固定样本数目的QR分解递推算法  

Recursive algorithm based on QR decomposition for fixed sample size

作　　者：倪淑燕[1] 程乃平[1] 倪正中 

机构地区：[1]装备指挥技术学院,北京101416 [2]遥感信息研究所,北京100192

出　　处：《通信学报》2010年第S1期195-200,共6页Journal on Communications

摘　　要：自适应波束形成中基于QR分解的递推算法大都在样本不断累积下推出,即每次快拍增加一个样本。实际中可利用的样本不可能无限多,在到达一定样本数目后,每增加一个新样本的同时需要剔除一个旧样本。针对这种样本数目固定的数据更新方式,利用双曲Householder变换,提出了一种更实用的QR分解递推算法,并对其进行了简化,大大减小了运算量;之后利用逆QR分解的思想对其进行了进一步改进,使算法更利于系统的实时实现;在此基础上研究了更为稳健的对角加载逆QR分解的递推实现方式。计算机仿真证明,在有限样本情况下,本算法比常规QR分解算法具有更高的阵增益和更好的波束性能。Most of recursive algorithms based on QR decomposition(QRD) for adaptive beamforming are deduced by assuming that at each incoming snapshot an additional data vector is added.However,in practical use,the sample size which can beuse is always limited.After the sample data are cumulated to a certain amount,an old data vector should be removed at the same time when a new data vector is added.For this way of data updating,a more practical recursive QRD algorithm was proposed utilizing hyperbolic householder transformation.To reduce the computational cost of the algorithm a simplified method was also studied.Then based on the theory of the inverse QRD,further improvements were made for the algorithm more fit for real time use.Finally,the recursive algorithm was applied to the diagonal load-ing beamformer which had more robust capabilities.Several computer simulations illustrates that,when the sample size is limited,the proposed algorithms provide large array gain and better beam pattern than the traditional QRD algorithm.

关 键 词：阵列信号处理 波束形成 QR分解 有限快拍 

分 类 号：TN911.7[电子电信—通信与信息系统]