一种新的基于噪声预测的部分判决反馈MIMO接收算法

A Noise-Predictive Partial Decision-Feedback Detection for MIMO Systems

传统垂直一贝尔实验室分层空时结构（V-BLAST）系统采用全排序判决反馈（ODF）算法，能提高线性接收机性能，但也增加系统复杂度。该文在分段检测理论基础上，提出一种改进的采用噪声预测技术的部分判决反馈（PDF）接收算法。它基于MMSE准则，将接收到信号分为两段，只进行一次判决反馈操作，与传统重复排序判决反馈算法相比，大大降低了接收机计算复杂度。仿真结果表明SUI-3信道中天线数为2发4收，误码率为10^-4量级韵情况下，其性能与V-BLAST相比只下降了0．6dB左右，但要明显好于线性接收机以及其他改进型MMSE算法，有2．3dB左右的增益。

The traditional Vertical-Bell Laboratory layered Space-Time（V-BLAST） systems adopt an Ordered Decision-Feedback（ODF） detection strategy, which can significantly outperform the liner detector, however, it increases the computational complexity. Based on the conception of segmented detection, an improved noise-predictive Partial Decision-Feedback（PDF）detector is proposed, it introduces the MMSE principle and divides the receive signals into two segments. Since only feeding back the decision once, the MMSF-PDF detector reduces complexity significantly compared to the conventional ODF detector. The simulation results show that in the condition of SUI-3 channel model, 2 transmission antennas and 4 receive antennas, in the 10^-4 level, compared with the complicated V-BLAST system, the novel receiver only suffers 0.6dB loss, however, it has more or less 2-3 dB performance gains compared with the linear and other improved receiver based on the MMSE principle.