Research On Robust Adaptive Beamforming In The Presence Of Array Steering Vector Mismatch

In recent decades, the digital antenna array (DAA) technology has been widely used in military and defense industries, wireless communication, sonar and other areas. As the core of DAA technology, the adaptive beamforming has given a big push for the research of DAA. The adaptive beamformemer has much better resolution and interference rejection capability than data-independent beamformer, provided that the prior knowledge about the array steering vector (ASV) of the desired signal is accurate. Similar to the conventional phased array antenna, there are also various errors in DAA system, such as mutual coupling, gain and phase errors, element position errors and so on. However, the prior information on the ASV of desired signal is often imprecise due to the errors. Unfortunately, the traditional adaptive beamformers are sensitive to the mismatch between the presumed ASV and actual ASV. Consequently, many researchers pay attention to the improvement on robustness of adaptive beamformer to ASV mismatch. Based on the previous work, this dissertation proposes some robust adaptive beamforming (RAB) algorithms with higher ability to tackle with errors. Then, the main work is summarized as follows.Firstly, this dissertation designs the robust beamformer by minimizing the error sensitivity (MS-RAB) under the ellipsoid uncertainty set. On the premise of ASV mismatch level, the upper-bound on norm of the ASV error in MS-RAB is not critical for the output performance. Theoretically, the MS-RAB is also belonging to the RAB based on ASV uncertainty set (RAB_un). Similarly, the output SINR of the MS-RAB suffers deterioration in the presence of large ASV mismatch.To improve the output performance of traditional RAB_un methods, this dissertation proposes several estimating equations of ASV mismatch level, which are derived from on the eigen-subspace projection theory. Based upon these equations, the iterative robust adaptive beamformer with adjustable error radius (AR-RCB) is developed. In the AR-RCB, the iteration is started with estimation of error radius, which is the optimal solution of the estimating equation. Then, the estimated error radius is used by the robust capon beamformer (RCB) to calculate the corresponding optimal ASV. The experimental results indicate that the AR-RCB achieves higher output SINR than common RAB_un approaches in the condition of large ASV mismatch. In a similar way, this methodology can be extended to deal with multiple errors using adaptively updated ellipsoid sets. Although the iterative beamfomers proposed above maintain better output performance, but much more computation source is consumed because of the iterations. To make a balance between the output SINR and the processing complexity, a new RAB is designed to iteratively search the ASV of desired signal on the basis of the Taylor series expansion. In this beamformer, the ASV uncertainty set is unnecessary, and then the amount of calculation is reduced without SINR degradation.At last, considering the symmetry of mutual coupling between the neighboring elements in the uniform rectangular array (URA), this dissertation introduces a robust adaptive beamformer with low coupling sensitivity. Through setting a plurality of auxiliary elements around the URA, the performance of adaptive beamformer can be preserved without construction of ASV uncertainty set. However, there was another important problem that is the auxiliary elements in URA may alleviate the mutual coupling sensitivity of adaptive beamformer along with the blind angle in the process of spatial spectrum estimation. This challenge can be easily overcome with the aid of circular spatial spectrum estimation using noise subspaces with different dimensions.