Study On Fast Robust Beamforming Algorithms

Fast robust adaptive beamforming is studied in this thesis. The effect of array errors on the performance of adaptive beamforming is analyzesd. And considering the practical demand of the computational complexity of beamformers, two fast robust adaptive beamformers are proposed, which are summarized as follows:1. Robust constrained LMS adaptive beamformer is proposed. The basic idea of the algorithm proposed is to search the optimal weight vector with recursive linearly constrained LMS updating, based on steering vector-expanded algorithm, jointly search the actual array steering vector of the desired signal. Both of the steering vector and the weight can converge to the optimal vectors. And the reference for choosing the initial values is indicated in the thesis. The algorithm enjoys lower computational complexity. Simulations under a few scenarios in the presence of steering error, array geometry error and sensor phase error demonstrate that this algorithm has excellent robust performance.2. A fast eigenspace-based (ESB) adaptive beamforming algorithm is proposed. ESB technology has robust performance in the presence of array errors. However it has the following problems. First, its computational complexity of the eigendecomposition of the covariance matrix is high. Second, under the nonstationary situation, we cannot get enough useful data snapshots, which may produce errors in the estimated signal-plus-interference subspace. The beamformer proposed in the thesis is accomplished by two steps. Firstly, the sensor array is partitioned into several subarrays without overlapped sensors, and then, the ESB adaptive beamforming of each subarray is performed. Secondly, the weight of the beamforming at subarray level is obtained by the projection of the steering vector at subarray level onto the signal-plus-interference subspace. As a reduced-rank technology, the proposed algorithm enjoys low computational complexity and fast convergence, and exhibits improved robust performance under convergence situation.