1. A Factor-Graph Approach to Joint OFDM Channel Estimation and Decoding in Impulsive Noise Channels Philip Schniter The Ohio State University Marcel Nassar, Brian L. Evans The University of Texas at Austin

2. Outline Uncoordinated interference in communication systems Effect of interference on OFDM systems Prior work on OFDM receivers in uncoordinated interference Message-passing OFDM receiver design Simulation results Introduction |Message Passing Receivers | Simulations | Summary 2

3. Uncoordinated Interference 3 Typical Scenarios: –Wireless Networks: Ad-hoc Networks, Platform Noise, non-communication sources –Powerline Communication Networks: Non-interoperable standards, electromagnetic emissions Statistical Models: Interference Model Two impulsive components: 7% of time/20dB above background 3% of time/30dB above background Gaussian Mixture (GM) Introduction |Message Passing Receivers | Simulations | Summary

4. OFDM Basics 4 System Diagram + Source channel DFT Inverse DFT Symbol Mapping 0111 … 1+i 1-i -1-i 1+i … noise + interference Noise Model where total noise background noise interference GM or GHMM and Receiver Model LDPC Coded After discarding the cyclic prefix: After applying DFT: Subchannels: Introduction |Message Passing Receivers | Simulations | Summary

5. OFDM Symbol Structure 5 Data Tones Symbols carry information Finite symbol constellation Adapt to channel conditions Pilot Tones Known symbol (p) Used to estimate channel pilots → linear channel estimation → symbol detection → decoding Null Tones Edge tones (spectral masking) Guard and low SNR tones Ignored in decoding Coding Added redundancy protects against errors But, there is unexploited information and dependencies Introduction |Message Passing Receivers | Simulations | Summary

6. Prior OFDM Designs 6 CategoryReferencesMethodLimitations Time-Domain preprocessing [Haring2001] Time-domain signal MMSE estimation ignore OFDM signal structure performance degrades with increasing SNR and modulation order [Zhidkov2008,T seng2012] Time-domain signal thresholding Sparse Signal Reconstruction [Caire2008,Lam pe2011] Compressed sensing utilize only known tones don’t use interference models complexity [Lin2011] Sparse Bayesian Learning Iterative Receivers [Mengi2010,Yih2 012] Iterative preprocessing & decoding Suffer from preprocessing limitations Ad-hoc design [Haring2004] Turbo-like receiver All don’t consider the non-linear channel estimation, and don’t use code structure Introduction |Message Passing Receivers | Simulations | Summary

7. Joint MAP-Decoding 7 The MAP decoding rule of LDPC coded OFDM is: Can be computed as follows: non iid & non-Gaussian depends on linearly-mixed N noise samples and L channel taps LDPC code Very high dimensional integrals and summations !! Introduction |Message Passing Receivers | Simulations | Summary

8. Belief Propagation on Factor Graphs 8 Graphical representation of pdf-factorization Two types of nodes: variable nodes denoted by circles factor nodes (squares): represent variable “dependence “ Consider the following pdf: Corresponding factor graph: Introduction |Message Passing Receivers | Simulations | Summary

9. Approximates MAP inference by exchanging messages on graph Factor message = factor’s belief about a variable’s p.d.f. Variable message = variable’s belief about its own p.d.f. Variable operation = multiply messages to update p.d.f. Factor operation = merges beliefs about variable and forwards Complexity = number of messages = node degrees Belief Propagation on Factor Graphs 9 Introduction |Message Passing Receivers | Simulations | Summary

10. Coded OFDM Factor Graph 10 unknown channel taps Unknown interference samples Information bits Coding & Interleaving Bit loading & modulation Symbols Received Symbols Introduction |Message Passing Receivers | Simulations | Summary

11. BP over OFDM Factor Graph 11 LDPC Decoding via BP [MacKay2003] MC Decoding Node degree=N+L!!! Introduction |Message Passing Receivers | Simulations | Summary

12. Generalized Approximate Message Passing [ Donoho2007,Rangan2010 ] 12 Estimation with Linear Mixing Decoupling via Graphs observations variables Generally a hard problem due to coupling Regression, compressed sensing, … OFDM systems: coupling Interference subgraph channel subgraph given and 3 types of output channels for each Introduction |Message Passing Receivers | Simulations | Summary

13. Message-Passing Receiver 13 Schedule 1.coded bits to symbols 2.symbols to 3.Run channel GAMP 4.Run noise “equalizer” 5. to symbols 6.Symbols to coded bits 7.Run LDPC decoding Turbo Iteration: 1.Run noise GAMP 2.MC Decoding 3.Repeat Equalizer Iteration: Initially uniform GAMP LDPC Dec. Introduction |Message Passing Receivers | Simulations | Summary

14. Receiver Design & Complexity 14 Not all samples required for sparse interference estimation Receiver can pick the subchannels: Information provided Complexity of MMSE estimation Selectively run subgraphs Monitor convergence (GAMP variances) Complexity and resources GAMP can be parallelized effectively OperationComplexity per iteration MC Decoding LDPC Decoding GAMP Design Freedom Notation : # tones : # coded bits : # check nodes : set of used tones Introduction |Message Passing Receivers | Simulations | Summary

15. Simulation - Uncoded Performance 15 Matched Filter Bound: Send only one symbol at tone k within 1dB of MF Bound 15db better than DFT 5 Taps GM noise 4-QAM N=256 15 pilots 80 nulls Settings use LMMSE channel estimate 2.5dB better than SBL use only known tones, requires matrix inverse performs well when interference dominates time- domain signal Introduction |Message Passing Receivers | Simulations | Summary

16. 16 Simulation - Coded Performance 10 Taps GM noise 16-QAM N=1024 150 pilots Rate ½ L=60k Settings one turbo iteration gives 9db over DFT 5 turbo iterations gives 13dB over DFT Integrating LDPC-BP into JCNED by passing back bit LLRs gives 1 dB improvement Introduction |Message Passing Receivers | Simulations | Summary

17. Summary 17 Huge performance gains if receiver account for uncoordinated interference The proposed solution combines all available information to perform approximate-MAP inference Asymptotic complexity similar to conventional OFDM receiver Can be parallelized Highly flexible framework: performance vs. complexity tradeoff Introduction |Message Passing Receivers | Simulations | Summary

18. Thank you 18

19. BACK UP 19

20. Interference in Communication Systems 20 Wireless LAN in ISM band Coexisting Protocols Non-Communication Sources Co-Channel interference Platform Powerline Communication Electromagnetic emissions Non-interoperable standards Introduction |Message Passing Receivers | Simulations | Robust Receivers| Summary

21. Empirical Modeling 21 WiFi Platform Interference Powerline Systems [Data provided by Intel] Gaussian HMM Model 12 1215 Partitioned Markov Chain Model [Zimmermann2002] 1 2 … Impulsive Background Introduction |Message Passing Receivers | Simulations | Robust Receivers| Summary

22. Statistical-Physical Modeling 22 Wireless Systems Powerline Systems Rural, Industrial, Apartments [Middleton77,Nassar11] Dense Urban, Commercial [Nassar11] WiFi, Ad-hoc [Middleton77, Gulati10] WiFi Hotspots [Gulati10] Gaussian Mixture (GM) Middleton Class-A Introduction |Message Passing Receivers | Simulations | Robust Receivers| Summary

23. Effect of Interference on OFDM 23 Single Carrier (SC)OFDM Impulse energy concentrated Symbol lost with high probability Symbol errors independent Min. distance decoding is MAP-optimal Minimum distance decoding under GM: Impulse energy spread out Symbol lost ?? Symbol errors dependent Disjoint minimum distance is sub-optimal Disjoint minimum distance decoding: Single Carrier vs. OFDM Impulse energy high → OFDM sym. lost Impulse energy low → OFDM sym. recovered [Haring2002] tens of dBs (symbol by symbol decoding) DFT Introduction |Message Passing Receivers | Simulations | Robust Receivers| Summary