基于对角加载奇异值分解的波束形成算法

在自适应波束形成算法中,QR分解具有很好的数值特征和固有的高度并行性。但当采样数较少,采样协方差矩阵估计值的噪声特征值分散会导致波束形成算法的性能下降问题,QR算法的性能就会下降。针对此缺陷,提出了对角加载奇异值(DSVD)分解的算法,该算法先对采样数据所构成的矩阵进行重构、分解、再重构、再分解,最后实现对角加载。通过仿真结果可以看到,DSVD算法不仅避免了对阵列协方差矩阵的估计和求逆,而且减少了估计运算量和估计误差,在复杂度与性能之间进行折衷。

Beamforming algorithm based on diagonal loading singular value decomposition

In the adaptive beamforming algorithm,QR decomposition algorithm has better numerical characteristics and inherently high degree of parallelism.But when the smaller sample number,the noise eigenvalue dispersion of the sample covariance matrix leads to the degradation of the beamforming algorithm performance,and QR algorithm performance will be decreased.According to this shortcoming,the diagonal loading singular value decomposition algorithm(DSVD) which is proposed in this paper,In this algorithm,the matrix which is composed of sampled data is firstly reconstruction,decomposition,then reconstruction,and then decomposition,and finally do the diagonal loading.DSVD algorithm avoid the estimation and inversion of the array covariance matrix,reducing the estimated computation and estimation error and improving numerical stability,can be obtained tradeoff between complexity and performance.