A Fast ML Sphere Decoder with Multi-Layer Multi-Path Search

Spatial multiplexing is a key feature for upgrading transmission rate in recent wireless communication systems. However, the performance improvement comes at the cost of decoding complexity. Sphere decoders have shown the distinction to provide the maximum likelihood (ML) solution to this signal detection problem with much lower complexity than the exhaustive search. In this paper, the algorithm of a multi-layer multi-path sphere decoder is proposed. The common information of the same signal carried at different layers can be discovered. Moreover, multiple nodes are visited simultaneously to speed up the search process. Meanwhile, the completeness in the whole search space is still retained to derive the ML solution without conflicts among the different search paths. Simulation results show that more than 50% decoding cycles can be reduced and hence the average throughput can be increased.