Analysis of CDMA multiuser demodulation based on statistical mechanics

点击翻译按钮获取中文摘要

We have evaluated analytically performance of CDMAmultiuser demodulation using the replica method. Under random spreading assumption and large-system limit, we have considered a quite generic case where the priors of the users' information symbols are arbitrary, and where the received signal follows a distribution conditional on the sum of the user's signal. The result has successfully extended that of Guo and verdu. In addition, their "decoupling principle," meaning that each user in CDMA communication can be regarded as occupying a single-user Gaussian channel, also holds under the generic condition, thereby demonstrating the power of the statistical-mechanics approach. The following problems have also been addressed : a coded CDMA system in which low-density parity-check (LDPC) codes are used as channel coding, effects of simultaneous estimation of information symbols and signal amplitudes in models with fading channels, and models for. frequency-hopping CDMA systems.We have also in … More vestigated the parallel interference cancellation scheme to the CDMA multiuser demodulation problem. In particular, on the basis of formal correspondence with neural associative memory models, we have applied the framework of statistical neurodynamics to study demodulation dynamics of parallel interference cancellation. We have also discussed analytical framework describing dynamics of the partial interference cancellation scheme proposed by Divsalar et al. Furthermore, we have elucidated relation between parallel interference cancellation and belief propagation, and have studied dynamics of the former on the basis of the density evolution, a framework of describing dynamics of the latter. The study has been applied to understanding linear parallel interference cancellation schemes, in particular their conditions of equivalence with one-shot algorithms such as the decorrelator and the linear minimum mean-squared-error (MMSE) demodulator. We have also demonstrated that our analytical framework is also useful in evaluating performance of stripping-based parallel interference cancellations when signal amplitudes differ from user to user. Less