Reduced complexity K-best sphere decoding algorithms for ill-conditioned MIMO channels

Abstract

The traditional K-best sphere decoder retains the best K-nodes at each level of the search tree; these K-nodes, include irrelevant nodes which increase the complexity without improving the performance. A variant of the K-best sphere decoding algorithm for ill-conditioned MIMO channels is proposed, namely, the ill-conditioned reduced complexity K-best algorithm (ill-RCKB). The ill-RCKB provides lower complexity than the traditional K-best algorithm without sacrificing its performance; this is achieved by discarding irrelevant nodes that have distance metrics greater than a pruned radius value, which depends on the channel condition number. A hybrid-RCKB decoder is also proposed in order to balance the performance and complexity in both well and ill-conditioned channels. Complexity analysis for the proposed algorithms is provided as well. Simulation results show that the ill-RCKB provides significant complexity reduction without compromising the performance.