A Comparison of Iterative Demapping with Iterative Soft Decision Interference Cancellation.pdf

A Comparison of Iterative Demapping with Iterative Soft Decision Interference Cancellation Employing the Distribution of Interference Ju?rgen F. Ro??ler and Johannes B. Huber Institute for Information Transmission, University of Erlangen–Nuremberg Cauerstra?e 7/LIT, 91058 Erlangen, Germany Abstract— A well-known receiver strategy is iterative soft decision interference cancellation (ISDIC) which can be used for CDMA (code–division multiple–access) transmission as well as PAM (pulse amplitude modulation) transmission over dispersive channels. ISDIC can also be adopted with a decoding unit in the loop yielding a turbo multiuser detector or a turbo channel equalizer. The distribution of residual interference after cancellation is commonly assumed to be Gaussian. Recently, an analysis of matched-filter based iterative soft decision in- terference cancellation (MF ISDIC) when utilizing the actual probability density function (pdf) of residual interference has been given. Another well-known receiver strategy is iterative demapping of non-Gray mapped symbols. In this paper we analytically compare both receiver strategies. It first turns out, that the cancellation step in the MF ISDIC receiver is redundant when considering the actual pdf of residual interference for calculation of improved a posteriori probabilities of symbols. Secondly, it can be shown that MF ISDIC employing the actual pdf of interference and iterative demapping are equiv- alent as the calculations needed for computation of improved a posteriori information supplied to a channel decoder in each iteration have the same structure. Additionally, it is shown ,that both schemes coincide exactly if the mapping can be written as a weighted superposition of interfering bits. I. INTRODUCTION The demands on the data rate provided by future mo- bile communications systems will further increase. Provision of higher data rates requires e.g. higher order modulation schemes and more efficient receiver algorithms to over