中文摘要：

针对MIMO通信系统，提出一种非线性译码算法，并根据Fisher准则选择最有利于信号分离的空间并降低信号维数，以克服通常非线性方法计算量较大的不足。另外，由于利用接收信号的高阶统计特性，其性能优于常规线性最小均方和排序连续干扰抵消算法。计算机仿真表明，即使在各接收通道存在一定增益不一致的情况下，所提出算法的性能也较常规算法有明显提高。

英文摘要：

A nonlinear decoding method was proposed for MIMO communication systems, and dimension of the matrix and computation complexity were reduced significantly by selecting discriminate space based on Fisher ratio for signal separation. Performance improvement was obtained by utilizing high-order statistic characteristic of the received signals. Compared with the conventional linear detection algorithms, MMSE and OSIC, simulation results show that the nonlinear algorithm reduces largely bit error rate even though the responses of receiver antennas are not identical to each other.