几种MIMO最大似然检测算法性能与复杂度比较及改进  

Performance, Complexity comparison and improvement of several ML detectors for MIMO system

作　　者：孙艳华[1] 张延华[1] 龚萍[2] 吴伟陵[2] 

机构地区：[1]北京工业大学,北京100022 [2]北京邮电大学,北京100876

出　　处：《电路与系统学报》2008年第3期93-99,共7页Journal of Circuits and Systems

基　　金：863项目:B3G-TDD迭代接收机设计(2001AA123016)

摘　　要：最大似然检测在误比特率最小的意义下是最优接收,但是其复杂度不可实现。本文介绍了半定松弛、分枝定界和堆栈三种低复杂度最大似然检测算法,并对其性能和复杂度进行了仿真分析,提出了改进的分枝定界和堆栈算法,仿真结果证明分枝定界和堆栈算法性能要优于半定松弛算法,分枝定界算法的复杂度低于堆栈算法且半定松弛算法以多项式复杂度取得了逼近最大似然的性能,同时改进算法加快了算法收敛速度,降低了计算复杂度和对存储空间的要求。Although ML detector is optimal in the sense of minimization of bit error rate, it is not practical for its complexity. Three ML detection algorithms are presented in this paper which are the SemiDefinite Programming algorithm （SDP）, the Branch and Bound algorithm （BB） and the Stack algorithm. We compare the performance and the complexity of these algorithms, and propose the improved BB and Stack algorithms. Simulations show that the BB algorithm and Stack algorithms outperform the SDP algorithm and the BB algorithm has a lower complexity compared to the Stack algorithm. It also proves that the improved algorithms accelerate the rate of convergence to ML solution and decrease the complexity and the required storage of algorithm.

关 键 词：MIMO 最大似然检测 半定松弛 分枝定界 

分 类 号：TN911.23[电子电信—通信与信息系统]