A variable breadth based adaptive tree search algorithm for MIMO systems

Abstract

Near maximum-likelihood (ML) detections based on the tree search can approach the optimal performance with reduced complexity in Multiple-Input Multiple-Output (MIMO) systems. The breadth-first scheme is widely applied in practical systems for its stable and upper-bounded throughput. However, the major drawback of breadth-first detection is still the relatively high computational complexity. In this paper, we propose a variable breadth based adaptive tree search (VBA) scheme to further reduce the complexity. In particular, we introduce a variable metric constraint to dynamically regulate the searching breadth, which is determined by the accumulated metric of the partial zero-forcing (ZF) sequence at each layer of the searching tree during the adaptive candidate selection process. Simulation results and analysis show that the proposed algorithm can further reduce the detection complexity without degrading bit-error-rate (BER) performance.