Reconfigurable K-best algorithm for MIMO detection systems

The maximum likelihood (ML) detection is the optimal detection method for multiple-input multiple-output (MIMO) communication systems. The normal K-best sphere decoding algorithm (SDA) can guarantee a fixed throughput, but it induces a large bit error rate (BER) degradation. In order to achieve close-to-ML performance, the K value needs to be sufficiently large. Thus, it needs large computation with long latency and low throughput. In this paper, a reconfigurable K-best SDA is proposed by using fixed K values for different layers of K-best detection, which can increase the MIMO detection performance. The architecture of 4×4 MIMO detector is designed for both 16QAM and 64QAM modulation.