Channel-adaptive coding for coarsely quantized MIMO systems

Abstract

We present iterative algorithms which efficiently compute an optimized subset of equiprobable input symbols that achieves near-capacity on a large asymmetric discrete memoryless channel (DMC). The optimized subset defines a delay-free channel-adapted inner code which can be combined with a standard code. The introduction of a dense data structure renders it possible to apply channel-adaptive coding to 2-bit quantized MIMO systems, which exhibit a very large output set. The coding approach is shown to be robust to imperfect channel state information (CSI). Finally, we present a binary index switching algorithm that minimizes the BER by optimizing the mapping from code bits to input symbols.