一种分步ML的VBLAST检测算法  

A split-step ML VBLAST detection algorithm

作　　者：吴君钦[1] 金巧[1] 吴云龙[1] 

机构地区：[1]江西理工大学信息工程学院,江西赣州341000

出　　处：《信息技术》2015年第3期17-20,24,共5页Information Technology

基　　金：国家自然科学基金项目(11062002)

摘　　要：针对在VBLAST架构下的现有算法不能很好地平衡系统的检测性能和检测算法复杂度,提出一种基于最大似然的分步检测算法。该算法将串行干扰消除算法和QR分解算法结合在一起,运用了有限步的最大似然检测,最大限度地提高了每步检测的信号的性能,减少了误码传播,与传统的ML算法相比,其复杂度大大降低。仿真结果表明,分步ML算法比QR算法和迫零算法在误码性能上要好很多,尤其在多天线和高信噪比的情况下性能胜出更为明显。In VBLAST system,the existing algorithms can not balance detection performance and detection algorithm complexity very well. In this paper,a split-step detection algorithm based on maximum likelihood was proposed. The algorithm combined successive interference cancellation algorithm and QR decomposition algorithm together. It used finite steps of maximum likelihood detection,and improved the performance of each detection signal which reduced the error propagation. The complexity of this algorithm is greatly reduced as against the traditional ML algorithm. The proposed algorithm was simulated and compared with the conventional detection algorithms. The simulation results showed that the proposed split-step ML algorithm was superior to QR algorithm and zero forcing algorithm on the BER performance especially in the multiple antennas and under the condition of high SNR.

关 键 词：VBLAST 复杂度 QR分解 串行干扰消除 最大似然 

分 类 号：TN911.23[电子电信—通信与信息系统]