Performance Analysis and Improvements for the Future Aeronautical Mobile Airport Communications System (AeroMACS)(AeroMACS)

Candidate: Paola PuliniCandidate: Paola PuliniAdvisor: Marco Chiani

Outline

Introduction and Motivations

Thesis SummaryMain Contributions

Unequal Diversity CodingFundamentalsFundamentalsSystem DescriptionPerformance Results

Conclusions

Introduction and Motivations (1/3)

Current aeronautical communications system - air traffic control (ATC) and air traffic management (ATM)

Voice based - Double-sideband amplitude modulation (DSB-AM)Data link based - VDL (VHF digital link) mode 2Capacity of the system is already saturatedCapacity of the system is already saturated

Necessity of a new aeronautical communications infrastructureRobust, efficient, secure, flexibleAble to cope with the future long-term increasing demands

Introduction and Motivations (2/3)New global aeronautical communications system

Air to air communications

Satellite-based communications

Air-to-air communications

Ground-based communications

communications

Airport communications (AeroMACS)

Introduction and Motivations (3/3)Airport Surface Communications

High demand of capacityNew frequencies assignments (ITU world radio conference 2007)

C band (5091 – 5150 MHz)

IEEE 802.16 standard (mobile WiMAX) has been chosen as the base technology for the future system (AeroMACS)

Thesis Summary

Analysis and investigation of the performance of AeroMACS

Evaluation of potential solutions for enhancing the performance of the system

Improvement through the introduction of diversity techniquesMIMO schemes with multiple antennas only on the control O sc e es t u t p e a te as o y o t e co t otower (space diversity)Cooperative Communications with single relay (cooperative di ersit )diversity)Packet level coding (time diversity)

Main Achievements (1/3)Preliminary Studies

Development of a novel stochastic airport channel modelParameters based on measurement campaign at MUCp g

Analysis of the performance of two WiMAX profiles in a realistic i t i tairport environment

OFDM based waveformOFDMA based waveformOFDMA based waveform

Selection of the most suitable profile for AeroMACSp

Main Achievements (2/3)

Analysis of MIMO schemes and relative performance evaluationSIMO 1x2 with MRC (reverse link)( )MISO 2x1 STC (forward link)Introduction of a novel implementation of 2x1 STC for A MACSAeroMACS

Investigation of the use of cooperative communications strategiesInvestigation of the use of cooperative communications strategies in the airport contextStudy and performance analysis of single relay schemes

Amplify and forwardDecode and forward

Main Achievements (3/3)

Introduction of the novel concept of unequal diversity (UD) coding for the relay channel (high efficiency)

Development of a novel class of LDPC codesAnalytical study of the codesA li ti f th d th d t A MACSApplication of the proposed method to AeroMACS

Packet level codingPacket level codingDevelopment of two algorithms for the online design of LDPC codesAnalytical study of the proposed methodApplication of the proposed scheme to AeroMACS

FundamentalsCooperative Communications

New paradigm based on the utilization of heterogeneous resources to improve the overall performance of the systemoverall performance of the system

Virtual antenna array by the combination of antennas of different users (also singleantennas of different users (also single antenna)

Distributed MIMO networkSpatial diversity (cooperativeSpatial diversity (cooperative diversity)

In a multi-user systemIn a multi user systemEach user represents a potential cooperative-partnerNo requirement of a dedicated

Potential cooperative-tNo requirement of a dedicated

infrastructure partners

Cooperative Communications

Example: Single relay (cooperation between partner A and B)

t1Source

t2

Partner

t2

Cooperative CommunicationsBasic Methods

Amplify and Forward

Decode and Forward

Coded Cooperation

Diversity gain of order 2Reduction of the efficiency overall coding rate ≤1/2 forReduction of the efficiency, overall coding rate ≤1/2 for achieving diversity-2

Cooperative CommunicationsHigh-Rate Coded Cooperation

We introduce a coded cooperation scheme which allows high code rates

Diversity-2 is guaranteed for a part of the source message

U l Di it (UD) E t i f l t tiUnequal Diversity (UD) Extension of unequal error protection Relevant for messages composed by parts having differentpriority/QoS requirementsp o ty/QoS equ e e ts

Video streamingAircraft communications (messages with different level of criticality) within the airport domain

Cooperative CommunicationsHigh-Rate Coded Cooperation

We propose a novel construction based on Low-Density Parity-Check codes which achieves the promised performance

We provide an analysis on block-fading channels complemented by simulationsby simulations

Low-Density Parity-Check CodesBasics

Low-Density Parity-Check Codes (Gallager,1960)Near-Shannon limit error correcting codes with iterative (message-passing) decoding

Parity-check matrix:Check nodes

100111001011010011011

H Tanner graph:

Check nodes

1001110

05421 cccc

Parity-check equations:

000

7432

6431

5421

cccccccccccc

Variable nodes

Low-Density Parity-Check CodesProtograph Construction of the Tanner Graph

Protograph: small bipartite graph describing the macroscopic structure of an LDPC code

Tanner Graph: obtained by Q-fold replication of the protograph and by edge permutation among the protograph replicasedge permutation among the protograph replicas

Low-Density Parity-Check CodesProtograph Construction of the Tanner Graph

For the LDPC code associated with the Tanner graph,Minimum distance propertiesIterative decoding threshold

depend on the starting protograph only

Code design reduces to protographdesign!

Additionally, protograph LDPC codes have structured parity-checkAdditionally, protograph LDPC codes have structured parity check matrices which facilitate the decoder implementation

Low-Density Parity-Check CodesProtograph Extrinsic Information Transfer (EXIT)

Protograph EXIT analysis: track the evolution of the message probability densities over theprobability densities over the protograph edges

Allows to accurately predict the iterative decoding threshold, i.e. the signal-to-noise ratio (SNR) at which iterative decoding starts to converge

EXIT analysis can be adapted toEXIT analysis can be adapted to block-fading channels

Low-Density Parity-Check CodesProtograph Analysis over Block Fading Channels (1/2)

Block fading channel (BFC): The codeword is split into

N blocksN blocksEach block is transmitted over a different flat fading channelEach block experiences aEach block experiences a different SNRIn each channel, the SNR follows an exponentialfollows an exponential distribution

Accurate model forFrequency-hoppingFrequency hoppingRelay communications

Low-Density Parity-Check CodesProtograph Analysis over Block Fading Channels (2/2)

We introduced a modification of the protographEXIT analysis in order to account for different SNRsfor each variable nodefor each variable node

Protograph variable nodes = codeword blocksFeed each protograph variable node with a

different SNR levelGiven a SNR profile, we determine whether

iterative decoding converges or not (outage)g g ( g )Outage region of a protograph G,The definition of outage region can be

extended to each single variable node (= block)extended to each single variable node (= block)

We can characterize the UEP of the protographnodes!

Protograph Analysis over the Relay ChannelBlock Fading Channel Approximation

Conventional assumption: the Source-Relay (S-R) link is reliableValid if the SNR over the S-R link is larger than the decoding threshold of the code employed at the Sourcethe code employed at the SourceRealistic assumption (relay selection protocol)

Approximation with block fading channel withtwo independent channels (=two fading levels / SNRs), and

Conventional Approach for the Relay ChannelCoded Cooperation

The Source encodes the packet with a (nS,k) code CS and broadcasts (Time Frame 1)The Relay decodes and re-encodes nR additional parity bits with a code CR, whichThe Relay decodes and re encodes nR additional parity bits with a code CR, which are sent to the Destination (Time Frame 2)

S D

R

S DTime frame 1

Time frame 2

The overall code has block length nS+nR the overall code rate is R=k/(nS+nR)Diversity-2 can be achieved only if R≤1/2New solution:

By re-encoding just a fraction of the information bits at the Relay, we can provide diversity-2 for certain codeword bits even if R>1/2…

Unequal Diversity CodingA New Scheme for the Relay Channel

The Source encodes the packet with an (nS,k) code CS and broadcasts (Time Frame 1)(Time Frame 1)The Relay decodes and re-encodes nR additional parity bits with a code CR, out of kh<k information bits R h(Time Frame 2)

u = informationword (k bits)

Unequal Diversity CodingA New Scheme for the Relay Channel

The Source encodes the packet with an (nS,k) code CS and broadcasts (Time Frame 1)(Time Frame 1)The Relay decodes and re-encodes nR additional parity bits with a code CR, out of kh<k information bits R h(Time Frame 2)

uh = high-priorityfragment (kh bits)

Unequal Diversity CodingA New Scheme for the Relay Channel

The Source encodes the packet with an (nS,k) code CS and broadcasts (Time Frame 1)(Time Frame 1)The Relay decodes and re-encodes nR additional parity bits with a code CR, out of kh<k information bits R h(Time Frame 2)

ul = low-priorityfragment (kl bits)

Unequal Diversity CodingA New Scheme for the Relay Channel

The Source encodes the packet with an (nS,k) code CS and broadcasts (Time Frame 1)(Time Frame 1)The Relay decodes and re-encodes nR additional parity bits with a code CR, out of kh<k information bits R h(Time Frame 2)

Ti F 1Time Frame 1:S encodes u (low+high priority fragments)f g )

Unequal Diversity CodingA New Scheme for the Relay Channel

The Source encodes the packet with an (nS,k) code CS and broadcasts (Time Frame 1)(Time Frame 1)The Relay decodes and re-encodes nR additional parity bits with a code CR, out of kh<k information bits R h(Time Frame 2)

Ti F 2Time Frame 2:R encodes uh(high priority fragment ONLY)f g )

Unequal Diversity CodingProtograph Design

Encoding at the Source with a high-rate LDPC codeEncoding at the Relay with a short LDPC codeSi l ll l t ti f t LDPC dSimple parallel concatenation of two LDPC codes

S DTime frame 1

R Time frame 2

Unequal Diversity CodingProtograph Design

Encoding at the Source with a high-rate LDPC codeEncoding at the Relay with a short LDPC codeSi l ll l t ti f t LDPC dSimple parallel concatenation of two LDPC codes

At the Destination, joint decoding th ll hover the overall graph

EXIT analysis: only two SNRsThe channel profile is a 2-D

tvector

Unequal Diversity CodingPerformance

Overall code rate: 7/10Block length k=1792 bits

Probability Distribution of the channel profile

Outage regionOutage region for high-priority fragments

Outage regions Outage / Block Error ProbabilityOutage regions Outage / Block Error Probability

Unequal Diversity CodingPerformance

Overall code rate: 7/10Block length k=1792 bits

Outage region f l i i ffor low-priority fragments

Outage regions Outage / Block Error ProbabilityOutage regions Outage / Block Error Probability

Unequal Diversity CodingPerformance

Overall code rate: 7/10Excellent match between EXIT analysisand simulations

Outage regions Outage / Block Error ProbabilityOutage regions Outage / Block Error Probability

Unequal Diversity CodingPerformance of AeroMACS

Bandwidth 5 MHz, 512 subcarriers, ∆f = 10.94 KHz, OFDMA b lOFDMA symbolsParking scenarioLack of diversity, low Rice y,factor (K = 0 dB, no/limited mobility, low Doppler)

The design may be tailored to the different PER requirements of the COCR messagesof the COCR messages

Unequal diversity is achieved also over the aeronautical channel modelmodel

SummaryUnequal Diversity Coding

A coded cooperation scheme targeting high code rates (R>1/2)Diversity / coding rate trade-off by introducing an Unequal Diversity distributed coding scheme

High priority fragments enjoy diversity, low priority fragments do notAccurate EXIT analysis for protograph LDPC codes over block fadingAccurate EXIT analysis for protograph LDPC codes over block fading channelsDesign of distributed protograph codes for Unequal Diversity achieving the target performancethe target performancePotentially suitable for in-airport communications to protect messages with different priority levels

Conclusions

We investigated the performance of the future system for the airport surface communications and we analyzed and proposed methods for improving its performanceimproving its performance.

We focused on techniques that increase the diversity of the system, and in particular space diversity (MIMO and cooperative communications) and time diversity (packet level coding)

Generally, all the methods investigated may be suitable for AeroMACS

PublicationsConference Proceedings

P. Pulini, G. Liva, and M. Chiani “Protograph EXIT Analysis over Block Fading Channels with Application to Relays,” ICC’12, June 2012, Ottawa, CAP Pulini and M Chiani “Improving the performance of AeroMACS by cooperativeP. Pulini, and M. Chiani, Improving the performance of AeroMACS by cooperative communications,” DASC’30th, October 2011, Seattle, USG. Liva, P. Pulini, and M. Chiani, “Flexible on-line construction of IRA codes for packet erasure correction with application to aeronautical communications,” ICC’11,packet erasure correction with application to aeronautical communications, ICC 11, June 2011, Kyoto, JapanP. Pulini, “Forward Link Performance Analysis for the Future IEEE 802.16-based Airport Data Link,” ICC’2010, May 2010, Cape Town, South AfricaP. Pulini, and M. Chiani, “Improving the forward link of the future airport data link by space-time coding,” InOWo’10, September 2010, Hamburg, GermanyS. Gligorevic, and P. Pulini, “Simplified airport surface channel model based on the WSSUS assumption,” ICNS’10, May 2010, Washington, USP. Pulini, and S. Gligorevic, “WiMAX performance in the airport environment,” MCSS’09, May 2009, Hersching, Germany

PublicationsSubmitted

Journals:G. Liva, P. Pulini and M. Chiani, “On-Line Construction of Irregular Repeat Accumulate Codes for Packet Erasure Channels” submitted to IEEEAccumulate Codes for Packet Erasure Channels , submitted to IEEE Transaction on wireless communicationsP. Pulini, G. Liva, and M. Chiani, “Unequal Diversity LDPC Codes for Relay Channels”, submitted to IEEE Transaction on communicationsChannels , submitted to IEEE Transaction on communications

Patents:G Liva P Pulini “Method for flexible transmission with LDPC codes”G. Liva, P. Pulini, Method for flexible transmission with LDPC codes . (Sub. January 2011)P. Pulini, G. Liva, “Method for relay transmission with UEP”. (Sub. May 2011)2011)P. Pulini, G. Liva, “Coded cooperation with information appending” (Sub. May 2011)

Seminars

“Future Airport Data Link Based on WiMAX – Forward Link Performance”, 22 October 2009, Oberpfaffenhoffen, Germany

“Improving the Performance of AeroMACS by Cooperative Communications”, 11 November 2011, Oberpfaffenhoffen, Germany

“Unequal Diversity Coding for the Relay Channel”, 14 December 2011, Oberpfaffenhoffen, Germany

Thanks for your attention!

Introduction and Motivations (4/4)Airport Surface Communications

WiMAX foresees a large number of profiles with different efficiency/robustness trade-offs

The most suitable profiles should be selected, taking into account the airport environment peculiarities

Analysis of the strengths/weaknesses of the selected profile may reveal the need for enhancements

Low-Density Parity-Check CodesProtograph Analysis over Block Fading Channels (3/4)

Given an SNR profile,

determine whether iterative decoding converges or not (outage)

Outage region of a protograph G,

SNR ° (0) SNR ° (1) SNR ° (2)= set of channel profiles for which

iterative decoding does not converge

SNR ° (0) SNR ° (1) SNR ° (2)

Low-Density Parity-Check CodesProtograph Analysis over Block Fading Channels (4/4)

Block error probability (outage probability) = probability of having a channel profile inchannel profile in

Probability Distribution of the channel profile

Th d fi iti f t i b SNR ° (0) SNR ° (1) SNR ° (2)The definition of outage region can be extended to each single variable node (= block)

SNR ° (0) SNR ° (1) SNR ° (2)

We can characterize the UEP of the protograph nodes!