the MIMO-OFDM channel ca-
pacity. To verify this point, the normalized dynamic range, which is the ratio of the
dynamic range value with respect to the median capacity, is shown in Fig. 7.15.
While the trend of increasing dynamic range with the number of pedestrians is
maintained, the relationship between the normalized dynamic range and the num-
ber of antennas is reversed. Here it has been observed that, when the dynamic range
is scaled by the median capacity, a larger number of antennas tend to provide a
smaller variation in the normalized dynamic range. This is considered to be due
to increased path diversity by a larger number of MIMO channels with the larger
number of antennas. While it may be desired to obtain more stable (less absolute
dynamic range) MIMO-OFDM channel performance with increase in the number
of antennas, both measurements and simulations show that the absolute capacity
Chapter 7 Analysis of Results for Random Scenarios
7.4 Measurement Vs. Simulation 149
SM SS TM TS0
1
2
3
4
5
6
7
8
9
10
Dyn
amic
ran
ge (
bits
/s/H
z)
(a)
2 × 2
10 p7 p5 p4 p3 p2 p1 p
SM SS TM TS0
1
2
3
4
5
6
7
8
9
103 × 3
(b)SM SS TM TS0
1
2
3
4
5
6
7
8
9
104 × 4
(c)
Figure 7.14: Dynamic Range Variation with Different Number of Pedestrians and
Antennas for Random Scenarios using (a) 2 × 2 (b) 3 × 3 (c) 4 × 4 arrays. SM:
Fixed SNR, measurement. SS: Fixed SNR, simulation. TM: Fixed Tx power, mea-
surement. TS: Fixed Tx power, simulation.
Chapter 7 Analysis of Results for Random Scenarios
7.4 Measurement Vs. Simulation 150
dynamic range slightly increases with the number of antenna used.
It has also been noted that, a large deviation of the simulation results from the
measured results is observed for the Fixed Tx power criterion. This is considered
to be due to the simplicity of the models of randomly moving human bodies and
of the environment employed in the simulations. In addition, the increasing trend
of MIMO-OFDM channel capacity dynamic range with the number of people is
clearly visible in the graphs.
7.4.3 Capacity Dynamic Range vs Number of Pedestrians
A similar empirical analysis, as the one conducted for the deterministic measure-
ments in Section 6.4.3 has been performed for the random measurement scenarios.
Linear and quadratic regression analysis were conducted for all random scenarios.
The first order derivative of the linear and quadratic regression equations, give an
indication of the greatest rate of capacity dynamic range variation when the number
of pedestrians increases from one to ten. Table 7.5 shows the best-fit linear and
quadratic regression equations for different antenna combinations for all measure-
ment and simulation scenarios, where the independent variable ’x’ is the number
of pedestrians and ’y’ is the channel capacity dynamic range in bits/sec/Hz. Ta-
ble 7.6 shows the average linear and quadratic equations over all possible antenna
combinations.
Fig. 7.16 shows the linear regression plot for random Fixed SNR scenarios. Here
x axis is the number of people (up to ten people have been considered for random
scenarios), while y axis is the MIMO-OFDM channel capacity dynamic range in
bit/sec/Hz. All possible antenna combinations have been included in the graph. As
expected, the linear regression shows a increasing trend with the number of people.
Here, the mean first order linear coefficient is positive 0.134. The linear Regression
for random Fixed SNR equation is y = 0.134×x+0.95. From the results, an average
increase of 0.134 bit/sec/Hz per each additional pedestrian in the environment could
be expected. The positive gradient captures previously discussed measured and
Chapter 7 Analysis of Results for Random Scenarios
7.4 Measurement Vs. Simulation 151
SM SS TM TS0
20
40
60
80
100
120
140
160
Nor
mal
ized
dyn
amic
ran
ge (
%)
2 x 2
(a)
10 p7 p5 p4 p3 p2 p1 p
SM SS TM TS0
20
40
60
80
100
120
140
1603 x 3
(b)SM SS TM TS0
20
40
60
80
100
120
140
1604 x 4
(c)
Figure 7.15: Normalized Dynamic Range Variation with Different Number of
Pedestrians and Antennas for (a) 2 × 2 (b) 3 × 3 (c) 4 × 4 antenna combinations.
SM: Fixed SNR, measurement. SS: Fixed SNR, simulation. TM: Fixed Tx power,
measurement. TS: Fixed Tx power, simulation.
Chapter 7 Analysis of Results for Random Scenarios
7.4 Measurement Vs. Simulation 152
FSNR FTX
2× 2 (SimLin) 0.06× x + 1.19 −0.32× x + 7.53
2× 2 (SimQua) −0.02× x2 + 0.23× x + 0.95 −0.038× x2 − 0.015× x + 7.07
2× 2 (MesLin) 0.12× x + 0.64 = 0.098× x + 6.69
2× 2 (MesQua) −0.008× x2 + 0.19× x + 0.54 −0.03× x2 + 0.18× x + 6.27
3× 3 (SimLin) 0.15× x + 1.29 −0.33× x + 7.82
3× 3 (SimQua) −0.03× x2 + 0.38× x + 0.94 −0.03× x2 − 0.07× x + 7.43
3× 3 (MesLin) 0.12× x + 0.83 −0.15× x + 9.91
3× 3 (MesQua) −0.007× x2 − 0.17× x + 0.74 −0.056× x20.29× x + 9.24
4× 4 (SimLin) 0.23× x + 1.41 −0.45× x + 10.47
4× 4 (SimQua) −0.04× x2 + .6× x + 0.93 −0.063× x2 + 0.06× x + 9.68
4× 4 (MesLin) 0.12× x + 1.03 −0.20× x + 13.34
4× 4 (MesQua) −0.006× x2 + .18× x + 0.95 −0.03× x2 + 0.02× x + 13.01
Table 7.5: Linear and Quadratic Regression for Different Random Measured and
Simulated Scenarios (Sim: Simulation, Mes: Measurement, Lin: Linear Regres-
sion, Qua: Quadratic Regression, FSNR: Fixed SNR, FTX: Fixed Tx)
Linear Regression (FSNR) 0.134× x + 0.95
Quadratic Regression (FSNR) −0.02× x2 + 0.28× x + 0.84
Linear Regression (FTX) −0.26× x + 7.87
Quadratic Regression (FTX) −0.042× x2 + 0.077× x + 8.79
Table 7.6: Average Linear and Quadratic Regression for Different Random Mea-
sured and Simulated Scenarios (FSNR: Fixed SNR, FTX: Fixed Tx)
Chapter 7 Analysis of Results for Random Scenarios
7.4 Measurement Vs. Simulation 153
1ppl 2ppl 3ppl 4ppl 5ppl 7ppl 10ppl0.5
1
1.5
2
2.5
3
Dyn
amic
ran
ge (
bits
/s/H
z)
(a) Linear Regression (FSNR)
Simulation / Measurement Result for Random Fixed SNR
2x2sim2x2mes linear3x3sim3x3mes4x4sim4x4mes
Figure 7.16: Linear Regression for Random Scenarios in Fixed SNR
simulated increasing trend for MIMO-OFDM channel capacity dynamic range with
the number of pedestrians.
The quadratic regression equation shown in Fig. 7.17. Quadratic Regression for
random Fixed SNR equation is y = −0.019 × x2 + 0.285 × x + 0.84. A negative
gradient has been acquired through first derivative calculation (2× (−0.019)× x +
.284)), showing a negative increment of capacity dynamic range when the number of
pedestrians exceeds 9 using the Fixed Tx criteria. This is due to the sudden increase
on MIMO-OFDM channel capacity dynamic range between 4 and 6 pedestrian.
The increase in dynamic range is considered to be due to a particular geometrical
distribution of multipath. As discussed in Chapter 5, the simulations considered 4
multipath reflections. However, trials were conducted with 4 and 5 reflections. The
sudden increase in dynamic range at 4 pedestrians was less abrupt when 5 reflections
were considered. The difference is attributed to particular distributions of multipath
geometry in the simulations.
Chapter 7 Analysis of Results for Random Scenarios
7.4 Measurement Vs. Simulation 154
1ppl 2ppl 3ppl 4ppl 5ppl 7ppl 10ppl0.5
1
1.5
2
2.5
3
Dyn
amic
ran
ge (
bits
/s/H
z)
(b) Quadratic Regression (FSNR)
Simulation / Measurement Result for Random Fixed SNR
2x2sim quadratic2x2mes3x3sim3x3mes4x4sim4x4mes
Figure 7.17: Quadratic Regression for Random Scenarios in Fixed SNR
Fig. 7.18 shows linear regression and Fig. 7.19 shows quadratic regression for
the MIMO-OFDM channel capacity dynamic range with the number of pedestri-
ans. Due to the fixed transmission power, both measurements and simulations show
a decreasing trend in channel capacity dynamic range with growing number of peo-
ple.
The linear regression equation for dynamic range in random Fixed Tx scenar-
ios is y = −0.26 × x + 7.8. When up to ten pedestrians are considered, a linear
regression shows a negative gradient, of 0.26 bits/sec/Hz per pedestrian, showing a
decreasing MIMO-OFDM capacity dynamic range with the number of pedestrians
in the environment for all Fixed Tx scenarios. An average decrease of 0.26 bit/s/Hz
per each additional pedestrian in the environment can be expected when using a
Fixed Tx criteria. The quadratic regression equation for random Fixed SNR scenar-
ios is y = −0.042 × x2 + 0.78 × x + 8.78. The first derivative produces a linear
gradient of approximately 2× (−0.042)× x + 0.78, showing a negative increment
Chapter 7 Analysis of Results for Random Scenarios
7.5 Conclusions 155
1ppl 2ppl 3ppl 4ppl 5ppl 7ppl 10ppl4
6
8
10
12
14
Dyn
amic
ran
ge (
bits
/s/H
z)
(a) Linear Regression (FTx)
Simulation / Measurement Result for Random Fixed TX
2x2sim2x2mes3x3sim linear3x3mes4x4sim4x4mes
Figure 7.18: Linear Regression for Random Scenarios in Fixed Tx
of capacity dynamic range when the number of pedestrians exceeds 9.
7.5 Conclusions
Measurement and simulation results have been presented for the random scenarios
in conjunction with the analysis of average MIMO-OFDM channel capacity, dy-
namic range. Both measurements and simulations results show a similar incremen-
tal trend in MIMO-OFDM average channel capacity and MIMO-OFDM channel
capacity dynamic range with the number of antenna element combinations in the
indoor environment.
A positive gradient of 0.134 in the average dynamic range against the number
of pedestrians has been observed for Fixed SNR criteria. In addition, a linear re-
gression curve for the average dynamic range against the number of pedestrians
shows a negative gradient of -0.26 in dynamic range with the number of people
for Fixed Tx. With increasing number of pedestrian the average channel capacity
dynamic range can be expected to increase by 0.134 bits/sec/Hz per pedestrian in
Fixed SNR, while the average channel capacity dynamic range can be expected to
Chapter 7 Analysis of Results for Random Scenarios
7.5 Conclusions 156
1ppl 2ppl 3ppl 4ppl 5ppl 7ppl 10ppl4
6
8
10
12
14
Dyn
amic
ran
ge (
bits
/s/H
z)
(b) Quadratic Regression (FTx)
Simulation / Measurement Result for Random Fixed TX
2x2sim quadratic2x2mes3x3sim3x3mes4x4sim4x4mes
Figure 7.19: Quadratic Regression for Random Scenarios in Fixed SNR
decrease by 0.26 bits/sec/Hz per pedestrian for Fixed Tx. Wireless system designers
can improve the quality of service and system efficiency by dynamically managing
the selection of criteria (such as Fixed SNR or Fixed TX). However, using Fixed
SNR though the capacity can be controlled in a steady manner, the increment in
the average channel capacity dynamic range makes the MIMO-OFDM channels
more unpredictable. On the other hand, Fixed Tx criteria provides less variation in
average channel capacity dynamic range, hence more predictability, but there is a
significant reduction in average channel capacity with higher number of pedestrian.
Selection of either Fixed SNR or Fixed Tx depends upon the site survey considering
the amount of pedestrian traffic expected, for example in a location with expected
pedestrian traffic of less than 5 people a Fixed Tx criteria could provide an optimum
solution.
Chapter 8 will present the concluding remarks and suggestions for future work
directions.
Chapter 7 Analysis of Results for Random Scenarios
157
Chapter 8
Conclusions and Future Work
This thesis has focused on the deterministic modeling of indoor MIMO-OFDM
channel with particular emphasis on its application to the design of current or near
future indoor wireless communication services. The work in this thesis is motivated
essentially by a single fundamental question, How the human body is affecting the
MIMO-OFDM channel characteristics in an indoor environment? It is to design and
improve indoor radio channel modeling in the presence of pedestrians.
To validate the deterministic model, access to a systematic measurement cam-
paign is critical. This thesis has presented empirical results from an indoor MIMO-
OFDM channel characterization from both deterministic and random measurement
scenarios in Chapter 6 and 7. These results have been used for validation of the
proposed deterministic model. A FRTT algorithm has been deployed for the imple-
mentation of the deterministic model. The FRTT technique is quite different from
the conventional ray tracing methods. FRTT utilizes a fast line clipping algorithm,
rather than a time-consuming ray intersection test algorithm. The line clipping al-
gorithm has been used in the area of computer graphics to draw many polygons in
real time, such as in the case of a flight simulation programme. With the FRTT im-
plemented on a commercially available PC, it becomes feasible to obtain a thorough
GO solution for the channel characterization with many receiver locations. In a gen-
uine requirement of channel measurement validation, the innovative deterministic
Chapter 8 Conclusions and Future Work
158
model developed in this thesis, has been able to replicate the measurement scenarios
considering the effect of surrounding walls and simple human body models in the
indoor environment. To the best of the authors knowledge, such detailed analysis
considering pedestrian effect on MIMO-OFDM system is the first to be reported in
the literature.
The slight difference found between the measured and simulated results is at-
tributed to the fact that the simulation did not take into account details, such as
lighting fixtures, which may introduce scattered signals. The proposed determin-
istic model provides the designers of indoor wireless communication services, in-
cluding wireless PBX and wireless LAN with an inexpensive means of simulating
and characterizing the 5.24 GHz MIMO-OFDM indoor channel at various sites. In
order to support these findings exhaustive channel measurements have been con-
ducted on the floor of the CSIRO ICT Centre to validate the proposed deterministic
model.
The results were analyzed in terms of the MIMO-OFDM channel capacity, and
MIMO-OFDM channel capacity dynamic range. The system designer needs to con-
sider how one might mitigate the absolute dynamic range to provide stable perfor-
mance in the presence of moving objects with increasing number of antennas. With
financial freedom and bandwidth availability system designers can easily adopt the
Fixed SNR approach for indoor populated environment to achieve a certain steady
MIMO-OFDM average channel capacity.
The next section presents the research contribution from this thesis and the fol-
lowing section presents a list of the research outcomes obtained from this project.
This is followed by suggestions for further work that have arised from the analysis
conducted in this thesis.
Chapter 8 Conclusions and Future Work
8.1 Concluding Remarks 159
8.1 Concluding Remarks
As answer to the demand for ever increasing data rates and augmented mobility,
MIMO-OFDM provides an attractive and practical solution for future high-speed
indoor wireless data communication networks. For the purpose of discussing con-
clusions, this section presents the proposed solutions to the research questions stated
stated in Chapter 1
• How human body is affecting the MIMO-OFDM channel characteristics in
an indoor environment?
• How the average MIMO-OFDM channel capacity is behaving when more
people are introduced in an indoor environment?
• How the average MIMO-OFDM channel capacity is behaving when more
numbers of antenna elements are deployed in an indoor environment?
• How the average MIMO-OFDM channel capacity dynamic range changes
with the number of pedestrian in an indoor environment?
• How the average MIMO-OFDM channel capacity dynamic range changes
with the number of antenna elements in an indoor environment?
• Increase robustness of the simulation by incorporating realistic populated in-
door environment?
Propagation effects cause by body shadowing greatly affect the received signal
strength and can significantly vary transmission quality. Moving pedestrians can in-
tersect the direct path of the wave between the transmitting and receiving antenna,
potentially blocking the LOS path. Additionally, pedestrians moving within the
environment also act as scatterers, contributing to multipath fading through a com-
bination of absorbtion, reflection and diffraction mechanisms, often allowing the
communication link to be maintained by the contribution of reflected waves as the
Chapter 8 Conclusions and Future Work
8.1 Concluding Remarks 160
propagation conditions become NLOS. Despite different efforts to characterize in-
door propagation in populated environments, there have been no systematic studies
that comprehensively characterize the effect of human body effects in indoor propa-
gation. Therefore, in this thesis, measurements and statistical analysis of the indoor
MIMO-OFDM propagation channel at 5.24 GHz in populated environments were
considered. The results presented suggest that human body shadowing effects have
an important influence on the characteristics of the indoor MIMO-OFDM channel
at 5.24 GHz. A novel analysis of MIMO-OFDM channel capacity, channel capac-
ity dynamic range, CDF, linear regression and quadratic regression for the indoor
MIMO-OFDM channel from measurements and simulations have been illustrated
at 5.24 GHz. In order to validate the results, a customized GO based FRTT simula-
tion has been developed with exact replication of the measurement scenarios. The
measurements were performed during working hours to keep the realistic surround-
ings active and incorporate them in the measured result. It has also been noted that,
a large deviation of the simulation results from the measured results is observed for
the Fixed Tx power criterion. This is considered to be due to the simplicity of the
models of randomly moving human bodies and of the environment employed in the
simulations.
A systematic analysis of the effect of pedestrian movement on channel capacity
for a line-of-sight MIMO-OFDM system, of 4 × 4 elements in an indoor envi-
ronment was studied through a novel channel model. It was observed that temporal
variations due to the presence of pedestrians significantly affect the theoretical max-
imum channel capacity of indoor MIMO systems at 5.24 GHz, for indoor environ-
ments with different number of pedestrians. The mean channel capacity increased
linearly with the numbers of pedestrians present within the environment for deter-
ministic Fixed SNR, and decreased with the number of pedestrians in random Fixed
Tx condition. A strong similarity between the measurement and simulation results
was observed in most of the cases. The results obtained show that, in both Fixed
SNR and Fixed Tx for deterministic condition, the channel capacity dynamic range
Chapter 8 Conclusions and Future Work
8.1 Concluding Remarks 161
rose with the number of pedestrians as well as with the number of antenna combina-
tions. In random scenarios with 10 pedestrians, an increment in channel capacity of
up to 0.89 bits/sec/Hz in Fixed SNR and up to 1.52 bits/sec/Hz in Fixed Tx has been
recorded compared to the one pedestrian scenario. The proposed modeling tech-
nique offered a reliable solution to the performance evaluation of MIMO-OFDM
wireless systems for indoor radio communications by taking into account specific
location and pedestrian traffic parameters. The described technique could be di-
rectly applied to indoor MIMO-OFDM systems, where the analysis should focus
on the combination of multipath fading caused by the array moving in the environ-
ment and the effect of pedestrians. The need for adaptive coding schemes to make
best use of the dynamic fluctuations in available channel capacity in populated in-
door environments was also emphasized.
The increase in spectral efficiency offered by MIMO-OFDM systems is based
on the utilization of multiple antennas at both the transmitter and the receiver end.
Using multiple antenna a linear increase in spectral efficiency can be achieved. The
high spectral efficiencies attained by a MIMO-OFDM system are enabled by the
fact that in a rich scattering environment, the signals from each individual trans-
mitter appear highly uncorrelated at each of the receive antennas. In this thesis,
the analysis has been conducted using up to 4 antenna elements. Results obtained
clearly shows the increasing trend in average MIMO-OFDM channel capacity with
higher number of antennas for both measurements and simulations. The replication
of measurement scenarios using the GO based ray tracing simulation allowed the
generation of temporal profiles for the complex transfer function of each antenna
combination in the MIMO-OFDM system in the presence of specified pedestrian
movement in the indoor environment. From the results a maximum increase in av-
erage channel capacity of 49% has been measured while 4 antenna elements are
used, compared with 2 antenna elements. The highest measured average capacity,
11.75 bits/sec/Hz, corresponds to the 4x4 array with 10 pedestrians moving ran-
domly.
Chapter 8 Conclusions and Future Work
8.1 Concluding Remarks 162
Significant variation in channel capacity dynamic range as a function of increas-
ing numbers of people (1-5,7 and 10) has been measured. In general, we observe
that for both simulations and measurements the capacity dynamic range for the
Fixed Tx scenarios is larger than the Fixed SNR scenarios. This is due to the re-
duction in channel capacity values due to human body shadowing effects being
much more noticeable than the expected increase in capacity due to the decorrela-
tion of the channel caused by the obstruction of the direct LOS path for the Fixed
Tx power criteria. Additionally, the spread between the highest and lowest value of
the dynamic range is larger for Fixed Tx, predicted 5.5 bits/sec/Hz and measured
1.5 bits/sec/Hz, in comparison with Fixed SNR criteria, predicted 1.5 bits/sec/Hz
and measured 0.7 bits/sec/Hz. This has been confirmed by both measurements and
simulations ranging from 1 to 5, 7 and 10 pedestrians.
It has been observed in both measurements and simulations, that the MIMO-
OFDM capacity dynamic range increases with the number of antennas used.The
increase of the dynamic range as a function of the number of antennas is considered
to be due to the increase in the MIMO-OFDM capacity with the number of anten-
nas. To verify this point, the normalized dynamic range, which is the ratio of the
dynamic range value with respect to the median capacity, has been analyzed. While
the trend of increasing dynamic range with the number of pedestrians is maintained,
the relationship between the normalized dynamic range and the number of antennas
is reversed. Here it has been observed that, when the dynamic range is scaled by
the median capacity, a larger number of antennas tend to provide a smaller varia-
tion in the normalized dynamic range. This is considered to be due to increased
path diversity by a larger number of MIMO channels with the larger number of
antennas. While it may be desired to obtain more stable (less absolute dynamic
range) MIMO-OFDM channel performance with increase in the number of anten-
nas, both measurements and simulations show that the absolute capacity dynamic
range slightly increases with the number of antenna used. The system designer
needs to consider how one might mitigate the absolute dynamic range to provide
Chapter 8 Conclusions and Future Work
8.2 Research Outcomes 163
stable performance in the presence of moving objects with increasing number of
antennas.
In the customized section of the software, we have implemented several modules
to replicate the measurement scenarios which were dynamically simulated consider-
ing permeability and conductivity of materials in the environments. This innovative
section of the software provides the extra feature of a simplified human body, which
can be located at different positions in either a deterministic or a random fashion.
A very simple human body block was employed with a dimension of 0.62 m depth,
0.31 m width and 1.70 m height with the permittivity and conductivity character-
istics of a real human body. One simulation was performed for different receiver
antenna array locations defined on a grid within an area of two wavelengths times
two wavelengths with 0.1 wavelength resolution resulting in 400 locations, in order
to observe the variation of the capacity dynamic range as a function of small scale
displacement of the antennas. While small variations in the capacity dynamic range
were observed, depending on the exact location of the receiver antenna array, the
dynamic range results are averaged over 400 receiver antenna array locations, to
obtain the trend as a function of the number of antennas and of pedestrians. The
algorithms were implemented on MATLAB with double-precision floating-point
values. The OFDM parameters used in the simulations are identical to those used
for the measurements. Using the implemented simulation all the variation trends
of the capacity dynamic range due to the human body shadowing effects have been
captured.
8.2 Research Outcomes
8.2.1 Journals
1. Das Gupta, Jishu and Suzuki, Hajime and Ziri Castro, Karla (2009) Effect of
Pedestrian Movement on MIMO-OFDM Channel Capacity in an Indoor Environ-
ment. IEEE Antennas and Wireless Propagation Letters. vol.8, no.3, pages
Chapter 8 Conclusions and Future Work
8.2 Research Outcomes 164
(682-685), 2009.
2. Das Gupta, Jishu and Howard, Sreckko and Howard, Angela (2010) Com-
parison analysis of Range of Dynamic Variation in a populated indoor environ-
ment(Under Review).
3. Das Gupta, Jishu (2010) A Systematic Study of MIMO-OFDM Broadband
Channels in Populated Indoor Environment.(IETE Journal of Research)(In Press).
4. Das Gupta, Jishu and Ziri-Castro, Karla (2010) Effect of Pedestrian Move-
ment on MIMO-OFDM Channel Capacity in an Indoor Environment. Ieee Trans-
actions On Antennas And Propagation, (ERA rank: A, Impact factor: 2.479).
(Submitted/ Under Review).
8.2.2 Conferences
1. H. Tan, J. Das Gupta and K. Ziri-Castro (2010) HUMAN-BODY SHADOWING
EFFECTS ON INDOOR MIMO-OFDM CHANNELS AT 5.2 GHz European Con-
ference on Antennas and Propagation 2010, 12-16 April 2010, Barcelona.
2. Das Gupta, Jishu and Ziri-Castro, Karla I. (2009) Body-shadowing effects in
indoor MIMO-OFDM channel capacity. In: Proceedings of Australasian Telecom-
munications Networks and Applications Conference, 9-11 November 2009, Na-
tional Convention Centre, Canberra.
3. Das Gupta, Jishu and Ziri Castro, Karla (2009) Variations in MIMO-OFDM
channel capacity due to random human movement in an indoor environment. In:
Loughborough Antennas and Propagation Conference 2009, 16-17 November
2009, Loughborough, UK.
4. Das Gupta, Jishu and Ziri Castro, Karla (2009) Pedestrians Effects on In-
door MIMO-OFDM Channel Capacity. In: The 5th International Conference
on Wireless Communications, Networking and Mobile Computing (WiCOM
2009), September 24-26, 2009, Beijing.
5. Das Gupta, Jishu and Ziri-Castro, Karla I. and Suzuki, Hajime (2007) Ca-
pacity analysis of MIMO-OFDM broadband channels in populated indoor environ-
Chapter 8 Conclusions and Future Work
8.3 Future Research Topics 165
ments. In: 7th International Symposium on Communications and Information
Technologies, 17-19 October 2007, Australia.
6. Das Gupta, Jishu and Ziri-Castro, Karla I. and Suzuki, Hajime (2007) Cor-
relation analysis on MIMO-OFDM channels in populated time varying indoor en-
vironment. In: 10th Australian Symposium on Antennas, 14-15 February 2007,
Sydney, Australia.
7. Das Gupta, Jishu and Ziri-Castro, Karla I. and Suzuki, Hajime (2007) Dy-
namic range analysis on MIMO-OFDM broadband channels in a populated time-
varying indoor environment. In: 2007 Australian Telecommunication Networks
and Applications Conference (ATNAC), 2-5 December 2007, Christchurch, New
Zealand.
8. Das Gupta, Jishu and Ziri-Castro, Karla I. and Suzuki, Hajime (2006) Time
variation characteristics of MIMO-OFDM broadband channels in populated indoor
environments. In: 2006 Australian Telecommunications, Networks and Appli-
cations Conference (ATNAC), 4-6 December 2006, Melbourne, Australia.
8.3 Future Research Topics
8.3.1 Real time MIMO-OFDM Channel Modeling for Realistic
Environment
In this thesis, we have concentrated on developing a deterministic channel predic-
tion model and characterized a channel for implementing a reliable and effective
indoor channel model. For simplicity we have predicted and presented the results
based on controlled and somewhat uncontrolled pedestrian movement, which is im-
portant to understand the channel in its earlier stage. Future research should be di-
rected in a more realistic environment, with more people present in different com-
bined scenarios (such as standing, running, moving fast/slow etc). Besides this,
simulation also requires to be upgraded for such environment.
Chapter 8 Conclusions and Future Work
8.3 Future Research Topics 166
8.3.2 Quality Improvement using Controlled Scattering Fixture
Multipath plays a crucial role in improving the quality of the propagating channel,
hence increases the throughput significantly. To achieve higher data rate within the
indoor environment in the presence of pedestrian, industrial equipment controlled
scattering fixture can significantly increase the receivable power. Further research
on such phenomenon can improve the channel modeling technique and improve the
channel characterization for WLAN design.
Chapter 8 Conclusions and Future Work
REFERENCES 167
References
[1] Kozuch, “Mobile phone subscribers per 100 inhabitants 1997-2007,” Web,
2008.
[2] H. Hashemi, “The Indoor Radio Propagation Channel,” Proceedings Of The
IEEE, vol. 81, no. 7, pp. 943–968, July 1993.
[3] D. Gesbert, M. Shafi, D. s. Shiu, P. J. Smith, and A. Naguib, “From Theory
to Practice: An Overview of MIMO Space-Time Coded Wireless Systems,”
IEEE Journal On Selected Areas In Communications, vol. 4, no. 3, pp. 1186–
1194, May 2005.
[4] K. Ziri-Castro, W. G. Scalon, and F. Tofoni, “Dynamic Capacity Estimation
for the Indoor Wireless Channel with MIMO Arrays and Pedestrian Traf-
fic,” Proc. 1st Joint IEI/IEE Symposium on Telecommunications Systems
Research, Nov/Dec 2001.
[5] J. D. Gupta, K. Ziri-Castro, and H. Suzuki, “Capacity analysis of MIMO-
OFDM broadband channels in populated indoor environments,” 7th Inter-
national Symposium on Communications and Information Technologies, pp.
273–278, Oct 2007.
[6] ITU, “World telecommunication development report 2002,” Web, Oct 2003.
[7] M. J. Gans G. J. Foschini, “On limits of wireless communications in a fading
environment when using multiple antennas,” Wireless Personal Communica-
tion, vol. 6, no. 3, pp. 311–335, March 1998.
REFERENCES
REFERENCES 168
[8] G. G. Raleigh and J. M. Cioffi, “Spatio-temporal coding for wireless com-
munication,” IEEE Transactions On Communications, vol. 46, no. 3, pp.
357–367, March 1998.
[9] C. N. Chuah, J. M. Kahn, and D. Tse, “Capacity of multi-antenna array
systems in indoor wireless environment,” Proc. IEEE Globecom, vol. 4.
[10] S. M. Alamouti, “A Simple Transmit Diversity Technique for Wireless Com-
munications,” IEEE Journal on Selected Areas in Communications, vol. 16,
no. 8, pp. 1451–1458, Oct 1998.
[11] 802.11 Working Group of the 802 Committee, “Draft Standard for Informa-
tion Technology- Telecommunications and information exchange between
systems- Local and metropolitan area networks- Specific requirements- Part
11: Wireless LAN Medium Access Control (MAC) and Physical Layer
(PHY) specifications: Amendment 4: Enhancements for Higher Through-
put,” September 2007.
[12] IEEE Standard IEEE 802.11b, “Part 11: Wireless LAN Medium Access Con-
trol (MAC) and Physical Layer (PHY) specifications: Higher-Speed Physical
Layer Extension in the 2.4 GHz Band,” September 1999.
[13] S. Noh, Y. Jung, S. Lee, and J. Kim, “Low-Complexity Symbol Detector for
MIMO-OFDM-Based Wireless LANs,” IEEE Transactions On Circuits And
SystemsII: Express Briefs, vol. 53, no. 12, pp. 1402–1407, December 2006.
[14] S. A. Mujtaba, “TGn Sync proposal technical specification,” IEEE 802.11-
04/0889r7, July 2005.
[15] H. Suzuki, T. V. A. Tran, and I. B. Collings, “Characteristics of MIMO-
OFDM channels in indoor environments,” EURASIP Journal on Wireless
Communications and Networking, vol. 2007, no. 19728, January 2007.
REFERENCES
REFERENCES 169
[16] K. Ziri-Castro, W. G. Scalon, and N. E. Evans, “Prediction of variation in
MIMO channel capacity for the populated indoor environemnt using a radar
cross-section-based pedestrian model,” IEEE Transection on Wireless Com-
munications, vol. 21, no. 3, pp. 281–302, April 2003.
[17] H. Suzuki, “Characteristics of 4x4 MIMO-OFDM channels in indoor envi-
ronment,” Proc. ClimDiff 05, September 2005.
[18] J. W. Wallace, M. A. Jensen, A. L. Swindlehurst, and B. D. Jeffs, “Experi-
mental characterization of the MIMO wireless channel: data acquisition and
analysis,” IEEE Transactions On Wireless Communications, vol. 2, no. 2, pp.
335–343, March 2003.
[19] E. Telatar, “Capacity of Multi-antenna Gaussian Channels,” European Trans-
actions on Telecommunications, vol. 10, no. 6, pp. 585–595, Dec 1999.
[20] S. Y. Seidel and T. S. Rappaport, “914 MHz path loss prediction models for
indoor wireless communications in multifloored buildings,” IEEE Transac-
tions on Antennas and Propagation, vol. 40, pp. 207–217, Feb 1992.
[21] T. S. Rappaport and V. Fung, “Simulation of bit error performance of
FSK, BPSK, and DQPSK in flat fading indoor radio channels using a
measurement-based channel model,” IEEE Transactions on Vehicular Tech-
nology, vol. 40, pp. 731–740, Nov 1991.
[22] H. Suzuki, Measurement, Prediction, and Analysis of Indoor Wireless Chan-
nel Characteristic Maps, Ph.D. thesis, University of Technology, Sydney,
1999.
[23] D. Gesbert, M. Shafi, D. S. Shiu, and P. J. Smith, “From Theory to Prac-
tice: An Overview of MIMO Space-Time Coded Wireless Systems,” IEEE
Journal On Selected Areas In Communications, vol. 46.
REFERENCES
REFERENCES 170
[24] M. A. Jensen and J. W. Wallace, “A Review of Antennas and Propagation
for MIMO Wireless Communications,” IEEE Transactions On Antennas And
Propagation, vol. 52, no. 11, pp. 2810–2824, November 2004.
[25] Z. Tang and A. S. Mohan, “Characterize the Indoor Multipath Propaga-
tion for MIMO Communication,” Asia-Pacific Conference Proceedings,
APMC2005, vol. 4, no. 10.1109/APMC.2005.1606922, pp. 4–7, Dec 2005.
[26] K. Ziri-Castro, W. G. Scalon, and N. E. Evans, “Measured pedestrian move-
ment and bodyworn terminal effect for the indoor chanel at 5.2 GHz,” Euro-
pean Transection on Telecommunications, Nov/Dec 2003.
[27] W. G. Scalon and K. Ziri-Castro, “modeling of MIMO channels for the
populated indoor environment,” IEE Seminar on MIMO: Communications
Systems from Concept to Implementations.
[28] G. L. Turin, F. D. Clapp, T. L. Johnston, S. B. Fine, and D. Lavry, “A sta-
tistical model of urban multipath propagation,” IEEE Transactions on In
Vehicular Technology, vol. 21, no. 1, pp. 1–9.
[29] M. Kountouris, A. Pandharipande, H. Kim, and D. Gesbert, “QoS-based user
scheduling for multiuser MIMO systems,” IEEE 61st Vehicular Technology
Conference, vol. 1.
[30] P. Uthansakul and M. E. Bialkowski, “Multipath Signal Effect on the Ca-
pacity of MIMO, MIMO-OFDM and Spread MIMO-OFDM,” 15th Inter-
national Conference on Microwaves, Radar and Wireless Communications,
2004. MIKON-2004, vol. 3.
[31] G. Strang, Linear Algebra and its Applications, Harcourt Brace Jovanovich,
San Diego, third edition, 1988.
REFERENCES
REFERENCES 171
[32] G. Zou, S. Zhang, L. Wang, Y. Wang, and M. Li, “MIMO-OFDM channel
modeling and simulation for application in ocean exploring,” IET Conference
on Wireless, Mobile and Sensor Networks(CCWMSN07).
[33] V. Tarokh, N. Seshadri, and A. R. Calderbank, “Space- Time Codes for
High Data Rate Wireless Communication: Performance Criterion and Code
Construction,” IEEE Transactions on Information Theory, vol. 44, no. 2, pp.
744–765, March 1998.
[34] H. Bolcskei, “MIMO-OFDM wireless systems: Basics, perspectives, and
challenges,” IEEE Wireless Communications, vol. 13, no. 4, pp. 31–37, Au-
gust 2006.
[35] A. J. Paulraj and T. Kailath, “Increasing Capacity in Wireless Broadcast
Systems Using Distributed Transmission Directional Reception,” U.S. Patent
no. 5,345,599.
[36] G. J. Foschini, “Layered Space-Time Architecture for Wireless Communi-
cation in a Fading Environment when Using Multielement Antennas,” Bell
Labs Technical Journal.
[37] O. Edfors, M. Sandell, J. Beek, D. Landstrom, and F. Sjiberg, “An introduc-
tion to orgonal frequency-division multiplexing,” Report from Digital Signal
Labs.
[38] P. Hammarberg and O. Edfors, “A Comparison of DFT and SVD Based
Channel Estimation in MIMO OFDM Systems,” IEEE 17th International
Symposium on personal, Indoor and Mobile Radio Communications.
[39] R. J. C. Bultitude, “Measurement, characterization and modeling of indoor
800/900 MHz radio channels for digital communications,” IEEE Communi-
cations Magazine, vol. 25, no. 6, pp. 5–12, Jun 1987.
REFERENCES
REFERENCES 172
[40] T. S. Rappaport and C. D. McGillem, “UHF fading in factories,” IEEE
Journal on Selected Areas in Communication, vol. 7, no. 1, pp. 40–48, Jan
1989.
[41] T. S. Rappaport, “Indoor radio communications for factories of the future,”
IEEE Communication Magazine.
[42] S. J. Howard and K. Pahlavan, “Doppler spread measurements of the indoor
radio channels,” Electronics Letters, vol. 26, no. 2, pp. 107–109, Jan 1990.
[43] S. J. Howard and K. Pahlavan, “Fading results from narrowband measure-
ments of the indoor radio channel,” IEEE International Symposium on Per-
sonal, Indoor and Mobile Radio Communications.
[44] H. Suzuki, M. Hedley, G. Daniels, and C. Jacka, “Spatial distribution of
temporal variation caused by active fluorescent lights in office environment,”
In Proceedings of the URSI Commission F Triennium Opem Symposium, pp.
181–186, June 2004.
[45] P. Melanon and J. Lebel, “Effects of fluorescent lights on signal fading char-
acteristics for indoor radio channels,” Electronics Letters, vol. 28, no. 18, pp.
1740–1741, Aug 1992.
[46] W. J. Vogel, H. Ling, and G. W. Torrence, “Fluorescent light interaction with
personal communication signals,” IEEE Transactions on Communications,,
vol. 43, no. 2/3/4, pp. 194–197, Feb/Mar/Apr 1995.
[47] J. D. Gupta, H. Suzuki, and K. Ziri-Castro, “Effect of Pedestrian Movement
on MIMO-OFDM Channel Capacity in an Indoor Environment,” IEEE An-
tennas and Wireless Propagation Letters, vol. 8, no. 3, pp. 682–685, August
2009.
REFERENCES
REFERENCES 173
[48] T. Manabe K. Sato, “Estimation of Propagation-Path Visibility for Indoor
Wireless LAN Systems under Shadowing Condition by Human Bodies,” 48th
IEEE Vehicular Technology Conference, vol. 3.
[49] K. Yu, M. Bengtsson, B. Ottersten, D. McNamara, P. Karlsson, and
M. Beach, “Modeling of wide-band MIMO radio channels based on NLOS
indoor measurements,” IEEE Transactions on Vehicular Technology, vol. 53,
no. 3, pp. 655–665, May 2004.
[50] A. van Zelst and T. C. W. Schenk, “Implementation of a MIMO-OFDM-
based wireless LAN system,” IEEE Transactions on Signal Processing, vol.
52, no. 2, pp. 483–494, February 2004.
[51] A. J. Paulraj, D. A. Gore, R. U. Nabar, and H. Bolcskei, “An overview
of MIMO communications - a key to gigabit wireless,” Proceedings of the
IEEE, vol. 92, pp. 198 – 218, Feb 2004.
[52] C. Hermosilla, R. Feick, R. Valenzuela, and L. Ahumada, “Improving
MIMO Capacity with Directive Antennas for Outdoor-Indoor Scenarios,”
IEEE Transactions on Wireless Communications, vol. 8, pp. 2177–2188, May
2009.
[53] M. Debbah and R. R. Muller, “MIMO channel modeling and the principle of
maximum entropy,” IEEE Transactions on Information Theory, vol. 51, pp.
1667–1690, May 2005.
[54] A. Goldsmith, S. A. Jafar, N. Jindal, and S. Vishwanath, “Capacity limits of
MIMO channels,” IEEE Journal on Selected Areas in Communications, vol.
21, pp. 684–702, June 2003.
[55] P. Kyritsi, D. C. Cox, R. A. Valenzuela, and P. W. Wolniansky, “Correla-
tion Analysis Based on MIMO Channel Measurements in an Indoor Envi-
ronment,” IEEE Journal On Selected Areas In Communications, vol. 21, no.
5, pp. 713–720, June 2003.
REFERENCES
REFERENCES 174
[56] D. S. Shiu, G. J. Foschini, M. J. Gans, and J. M. Kahn, “Fading Correlation
and Its Effect on the Capacity of Multielement Antenna Systems,” IEEE
Transactions On Communications, vol. 48, no. 3, pp. 502–513, March 2000.
[57] D. L. Goeckel and G. Ananthaswamy, “On the design of multidimensional
signal sets for OFDM systems,” IEEE Transactions On Communications,
vol. 50, pp. 442–452, Mar 2002.
[58] J. Gao, O. C. Ozdural, S. H. Ardalan, and H. Liu, “Performance Modeling
of MIMO OFDM Systems via Channel Analysis,” IEEE Transactions On
Wireless Communications, vol. 5, pp. 2358 – 2362, Sep 2006.
[59] G. Yi and K. B. Letaief, “An efficient space-frequency coded OFDM system
for broadband wireless communications,” IEEE Transactions on Communi-
cations, vol. 51, pp. 2019–2029, Dec 2003.
[60] N. Sasaoka, S. Danno, and Y. Itoh B. Sai, “Noise cancellation method for
OFDM systems,” IEEE-Eurasip Nonlinear Signal and Image Processing,
2005. NSIP 2005. Abstracts.
[61] Y. SUN, “Bandwidth-efficient wireless OFDM,” IEEE Journal on Selected
Areas in Communications, vol. 19, pp. 2267–2278, Nov 2003.
[62] C. Dubuc, D. Starks, T. Creasy, and Y. Hou, “A MIMO-OFDM prototype for
next-generation wireless WANs,” IEEE Communications Magazine, vol. 42,
pp. 82–87, Dec 2004.
[63] W. Zhang, X. Xiang-Gen, and K. B. Letaief, “Space-time-frequency coding
for MIMO-OFDM in next generation broadband wireless systems,” IEEE
Wireless Communications, vol. 14, no. 3, pp. 32–43, June 2007.
[64] B. M. Hochwald, G. Caire, and T. L. Marzetta, “The Academic and industrial
embrace of space-time methods,” IEEE Transection on Information Theory,
Oct 2003.
REFERENCES
REFERENCES 175
[65] S. Coffey, A. Kasher, and A. Stephens, “Joint Proposal: High throughput
extension to the 802.11 Standard: PHY,” IEEE 802.11-05/1102r4, January
2006.
[66] H. Suzuki, M. Hedley, G. Daniels, and J. Yuan, “Performance of MIMO-
OFDM-BICM on measured indoor channels,” In Proceedings of the IEEE
Vehicular Technology Conference, vol. 5, pp. 2073–2077, May 2006.
[67] M. Hedley, H. Suzuki, and G. Daniels, “Computational efficiency of list
sphere detection,” In Proceedings of the Australian Communications Theory
Workshop, pp. 66–70, February 2006.
[68] B. M. Hochwald and S. ten Brink, “Achieving near-capacity on a multiple-
antenna channel,” IEEE Transactions on Communications, vol. 51, no. 3, pp.
389–399, March 2003.
[69] B. Lu, Y. Guosen, and W. Xiaodong, “Performance analysis and design
optimization of LDPC-coded MIMO OFDM systems,” IEEE Transactions
on Signal Processing, vol. 52, pp. 348–361, Feb 2004.
[70] M. Hlaing, L. Yinghui, and N. Al-Dhahir, “Optimal Training Signals for
MIMO OFDM Channel Estimation in the Presence of Frequency Offset and
Phase Noise,” IEEE Transactions of Communications, vol. 54, pp. 1754–
1759, Oct 2006.
[71] R. W. Chang, “Synthesis of Band Limited Orthogonal Signals for Multi-
channel Data Transmission,” Bell System Technology Journal, vol. 45.
[72] IEEE Standard 802.11a, “Part 11: Wireless LAN Medium Access Con-
trol (MAC) and Physical Layer (PHY) Specifications: High-Speed Physical
Layer in the 5 GHz Band,” .
REFERENCES
REFERENCES 176
[73] IEEE Standard IEEE 802.16a, “Local and Metropolitan Area NetworksPart
16, Air Interface for Fixed Broadband Wireless Access Systems, IEEE Stan-
dard IEEE 802.16a,” .
[74] G. L. Stuber, J. R. Barry, S. W. McLaughlin, Y. G. Li, M. A. Ingram, and T. G.
Pratt, “Broadband MIMO-OFDM wireless communications,” Proceedings
of the IEEE, vol. 92, no. 2, pp. 271–294, February 2004.
[75] L. J. Cimini, Jr., “Analysis and simulation of a digital mobile channel using
orthogonal frequency division multiplexing,” IEEE Transactions on Com-
munications, vol. 33, no. 7, pp. 665–675, July 1985.
[76] A. Paulraj and T. Kailath, “Increasing capacity in wireless broadcast systems
using distributed transmission/directional reception (DTDR),” .
[77] G. J. Foschini, “Layered Space-Time Architecture for Wireless Communi-
cation in a Fading Environment When Using Multi-Element Antennas,” Bell
Labs Technical Journal, vol. 1, no. 2, pp. 41–59, Autumn 1996.
[78] S. Nanda, R. Walton, J. Ketchum, M. Wallace, and S. Howard, “A high-
performance MIMO OFDM wireless LAN,” IEEE Communications Maga-
zine, vol. 43, no. 2, pp. 101–109, February 2005.
[79] S. P. Alex and L. M. A. Jalloul, “Performance Evaluation of MIMO in
IEEE802.16e/WiMAX,” IEEE Journal Of Selected Topics In Signal Pro-
cessing, vol. 2, pp. 181–190, April 2008.
[80] C. Mecklenbrauker and S. Ivan, “Publishable Final Activity Report on MAS-
COT,” .
[81] H. Lin, R. C. Changand K. Lin, and C. Hsu, “Implementation of Synchro-
nization for 22 MIMO WLAN System,” IEEE Transactions on Consumer
Electronics, vol. 52, pp. 766–773, Aug 2006.
REFERENCES
REFERENCES 177
[82] H. Zhang, A. F. Molisch, and J. Zhang, “Applying Antenna Selection in
WLANs for Achieving Broadband Multimedia Communications,” IEEE
Transactions On Broadcasting, vol. 52, pp. 475–482, Dec 2006.
[83] Y. Q. Bian, A. R. Nix, E. K. Tameh, and J. P. McGeehan, “MIMOOFDM
WLAN Architectures, Area Coverage, and Link Adaptation for Urban
Hotspots,” IEEE Transactions On Vehicular Technology, vol. 57, pp. 2364–
2374, July 2008.
[84] WAVE Report, “OFDM Tutorial,” Covering Next Generation Computing
and Communication Technologies.
[85] J. P. Kermoal, L. Schumacher, K. I. Pedersen, P. E. Mogensen, and F. Fred-
eriksen, “A stochastic MIMO radio channel model with experimental vali-
dation,” IEEE Journal on Selected Areas in Communications, vol. 20, no. 6,
pp. 1211–1226, August 2002.
[86] H. Suzuki, M. Hedley, G. Daniels, and I. B. Collings, “Implementation of 4x4
MIMO-OFDM-LDPC achieving 600 Mbps packet transmission,” In Pro-
ceedings of the Australian Telecommunication Networks and Applications
Conference, pp. 440–444, December 2006.
[87] K. Bialkowski and A. Postula, “A Testbed For Evaluating Performance Of
An Indoor Mimo System,” 10th Australian Symposium on Antennas, Sydney,
February 2007.
[88] D. P. McNamara, M. A. Beach, P. N. Fletcher, and P. Karlsson, “Initial
investigation of multiple-input multiple-output (MIMO)channels in indoor
environments,” Symposium on Communications and Vehicular Technology
SCVT-2000.
[89] S. Lang and B. Daneshrad, “Design and implementation of a 5.25GHz ra-
dio transceiver for a MIMO testbed,” IEEE Wireless Communications and
Networking Conference, vol. 1.
REFERENCES
REFERENCES 178
[90] M. S. K. Heejung, Y. E. Y. Choi, T. Jeon, and S. K. Lee, “Design and Pro-
totype Development of MIMO-OFDM for Next Generation Wireless LAN,”
IEEE Transactions on Consumer Electronics, vol. 51, no. 4, pp. 1134–1142,
November 2005.
[91] K. Nishimori, R. Kudo, Y. Takatoti, A. Ohta, and K. Tsunekawa, “Perfor-
mance Evaluation of 8x8 Multi-User MIMO-OFDM Testbed in an Actual
Indoor Environment,” IEEE 17th International Symposium on Personal, In-
door and Mobile Radio Communications.
[92] N. Kita, W. Yamada, A. Sato, D. Mori, and S. Uwano, “Measurement
of Demmel condition number for 2x2 MIMO-OFDM broadband channels,”
Proceedings of 59th IEEE Vehicular Technology Conference (VTC 04), vol.
1.
[93] A. Gupta, A. Forenza, and R. W. Heath Jr., “Rapid MIMOOFDM software
defined radio system prototyping,” Proceedings of IEEE Workshop on Signal
Processing Systems Design and Implementation, Austin, Tex, USA.
[94] R. M. Rao, S. Lang, and B. Daneshrad, “Indoor field measurements with a
configurable multi-antenna testbed,” Proceedings of IEEE Global Telecom-
munications Conference (GLOBECOM04), Dallas, Tex,USA, vol. 6.
[95] H. Suzuki, T. V. A. Tran, I. B. Collings, G. Daniels, and M. Hedley, “Trans-
mitter noise effect on the performance of a MIMO-OFDM hardware imple-
mentation achieving improved coverage,” IEEE Journal on Selected Areas
in Communications, vol. 26, no. 6, pp. 867–876, August 2008.
[96] P. Almers, E. Bonek, A. Burr, N. Czink, M. Debbah, V. Degli-Esposti,
H. Hofstetter, P. Kyosti, D. Laurenson, G. Matz, A. F. Molisch, C. Oest-
ges, and H. Ozcelik, “Survey of Channel and Radio Propagation Models for
Wireless MIMO Systems,” EURASIP Journal on WIreless Communications
and Networking, vol. 2007.
REFERENCES
REFERENCES 179
[97] J. W. Wallace and M. A. Jensen, “Modeling the Indoor MIMO Wireless
Channel,” IEEE Transactions on Information Theory, vol. 50, no. 5, pp.
591–599, May 200.
[98] M. R. McKay and I. B. Collings, “Capacity and performance of MIMO-
BICM with zero-forcing receivers,” IEEE Transactions on Communications,
vol. 53, no. 1, pp. 74–83, January 2005.
[99] K. Z. Castro, Wireless Channel modeling for Indoor Populated Environ-
ments, Ph.D. thesis, Queens University Belfast, UK.
[100] O. Fernandez, M. Domingo, and R. P. Torres, “Empirical analysis of the cor-
relation of MIMO channels in indoor scenarios at 2GHz,” IEE Proceedings-
Communications, vol. 152, pp. 82–88, Feb 2005.
[101] H. Suzuki, B. Murray, A. Grancea, R. Shaw, J. Pathikulangara, and I. B.
Collings, “Real-time wideband MIMO demonstrator,” In Proceedings of the
7th International Symposium on Communications and Information Technolo-
gies, pp. 284–289, October 2007.
[102] H. Suzuki, Z Tang, and I. B. Collings, “Measurement and modeling of indoor
MIMO-OFDM channels,” In Proceedings of the International Symposium on
Advanced Radio Technologies, pp. 27–35, February 2007.
[103] B. N. Getu and R. Janaswamy, “The Effect of Mutual Coupling on the Capac-
ity of the MIMO Cube,” IEEE Antennas And Wireless Propagation Letters,
vol. 4, pp. 240–244, May 2005.
[104] P. N. Fletcher, M. Dean, and A. R. Nix, “Mutual coupling in multi-element
array antennas and its influence on MIMO channel capacity,” Electronics
Letters, vol. 39, no. 4, pp. 342–344.
REFERENCES
REFERENCES 180
[105] J. W. Wallace and M. A. Jensen, “Mutual coupling in MIMO wireless sys-
tems: a rigorous network theory analysis,” IEEE Transactions on Wireless
Communications, vol. 3, no. 4, pp. 1317–1325.
[106] T. Kaiser, A. Bourdoux, M. Rupp, and U. Heute, “Special Issue on ’Imple-
mentation aspects and testbeds for MIMO systems’,” EURASIP Journal on
Advances in Signal Processing, vol. 2006, pp. 1–3.
[107] H. Suzuki and A. S. Mohan, “Measurement and prediction of high spatial
resolution indoor radio channel characteristic map,” IEEE Transactions on
Vehicular Technology, vol. 49, no. 4, pp. 1321–1333, July 2000.
[108] U. Dersch and E. Zollinger, “Propagation Mechanisms in Microcell and
Indoor Environments,” IEEE Transactions On Vehicular Technology, vol.
43, no. 4, pp. 1058–1066, Nov 1994.
[109] S. H. Chen and S .K. Jeng, “An SBR/image approach for radio wave prop-
agation in indoor environments with metallic furniture,” IEEE Transactions
On Antennas And Propagation, vol. 45, no. 1, pp. 98–106, Jan 1997.
[110] K. Sato, T. Manabe, T. Ihara, H. Saito, S. Ito, T. Tanaka, K. Sugai, N. Ohmi,
Y. Murakami, M. Shibayama, Y. Konishi, and T. Kimura, “Measurement
of reflection and transmission characteristics of interior structures of office
building in the 60-GHz band,” IEEE Transactions On Antennas And Propa-
gation, vol. 45, no. 12, pp. 1783–1792, Dec 1997.
[111] H. Suzuki, “Accurate and efficient prediction of coverage map in an office
environment using frustum ray tracing and in-situ penetration loss measure-
ment,” IEEE Vehicular Technology Conference, pp. 236–240, April 2003.
[112] J. D. Gupta, K. Ziri-Castro, and H. Suzuki, “Dynamic range analysis on
MIMO-OFDM broadband channels in a populated time-varying indoor en-
vironment,” 2007 Australian Telecommunication Networks and Applications
Conference, Dec 2007.
REFERENCES
REFERENCES 181
[113] J. D. Gupta, K. Ziri-Castro, and R. Addie, “Time variation characteristics of
MIMO-OFDM broadband channels in populated indoor environments,” 2006
Australian Telecommunications Networks and Applications Conference, Dec
2006.
[114] J. D. Gupta and K. Ziri-Castro, “Pedestrians Effects on Indoor MIMO-
OFDM Channel Capacity,” The 5th International Conference on Wireless
Communications, Networking and Mobile Computing (WiCOM 2009), Sep
2009.
[115] T. Svantesson and J. Wallace, “Statistical characterization of the indoor
MIMO channel based on LOS/NLOS measurements,” 36th IEEE Asilomar
Conference in Signals, Systems, Computers, vol. 2.
[116] S. Howard, H. Inanoglu, J. Ketchum, M. Wallace, and R. Walton, “Results
from MIMO channel measurements,” The 13th IEEE International Sympo-
sium on Personal, Indoor and Mobile Radio Communications, vol. 4.
[117] G. Hankins, L. Vahala, and J. H. Beggs, “Electromagnetic propagation pre-
diction inside aircraft cabins,” IEEE Antennas and Propagation Society In-
ternational Symposium, vol. 3, pp. 2227–2230, June 2004.
[118] M. Youssef, L. Vahala, and J. H. Beggs, “Wireless network simulation in
aircraft cabins,” IEEE Antennas and Propagation Society International Sym-
posium, vol. 3, pp. 2223–2226, June 2004.
[119] J. D. Gupta and K. Ziri-Castro, “Variations in MIMO-OFDM channel capac-
ity due to random human movement in an indoor environment,” Loughbor-
ough Antennas and Propagation Conference 2009, Nov 2009.
[120] J. D. Gupta and K. Ziri-Castro, “Body-shadowing effects in indoor MIMO-
OFDM channel capacity,” Australasian Telecommunications Networks and
Applications Conference, Nov 2009.
REFERENCES
Top Related