适用于迫整接收机的排序格基规约算法  

On Sorted Lattice Reduction Algorithm for Integer Forcing Receiver

作　　者：陈志勇[1] 许小东[1] 戴旭初[1] 

机构地区：[1]中国科学技术大学电子工程与信息科学系,安徽合肥230027

出　　处：《信号处理》2014年第11期1292-1297,共6页Journal of Signal Processing

基　　金：国家自然科学基金(61271272);国家高技术研究发展计划(2012AA01A502)

摘　　要：针对单用户多输入多输出(MIMO)无线信号传输系统,本文采用一种低复杂度高性能的线性接收机结构—迫整接收机(Integer-Forcing),并在此基础上提出一种利用成对列变换依次减小格基的模长的格基规约(Lattice Reduction)算法—排序格基规约。迫整接收机利用整数域的闭环特性以及最大化每层信干噪比(SINR)的优化技术,可以用较低的复杂度获得较好的性能。鉴于最大化各层信干噪比等价于格理论中的最短独立向量组(SIVP)问题,本文提出的排序格基规约算法以较小的正交性能损失作为代价,有效的降低了解决该问题的复杂度,特别适合中大规模的MIMO系统。仿真实验和分析表明,不仅在格基模长性能上优于经典的LLL算法,而且具有更快的运算速度;同时,基于排序格基规约的迫整接收机显示出优于大部分线性接收机,且接近最大似然(ML)接收机的误比特率(BER)性能。Considering a single-user multiple-input multiple-output（MIMO） wireless system,this paper puts forward a novel sorted lattice reduction（SLR） algorithm for integer forcing（IF） linear receiver,which reduces the 2-norm of each lattice basis using a series of column-addition operations. The lower complexity and excellent performance of IF receivers are from the closed-loop property of integer domain and the optimization techniques of maximizing the signal-to-interferencenoise-ratio（SINR） of each layer,which is equivalent to a famous NP-Hard problem,referred to as shortest independent vector problem（SIVP）,in lattice theory. Thesorted lattice reduction algorithm in this paper can effectively alleviate the complexity of this problem in spite of a little orthogonality performance loss,which goes especially well with large MIMO systems. Computer simulations show that the proposed algorithm not only outperforms the famous LLL algorithm in terms of the length of the maximal lattice basis,but also has a relative lower practical complexity. Moreover,the integer forcing receiver based on the sorted lattice reduction algorithm defeats most linear receivers in terms of the bit-error-ratio（BER） performance and is closed to that of the optimal maximal likelihood（ML） receiver.

关 键 词：多输入多输出 迫整接收机 格基规约 误比特率性能 

分 类 号：TN929.5[电子电信—通信与信息系统]