Combined Serially Concatenated Codes and MMSE Equalization: an EXIT Chart Aided Perspective

Download

PDF Download (167Kb)

Description/Abstract

Iterative Minimum Mean Square Error (MMSE) equalization and channel decoding is appealing, since it exhibits a lower complexity than maximum a posteriori (MAP) turbo equalization. However, the MMSE equalizer’s attainable performance is upper-bounded by that recorded for transmission over the coded AWGN channel, a phenomenon usually referred to as ”error shoulder”. In order to circumvent this performance limitation, we propose a three-stage concatenated transceiver constituted by inner and outer coding as well as MMSE equalization, which achieves significant iteration gains, despite using low-complexity serially concatenated memory-1 convolutional codes. It is demonstrated that this three-stage scheme outperforms the traditional two-stage MMSE turbo equalization scheme above a certain Eb/N0 threshold. Furthermore, the convergence behavior of the proposed scheme is analyzed with the aid of 3D EXtrinsic Information Transfer (EXIT) charts and their 2D projections, leading to a number of practical iterative receiver design guidelines.