Design and VLSI implementation of novel pre-screening and simplified sorting based K-best detection for MIMO systems

The K-Best algorithm for MIMO detections has been widely used in wireless communication systems. In this paper, to increase the throughput cost-effectively, two novel computational schemes are applied to design the K-Best detector. To reduce the number of visiting leaf nodes, we propose the pre-screening based enumeration, and almost half of leaf nodes are reduced for fast searches. To reduce the complexity of the sorting scheme, we apply the simplified Batcher's odd-even sorting algorithm (SBOESA) without sorting all outputs, and the required comparators are reduced through the simplified fast sorting design. The proposed 4×4 MIMO K-Best detector at 64QAM is implemented by TSMC 90nm process. The throughput of the proposed MIMO detector achieves up to 4.4Gbps. Compared with other designs, the proposed K-Best detector provides larger throughputs and performs superior hardware efficiency.