Log-domain iterative sphere decoder with symbol sorting

Abstract

In this paper the authors propose modified branch and pruning metrics for the sphere decoder to facilitate the use of apriori information in the sphere decoder. The proposed sphere decoder operates completely in the log-domain. Additionally the effect of order in which the symbols are decoded on the computational requirements of the decoder are investigated with the authors proposing a modification of the sorted QR decomposition (SQRD) algorithm to incorporate apriori information. The modified SQRD algorithm is shown to reduce the average number of computations by up to 95%. The apriori sphere decoder is tested in an iterative multiple input multiple output (MIMO) decoder and shown to reduce the bit error rate (BER) by an order of magnitude or provide approximately a one decibel improvement.