1. 1 UTILIZING CHANNEL CODING INFORMATION IN CIVA- BASED BLIND SEQUENCE DETECTORS Xiaohua(Edward) Li Department of Electrical and Computer Engineering State University of New York at Binghamton xli@binghamton.edu, http://ucesp.ee.binghamton.edu/~xli

2. 2 Contents I.Introduction II.System models III.Joint CIVA and channel decoding IV.Simulations V.Conclusions

3. 3 I Introduction Problems in blind equalization –Not robust to ill-conditioned channels –Local convergence, slow convergence (SISO methods) –Can not resolve common roots among sub-channels (SIMO methods)

4. 4 CIVA (channel independent Viterbi algorithm) [1]: effective blind equalizer –Robust to all channel conditions –Superior near MLSE performance –Fast convergence Propose: CIVA/decoder –Integrate convolutional decoder in CIVA –Use channel coding information to reduce complexity, save hardware

5. 5 II System Model

6. 6 Vector model

7. 7 III CIVA/Decoder Blind sequence detection

8. 8 Implement as trellis search –Metric updating rule –Example of trellis and probe

9. 9 CIVA/Decoder Properties –Have the advantages of both CIVA and channel decoder –Reduce complexity and hardware

10. 10 IV Simulations Convolutional encoder: rate ½ Channel length: 3 CIVA/Decoder: 32 states CIVA: 128 states CIVA/decoder performs better with reduced complexity

11. 11 Random 3-tap channels. BPSK. 400 samples. Compare: CIVA: blind CIVA/decoder MLSE: optimal VA: training MMSE: training PSP: with decision feedback

12. 12 V Conclusions CIVA/Decoder: new blind equalizer, integrate convolutional decoding in the channel- independent Viterbi algorithm –Superior near MLSE performance –Robust to even ill-conditioned channels –Reduced complexity and hardware Reference: –X. Li, “Blind sequence detection without channel estimation,” to appear in IEEE Trans. Signal Processing. A part appears in the 35 th Asilomar Conf. Signals, Syst., Comput., Oct 2001.