Adaptive/Reconfigurable Wireless Communications · source source encoder {1, 2, …, q} discrete...
Transcript of Adaptive/Reconfigurable Wireless Communications · source source encoder {1, 2, …, q} discrete...
informationsource
sourceencoder
{1, 2, …, q}
discrete memoryless source
encryptor channelencoder
datamodulator
spreadspectrum
modulator
channel &network
receiverfrontend
spreadspectrum
despreader
channelestimation
Spreading codegenerator
receive diversity
decryptorsourcedecoder
informationsink
transmit diversity(multiple access)
DatademodulatorMU MLSE
Transceiver configurationcontrol
poweramplification
(powerlimitation)
channeldecoder
Higher layers
Adaptive segments of the system
• adaptive coding, modulation& equalization• power control• time/space/frequency domainspace-time coding/multi-carrier/OFDM
• intertechnology adaptation/roamingTDMA/CDMA/ad hock networks/UWB• minimum complexity/power consumption• bit rate /packet length• adaptive access control/Kalman filter/Fuzzy logic• routing/• source coding/• reconfigurable network architecture/ad - hock network/clustering/node mobility & activity
active networks /operators
PSTNPSTN PLMNPLMN
IP NetworkIP Network PrivateNetworkPrivate
NetworkIMT-2000Network
IMT-2000Network
IMT-2000Access
IMT-2000Access
BRANAccess
BRANAccess
MobileTerminalsMobile
Terminals
TDMA IS 136UMTS WCDMAup to 2MBit/scdma2000
802.112.4GHz (ISM)FHSS & DSSS5GHz
OFDM > 10Mbit/sHiperlan 25Mb(indoor)Hiperaccess(Wider area)Hiperlink 155MbSpace-time-frequency coding
Recofigurabity range
System Architecture for Maintaining QoS in mobile wireless System Architecture for Maintaining QoS in mobile wireless Multimedia NetworkingMultimedia Networking
Compression andPayload Control
Quality of ServiceManager
Quality of ServiceMeasurement
VC/RoutingTable
ConnectionManager
Error Controland
Frame Length
LinkController
ModemController
ChannelModel
Estimator
Processing Gainand
EqualizerCoefficients
ChannelEstimator
Set QoS Level
Set QoS Level
Update QoS Level
Update QoS Level
Adaptive Link Control
Adaptive Physical Layer
Processing Gain SpreadCode
CSI: SIR, CIR, fm
ReactiveApplication
Network
AdaptiveRadio
Scalable Codec
CompressionRadio
Block-Diagram
On-LineEstimation
PNGenerators
ReconfigurableBlock 1
ReconfigurableBlock 2
On-LineEstimation
Control andReconfiguration
Management
Control andReconfiguration
ManagementCorrelator
M
I/Q Baseband Signals
On-Line Estimates
PN Seq.
Qr
Qr
FPGA’s
DSP’s
Block-diagram of DSP/FPGA software radio implementation.
Reconfigurable ASIC Digital Communications
Data rate 125 Msymbols/sec
Green Wireless Network Concept
• Focus on human well-being and environmental concerns (reducing EM pollution) while preserving high-performance for the wireless network operation
• Give space for introducing new additional radio systems without health risk
• Should be incorporated in any future wireless networks, therefore referred to as All Generation (AG) networks
Environment-Friendly or Green Wireless Network (GWN)
• reduces inter cell interference (63%>5%) in FDD and • cross frame interference in TDD mode
Space/Rate Adaptive CLSP/DS-CDMA PRN
rm
r0
• Compensate the near-far effects and TPC • difficulty of datagram packet transmissions • under changing channel conditions;• Enhance the system performance and coverage;• Increase the energy-efficiency of mobile terminals;• Reduce inter-cell interference
Centralized Unslotted CLSP/DS-CDMA PRN
Allow MS operating with much lower max. power and reduce EM pollution.
System Parameters
Pt=P
Er=PT0
(rm=2-m/ξ)
r0=1(ring#)distance
rM r1
a) Adaptive System
Pt=Prξ
Tm=2-mT0
(Rm=2mR0)
Tm=T
Er=PT0
r0=1(ring#)log-distance
rM r1
b) Traditional System
• Mobile transmitter power is kept at a normalized level depending on mobile-location space resolution, which is based on propagation attenuation;
• Dynamic range of TPC is closer to 0 instead of e.g. approaching 80[dB] in traditional system that improve TPC resulting in smaller standard derivation of log-normal SINR;
• Transmission bit-rate in [kbps] and duration in [ms] compensate each other resulting in the same bit-energy given a constant packet-length in [bit];
• Higher bit-rate CDMA transmissions tends to need smaller SINR;
• Adaptive system is expected to have better multiplexing gain;
System Parameters (Cont.)
• Radio transmission of the mobile is not allowed in the user's direction: restriction angleα [o]
• The restricted-zone problem (with Pr=α [o]/360) can be resolved by:
– using receiving antenna (omni-directional) also for transmitting when the nearest hub enters the restricted zone;
– advising user to change holding position with control signal;
– redirecting the mobile to connect to othe hub which is not in the restricted zone though farther away;
directivity of receiving antenna
directivity of transmitting antenna
Attenna Patterns of the Mobile's
users directionα
Two-State HMM and Performance Analysis
HMM of System Load State
• Taking into account the impacts of TPC inaccuracy characterized by log-normal SINR, spatial user distributions and channel attenuation;
• Based on the stochastic Knapsack approximation of multi-rate loss network model;
• Mean of log-normal SINR is set the same regardless of bit-rates, also its standard deviation regardless of systems that is not to take advantages of the fixed system in performance comparison
Throughput-Delay Characteristics (1)
0 5 10 15 20 25 30 35 400
0.5
1
1.5
2
2.5
System Throughput
Nor
mal
ized
Ave
rage
Pac
ket D
elay
fixed system adaptive sys. with 1-D uni. SUDadaptive sys. with 2-D uni. SUD
• Performance of the adaptive system is sensitive to Spatial User Distribution (SUD) that affects offered traffic patterns;
• That more users are put closer to the hub can boost up the rate, and reduce the time of transmissions resulting in better throughput-delay tradeoffs
• This is more desirable if the advantages of less required SINR for higher bit-rate and less dynamic range of TPC are taken into account
Effects of Spatial User Distribution (SUD)
Throughput-Delay Characteristics (2)
0 5 10 15 20 25 30 35 400
0.5
1
1.5
2
2.5
System Throughput
Nor
mal
ized
Ave
rage
Pac
ket D
elay
fixed system adaptive sys. with atten. exponent=2adaptive sys. with atten. exponent=3adaptive sys. with atten. exponent=4
L=2560[bit], 2-D uni. SUD, Peb=5e-4
• Performance of the adaptive system is sensitive to attenuation path-loss exponent that affects offered traffic patterns, thus in the same way as of SUD;
• However, SUD is changing more dynamically
Effects of Radio Propagation Attenuation
Throughput-Delay Characteristics
0 5 10 15 20 25 30 35 400
0.5
1
1.5
2
2.5
3
3.5
4
4.5
System Throughput
Nor
mal
ized
Ave
rage
Pac
ket D
elay
fixed sys. with Peb=5e-4 adaptive sys. with Peb=5e-4fixed sys. with Peb=1e-3 adaptive sys. with Peb=1e-3
L=2560[bit], atten. exponent=3, 2-D uni. SUD
• Effects of bad channel conditions due to fading are investigated with introduction of Peb as BER in bad channel state of HMM;
• Performance of the adaptive system is more stable.
Effects of Bad Channel Conditions due to Fading
Throughput-Delay Characteristics
0 5 10 15 20 25 30 35 400
0.5
1
1.5
2
2.5
3
3.5
4
4.5
System Throughput
Nor
mal
ized
Ave
rage
Pac
ket D
elay
fixed sys. L=2560[bit] adaptive sys. L=2560[bit]fixed sys. L=5120[bit] adaptive sys. L=5120[bit]
2-D uni. SUD, atten. exponent=3, Peb=5e-4
• Performance of the adaptive system is also more stable under effects of packet-length in [bit];
• Smaller packet-length makes better system throughput-delay characteristics, but heavier load factor of the protocol overhead as well;
• This serves as the basis for optimum or adaptive packet-length problems subject to optimal systemgoodputEffects of Packet-Length
Conclusions
• The space/rate adaptive CLSP/DS-CDMA PRN system not only can accommodate the GWN concept but also outperforms the fixed counterpart in throughput-delay given the same coverage, offered traffic load and link-quality requirements
• The rate/space adaptive scheme also increases the robustness against the uncertainty of radio channel
PRNNinterference
predictor
Fuzzy/Neuralaccess probability
controller(FAPC/NAPC)
Fuzzyperformance
indicator
PV(n+1)PD(n+1)
LV(n)RC(n)U(n)
DD(n)
Is(n)
( )1~ +nIs
A(n)
Fuzzy/Neural Congestion Controller
A pipeline recurrent neural network (PRNN)
A DS-CDMA/FRMA cellular system with the fuzzy/neural congestion controller.
fuzzy/neural adaptive congestion controller
The rule structure for the fuzzy performance indicatorRule LV RC U DD A1 Hi Lt Sm Lg A12 Hi Lt La Lg A13 Hi Lt Sm Sh A14 Hi Lt La Sh A25 Lo Lt Sm Lg A26 Lo Lt La Lg A37 Lo Lt Sm Sh A38 Lo Lt La Sh A49 Lo Bg Sm Sh A510 Lo Bg La Sh A611 Lo Bg Sm Lg A612 Lo Bg La Lg A713 Hi Bg Sm Sh A714 Hi Bg La Sh A815 Hi Bg Sm Lg A816 Hi Bg La Lg A8
Rule IS A ∆P1 Hi La ∆P12 Hi Md ∆P13 Hi Sm ∆P24 Me La ∆P25 Me Md ∆P36 Me Sm ∆P47 Lo La ∆P58 Lo Md ∆P69 Lo Sm ∆P6
Fuzzy/neural congestion controller with NAPC
Fuzzy/neural congestioncontroller with FAPC
Channel access function(chapter 10.4.)
150 160 170 180 190 200
10-4
10-3
10-2
10-1
Number of users in a cell
Voice packet dropping ratio LV
NAPC
FAPC
Channel access function
150 160 170 180 190 20010-7
10-2
10-1
100
Number of users in a cell
Corruption ratio RC
NAPC
FAPC
Channel access function
150 160 170 180 190 200
0.6
0.7
0.8
0.9
Utilization U
Number of users in a cell
transmitter
transmitters
transmitters
Bandwidthmanager
(neighbour)
Bandwidthmanager
(neighbour)
Bandwidthmanager
receiver
Existing localconnections
Connectiondemand
Ad - traffic declarations
H- channelcharacteristics
Out-of-cell interference
Decision (accept or reject)
G - traffic statechannel characteristics
Kalman filter based adaptive access control
Receiver
RAKE
Decoded information
Interferencesamples
Is
BERc - channel bit error rate
Measurement(mean andvariance)
Interferenceprediction
adtraffic declaration(new connection)
ip, vp
( ) rIiyI P −<+ maxˆ Estimation
(Kalmanfilter)
( )yV
( )yI
Maximuminterference
threshold
Reservationthreshold
Imax
r
noyes
ACCEPT REJECT
( ) rIiyI P −<+ maxˆ
Estimation Decision
Average number of voice connections per cell with fixed power control [80]Fixed strategy Local strategy Global strategyA ∆A r2 ∆A r2 r3SU2 34.5 +4% 9 +7% 5 1SU4 60.3 +10% 9 +12% 5 1SC2 11.6 +20% 9 +22% 5 1SC4 14.7 +13% 9 +17% 5 1SX2 29.9 +3% 9 +8% 5 1SX4 50.8 +7% 9 +10% 5 1NX2 28.1 +5% 9 +10% 5 1NX4 50.8 +6% 9 +8% 5 1
Mean connections per cell with SIR power control [80]Fixed strategy Local strategy Global strategyA ∆A r2 ∆A r2 r3SC2 11.5 +20% 10 +24% 5 1SC4 14.7 +27% 10 +33% 5 1SX2 29.3 +15% 10 +31% 5 1SX4 51.3 +14% 10 +24% 5 1NX2 28.8 +16% 30 +17% 25 21NX4 52.2 +10% 30 +17% 25 21
Software distribution & business model(upgrading procedures/cycles)