1. 1 Channel Equalization for STBC- Encoded Cooperative Transmissions with Asynchronous Transmitters Xiaohua (Edward) Li, Fan Ng, Juite Hwu, Mo Chen Department of Electrical and Computer Engineering State University of New York at Binghamton {xli, fanng1,jhuw1,mchen0}@binghamton.edu http://ucesp.ws.binghamton.edu/~xli

2. 2 Summary Equalization for STBC-encoded cooperative transmissions –Asynchronous transmitters create ISI even in flat- fading environment –ISI channels adjustable by receiver –Viterbi equalizer for near-optimal performance –Efficient linear-prediction-based equalizer –Performance of cooperative transmission studied by simulations

3. 3 Contents 1.Introduction 2.Cooperative transmissions with asynchronous transmitters 3.Viterbi equalizer 4.Linear equalizers: linear prediction 5.Simulations 6.Conclusions

4. 4 1.Introduction Cooperative transmissions –Use STBC for diversity, power efficiency Challenges: –Imperfect synchronization among transmitters: conventional STBC receiver not applicable –Performance degradation: compromise advantage of cooperative transmissions Objectives: –New receiver equalization techniques –Performance comparison: asynchronous cooperative, or non-cooperative transmissions

5. 5 2. Cooperative transmissions with asynchronous transmitters Assume –Transmit nodes 1 to J transmit symbols {s(n)} with STBC –No perfect synchronization in time (local clock, transmission delay, propagation delay) –Frequency synchronization not addressed, dealt with by adaptive equalizer

6. 6 Channel model (J transmitter, a single receiver, flat fading)

7. 7 3. Viterbi equalizer Consider J=2 and Alamouti STBC for simplicity Receiver adjust δ: short channel, strong h 2 (0) Even delay d between transmitters Odd delay d

8. 8 Channel model with uncoded symbols –Even delay d. Even/odd samples are –Odd delay d, similarly available Viterbi equalizer available –Complexity: With decision feedback: Complexity reduced by adjusting δ

9. 9 4. Linear-prediction-based equalizer Choose proper δ to make h 1 (0) dominating Construct vector model –Special structure: H has dominating diagonal, Good for linear prediction –Example:

10. 10 Linear prediction: –Proposition:

11. 11 Properties –Symbols estimated from linear prediction error y(2n) and y(2n+1) –Efficient adaptive implementation: complexity O(N), track residue carrier induced time- variation –Robust: most ill channel conditions avoided by selecting proper δ

12. 12 5.Simulations Color codes: Convention STBC decoder used in asynchronous coop transmission. Non-cooperative transmission, flat fading channel Proposed Viterbi equalizer with asynchronous coop transmission Optimal STBC with perfect synchronization QPSK, J=2. d=1. VA has 128 states.

13. 13 Viterbi equalizer with decision feedback. 4 trellis states. VA with DF, delay d=10. VA with DF, delay d=2. VA without DF, delay d=2.

14. 14 Conventional STBC decoder used in asynchronous coop transmission MMSE equalizer used in asynchronous coop transmission Non-cooperative transmission, dispersive channel Proposed linear-prediction-based equalizer Conventional STBC with dispersive channel Linear equalizers: QPSK. d=10. Equalizer length N=20.

15. 15 6.Conclusions Equalizers for STBC cooperative transmissions when transmitters are not synchronized –Viterbi equalizer: performance near conventional STBC, high complexity –Viterbi equalizer with feedback: slight performance loss, extremely reduced complexity –Linear prediction-based equalizer: linear complexity, performance better than non-cooperation, much worse than conventional STBC (all in dispersive channel)