MIMO系统中基于条件数阈值的格基约减信号检测算法

Lattice reduction aided signal detection algorithm based on condition number threshold in MIMO systems

在多输入多输出系统中,最大似然(maximum likelihood,ML)检测算法性能最优但复杂度很高,最小均方误差(minimum mean-square error,MMSE)检测算法复杂度低但其性能较差。较高的信道矩阵条件数会给信号检测算法的误码率性能带来不利影响。针对这些问题,提出一种基于信道矩阵条件数阈值的信号检测算法来提升高条件数下传统检测算法的性能。该算法通过比较信道矩阵的条件数与预先设定的条件数阈值,选取相应的检测算法:当条件数低于阈值时,采用复杂度较低的LLL(lenstra-lenstra-lovasz)约减的MMSE(LLL-MMSE)算法来减少计算量;当条件数高于阈值时,采用基于排序分组的ML与LLL-MMSE联合的检测算法,通过增加一定的计算量来保证检测性能。对不同阈值下的误码率性能进行了仿真,结果表明算法的性能明显优于传统的LLL-MMSE检测算法,并且通过预先设定条件数阈值可以使得算法在性能和复杂度之间获得较好的平衡,最终达到优化检测算法性能的目的。

In multiple-input multiple-output(MIMO) systems,the ML detection algorithm has optimal performance but has high complexity,and the MMSE detection algorithm has lower complexity but its performance is poor.The higher condition number of channel matrix has bad effects on the detection performance of signals.Aimed at these problems,this paper provides a signal detection algorithm based on the condition number threshold of channel matrix,which can enhance the performance of traditional detection algorithm when the condition number is high.The algorithm selects the corresponding detection algorithm by comparing the condition number of channel matrix with the preset threshold: When the condition number is lower than the threshold,the algorithm chooses the LLL-MMSE algorithm which has lower complexity to reduce computational load; when the condition number is higher than the threshold,it chooses the sorted and grouped detection algorithm which combines ML and LLL-MMSE,ensuring the detection performance by increasing a certain amount of computational load.Through the simulation of the BER performance of the algorithm in different thresholds,results show that the proposed algorithm can outperform the traditional LLL-MMSE detection algorithm obviously.By presetting the threshold ofthe condition number,the algorithm can achieve a better balance between performance and complexity,and finally realize the optimizing of the performance of detection algorithm.