Iterative Multiuser Detection for Convolutionally Coded Asynchronous DSCDMA

Iterative Multiuser Detection for Convolutionally Coded Asynchronous DS-CDMA 9 th IEEE International Symposium on Personal, Indoor, and Mobile Radio Communications Boston, MA September 9, 1998 VIRGINIA POLYTECHNIC INSTITUTE & STATE UNIVERSITY MPRG MOBILE & PORTABLE RADIO RESEARCH GROUP Matthew Valenti and Brian D. Woerner Mobile and Portable Radio Research Group Virginia Tech Blacksburg, Virginia

Introduction n Performance of multiple access systems can be improved by multiuser detection (MUD). Introduction u u u F n Suboptimal approximations • Decorrelator, MMSE, DFE, PIC, SIC, etc. Most studies on MUD concentrate on the uncoded performance. u u 9/9/98 Verdu, Trans. Info. Theory ‘ 86. Implemented with Viterbi algorithm, complexity O(2 K). Optimal MUD is too complex for large K. u Here we consider the effects of coding. We propose a receiver structure that approximates joint MUD and FEC-decoding. The algorithm allows for asynchronous users and fading.

MUD for Coded DS-CDMA n Practical DS-CDMA systems use error correction coding (convolutional codes). u Introduction u u n If MUD and FEC are to be used, the interface should be improved. u 9/9/98 Soft-decision decoding outperforms hard-decision decoding (by about 2. 5 d. B). However, the optimal MUD passes hard-decisions to the channel decoder! Therefore it is possible for the coded performance of a system with MUD to be worse than the coded performance without the MUD. u The decoder for turbo codes gives insight on how to improve this interface. Use soft-decisions and feedback.

Relation to Other Work n T. Giallorenzi and S. Wilson u Optimal joint MUD/FEC-decoding Background F F F u Suboptimal approaches. F F F 9/9/98 Trans. Comm. Aug. 1996 Uses a “super-trellis”. High complexity: O(2 WK) Trans Comm. Sept. 1996 Separate MUD and Channel decoding. Soft values passed from MUD to channel decoder. No feedback used. See also P. Hoeher’s paper at ICUPC ‘ 93.

Relation to Other Work n M. Reed, C. Schlegel, et al u Feedback from FEC-decoder to MUD Background F u Synchronous DS-CDMA F u PIMRC ‘ 97 Close to single-user bound for K=5 users and spreading gain of N=7. F 9/9/98 Turbo Code Symp ‘ 97, ICUPC ‘ 97 Turbo codes F u “One-shot” detector. Convolutional codes F u Similar to the decoder for turbo codes. AWGN channel

Relation to Other Work n M. Moher u Background u Feedback from FEC-decoder to MUD. Multiuser systems with high signal correlation. F u u Random interleaving. Synchronous systems F u Comm. Letters, Aug. 1998 Close to single user bound for K=5, 10 and =0. 6, 0. 75 F F 9/9/98 Trans. Comm. , July 1998 Asynchronous systems F u FDMA with overlapping signals. K-symmetric channel. AWGN

Turbo Codes and Iterative Decoding Turbo Processing n 9/9/98 A turbo code is the parallel concatenation of two convolutional codes. u u An interleaver separates the code. Recursive Systematic Convolutional (RSC) codes are typically used. Dat a interleav er RSC Encod er #1 Output

Turbo Decoding Turbo Processing n A turbo decoder consists of two elementary decoders that work cooperatively. u Soft-in soft-out (SISO) decoders. F u Implemented with Log-MAP algorithm Feedback. F F Each decoder produces a posteriori information, which is used as a priori information by the other decoder. Iterative A priori probability A priori Received Data 9/9/98 SISO probability Decod SISO er Decoder #1 #2 Estimated Data

Serial Concatenated Codes Turbo Processing n The turbo decoder can also be used to decode serially concatenated codes. u Data Typically two convolutional codes. Outer Convolution al Encoder Inner SISO Decode r deinterleav er n(t) AWGN Turbo Decoder interleaver APP 9/9/98 interleaver Inner Convolution al Encoder Outer SISO Decoder Estimated Data

Turbo Equalization Turbo Processing n The “inner code” of a serial concatenation could be an Intersymbol Interference (ISI) channel. u Data ISI channel can be interpreted as a rate 1 code defined over the field of real numbers. (Outer) Convolution al Encoder ISI Channel Turbo Equalizer interleaver APP 9/9/98 interleaver SISO Equaliz er deinterleav er n(t) AWGN (Outer) SISO Decoder Estimated Data

Turbo Multiuser Detection n The “inner code” of a serial concatenation could be a MAI channel. Turbo MUD u 9/9/98 u u MAI channel can be thought of as a time varying ISI channel. MAI channel is a rate 1 code with time-varying coeficients over the field of real numbers. The input to the MAI channel consists of the encoded and interleaved sequences of all K users.

System Diagram Convolution al Encoder #1 “multiuser interleaver” interleaver #1 Turbo MUD MUX Convolution al Encoder #K n(t) AWGN interleaver #K Turbo MUD multiuser interleaver APP SISO MUD 9/9/98 MAI Channel multiuser deinterleav er Bank of K SISO Decoders Estimated Data

System Model MAI Channel Model n Received Signal: n Where: u u u n 9/9/98 ak is the signature waveform of user k. k is a random delay (i. e. asynchronous) of user k. Pk[i] is received power of user k’s ith bit (fading ampltiude). Matched Filter Output:

Optimal Multiuser Detection Algorithm: Setup Place y and b into vectors: n Place the fading amplitudes into a vector: n Compute cross-correlation matrix: MUD n 9/9/98

Optimal MUD: Execution n Run Viterbi algorithm with branch metric: where MUD u n Note that most derivations of the optimal MUD drop the p(b) term. u u n 9/9/98 Here we keep it. The channel decoder will provide this value. The algorithm produces hard bit decisions. u Not suitable for soft-decision channel decoding.

Soft-Output MUD n n Several algorithms can be used to produce softoutputs (preferably log-likelihood ratio). Trellis-based. u MAP algorithm MUD F F u SOVA algorithm F n Hagenauer & Hoeher, Globecom ‘ 89 Non-trellis-based. u u 9/9/98 Log-MAP, Robertson et al, ICC ‘ 95 OSOME, Hafeez & Stark, VTC ‘ 97 u Suboptimal, reduced complexity. Linear: decorrelator, MMSE. Subtractive (nonlinear): DFE, SIC, PIC.

Simulation Parameters n K=5 users u u Example u n Convolutional Code u u n 24 by 22 block interleaver (L=528). Log-MAP decoding. u 9/9/98 Constraint length 3. Rate 1/2. Interleaving u n Power controlled (same average power). N=7 (processing gain), code-on-pulse. Random spreading codes. u Both MUD and channel decoder. 3 iterations.

Simulation Results: AWGN Channel Introduction n After the second iteration, performance is close to single-user bound for BER greater than 10 -4. u u n Only a slight incremental gain by performing a third iteration. u 5/19/98 For BER less than 10 -4, the curves diverge. This behavior is similar to the “BER floor” in turbo codes. The extra processing for the third iteration is not worth it.

Simulation Results: Rayleigh Flat-Fading Channel n Fully-interleaved Rayleigh flat-fading. Introduction u n i. e. fades are independent from symbol to symbol. After second iteration, performance is close to the single-user bound. u u The curves do not diverge as they did for AWGN. Why? F F 5/19/98 The instantaneous received power is different for the different users. Therefore the MUD has one more parameter it can use to separate signals.

Conclusion n A strategy for iterative MUD/FEC-decoding is proposed. u Conclusion u F F n independently faded signals code and bit asynchronism. Proposed strategy was illustrated by simulation example. u u 9/9/98 Based on the concept of turbo processing. Similar to other researchers’ work, but the algorithm is generalized to allow: Significant performance gain by performing 2 iterations. When signals are independently faded, the algorithm exploits the differences in instantaneous signal power.

Future Work n The study assumes perfect channel estimates. u Conclusion u n The proposed strategy is still very complex u u n O(2 W+2 K) per iteration. Future work should consider the use of reduced complexity multiuser detectors. This structure could also be used for TDMA systems. u u u 9/9/98 The effect of channel estimation should be considered. The estimator could be incorporated into the feedback loop. u TDMA: only a few strong interferers, small K. Highly correlated signals, can take advantage of this system. Can use observations from multiple base stations. See our work at VTC, ICUPC, and Globecom CTMC.