采用改进ML-OSIC的MIMO联合检测

MIMO Signal Detection by Combing Modified Layered OSIC and ML Detection

在垂直分层空时码(V-BLAST)多输入多输出(MIMO)系统中,最大似然检测算法(MLD)可以获得最佳的差错概率性能,但是其算法复杂度随着天线数目及其搜索空间数目呈指数增长;排序连续干扰消除(OSIC)算法具有高可行性的优点,但是其误码率由于无法防止误码扩散的原因而下降。基于OSIC算法复杂度主要来源于矩阵求逆运算,提出了一种改进的OSIC算法,该算法采用并行检测与精确检测值相结合的方案,在维持相对满意性能的情况下,减少算法复杂度。考虑到传统OSIC算法的误码扩散,将改进的OSIC算法与ML算法相结合,利用ML的穷尽搜索方式来提高算法的整体性能。为了避免ML算法给系统复杂度带来负担,通过一定的复杂度公式优先选出b层进行改进的OSIC算法,剩余的误差累积层使用MLD执行。仿真结果表明,所提算法复杂度小于传统OSIC算法,同时得到了接近ML算法的性能。

In V-BLAST MIMO system, maximum likelihood detector （MLD） can achieve the best error probability performance, while the complexity grows exponentially with the number of antennas and the size of modulation dimension. Ordered successive interference cancellation （ OSIC ） is proposed for the advantage of high feasibility, however, OSIC has poorer performance compared with MLD, for it is unable to prevent the bit error diffusion. As the complexity of OSIC algorithm mainly comes from the pseudo-inverse compu- tation,this paper proposes a modified layered OSIC method which adopts parallel detection and detection value refinement procedures. If little performance loss can be tolerated, the proposed algorithm will benefit from reducing complexity. Then the modified layered OSIC algorithm and MLD are combined to avoid the bit error diffusion, since exhaustive search in weak layers can improve the entire performance, b layers are chosen to be detected by the modified layered OSIC and MLD is used to detect the remaining error propagation layers. The simulation results indicate that the modified ML-OSIC algorithm gets almost the same per- formance as ML method, while the complexity is less than that of traditional one.