Multiple-symbol M-sphere Detection based on Linear Radius for Differential Unitary Apace-time Modulation Systems

Abstract

Abstract In view of the problems of high computational complexity and difficulties in pipelining and parallel processing existed in sphere decoding (SD) algorithm for multiple-symbol differential detection (MSDD) system, the M-SD detection which combined M algorithm with SD is presented.. In M-SD, a new spherical radius scheme called linear radius is proposed to reduce the retained branches. It largely reduces the times of computing Frobenius norm square by a gradually decreasing radius with signal-to-noise ratio (SNR) increases. Based on the linear radius, M algorithm is combined with SD at low SNR to maintain a stable retained branch size and further reduce complexity. Simulation results show that M-SD detection with the constraint of linear radius significantly reduces the overall complexity compared with regular SD and its complexity is more stable. , Especially the complexity of M-SD can get more than 90% reduction when SNR £ 5dB, while the bit error rate (BER) performance is very close to SD within 0.1dB loss.