针对MVDR波束形成器对导向矢量误差和干扰加噪声协方差矩阵误差敏感,容易引起空间谱估计性能下降甚至失效的问题,文中引入”畸变因子”对方向图畸变程度进行量化和归一化,在MVDR波束形成器基础上增加基于畸变因子的附加约束,并通过非线性最优化方法获得波束形成器的最优权矢量,进而实现稳健的MVDR空间谱估计。理论推导证明该方法本质上是将对角加载量进行自适应择优,不同于现有的其他方法,该方法不需要对导向矢量误差或采样误差做任何假设,具有很高的稳健性和实用性。

The MVDR method for spatial spectrum estimation is implemented by spatial scanning with MVDR beamformer. While the MVDR beamformer suffers from distortion of beam pattern when its mainlobe points to the signal source. As a result the desired signal is weaken and spatial spectrum estimation failed. Traditionally this problem can be relieved by diagonal loading directly on the sampling covariance matrix, yet the loading level is hard to select. In this paper, we first introduced a variable “distortion factor” to quantify and normalize the degree of the distortion, then constrain the weight vector based on the MVDR algorithm with it. The optimal weight vector can be obtained through method of nonlinear programming and used to perform robust spatial scanning. It′s proved theoretically that the diagonal loading level is optimized adaptively in this method. Furthermore, the proposed method doesn’t need any prior knowledge of the steering vector error or the sampling covariance matrix error, its robust performance is shown via simulated experiments.

TN957. 51