A novel Segregated K-Best detection algorithm for low complexity near optimal MIMO receivers

This paper addresses the problem of computational and hardware complexity in higher order constellation Multiple Input Multiple Output (MIMO) detectors. A novel Segregated K-Best detection (SKD) algorithm is proposed to serve as low complexity near optimal detector for MIMO receivers. The algorithm reduces detection complexity by choosing an optimal value of K depending upon the layer of traversal. This allows lesser number of nodes to be traversed leading to reduced complexity with only slight degradation in error rate performance. Hardware Descriptive Language (HDL) implementation of SKD is performed using 'one symbol detection per cycle architecture' and a compatible 'sequential linear sort' algorithm is proposed for sorting in SKD. The proposed design achieves a throughput as high as 802 Mbps with latency of 19.83 ns.