MIMO系统的改进序贯蒙特卡罗迭代检测算法  

A Revised Sequential Monte Carlo Iterative Detection for MIMO System

作　　者：丁睿[1] 高西奇[1] 尤肖虎[1] 

机构地区：[1]东南大学移动通信国家重点实验室,南京210096

出　　处：《电子与信息学报》2010年第2期307-312,共6页Journal of Electronics & Information Technology

摘　　要：为了得到最优的MIMO迭代接收机,需要精确计算软输入软输出检测器输出的外信息,但精确计算的复杂度随调制阶数和天线数指数增长,不适合多天线高阶调制的情况。该文首先将外信息的估计归结为一个目标集合的选取,并提出通过序贯蒙特卡罗抽样方法获取目标集合。但是研究表明传统抽样方法不能有效获得合适的集合;因此一种改进的序贯蒙特卡罗抽样方法被提出,用于解决有限元离散概率空间的样本近似。最终,基于改进序贯蒙特卡罗抽样的外信息近似计算应用于迭代检测算法中。分析表明,该文提出的迭代检测算法的复杂度和抽取的样本数量呈线性比例;而仿真结果证明,较少的样本就可以取得逼近最优的误码率性能。An optimal iterative receiver for MIMO system need exact calculation of extrinsic information in Soft-Input-Soft-Output （SISO） detector. This optimal receiver does not fit the system with large numbers of antennas and high modulation order, because its complexity increases exponentially with modulation order and antenna number. So in this paper, the estimation of extrinsic information is proved to be equal to a choice issue of a target collection, which will be obtained by Sequential Monte Carlo （SMC） sampling. But the research also indicates that the traditional sampling method can not draw a suited target collection, so a Revised SMC （R-SMC） method is proposed to approximate a finite element discrete probability space by drawn samples. Finally, an approximate computation of extrinsic information based on R-SMC sampling is applied in this new detection algorithm. By analyses, the proposed algorithm＇s complexity is linearly proportional to the number of drawn samples. And simulation results prove that the near-optimal Bit-Error-Ratio （BER） performance can be obtained by a small number of samples.

关 键 词：MIMO系统 软输入软输出 序贯蒙特卡罗 迭代检测 

分 类 号：TN92[电子电信—通信与信息系统]