Turbo detection for MIMO systems: bit labeling and pre‐coding

Abstract

Abstract Bit‐interleaved coded modulation with iterative detection for multi‐input multi‐output (MIMO) systems demands an efficient APP detector that delivers soft information about the coded bits. The list‐sequential detector is a reduced‐complexity approximation to an exhaustive a posteriori probability (APP) detector. It uses a tree search employing the stack algorithm to find dominant terms of the log‐likelihood ratio (LLR). The commonly used symbol‐by‐symbol Gray labeling is not necessarily the best approach to good performance of iterative systems. Therefore, we consider different bit labelings and differential pre‐coding adapted to MIMO systems. With the help of extrinsic information transfer (EXIT) charts and bit error rate (BER) simulations, we show that symbol‐by‐symbol Gray labeling in combination with an outer Turbo code is a good compromise for fast convergence and low error floor of MIMO systems with iterative detection. Copyright © 2004 AEI