| Adaptive beamforming is an important research area in the field of array signal processing, which utilizes multi-sensor array to build a signal processing system for transmission or reception of the space signals, and it is widely used in radar, sonar, communication system, smart appliance and smart conference system. However, the performance of adaptive beamforming algorithm depends on the assumptions of incoming signal, array and the environment; and it is very sensitive to the accuracy of the assumption. In practical engineering applictions, the error is always exists, so the research of robust beamforming algorithm is necessary.Currently, popular robust algorithms are mainly divided into three categories: feature subspace, diagonal loading and convex optimization. The first category of algorithms has a serious loss in low SNR (Signal to noise ratio), and it needs to know the number of sources; for the second one, the loading factor and the upper and lower threshold of the actual error have no reliable method to determine; the algorithm of the last category is studied most in recent years, which improves the performance of beamforming greatly, but it has some shortcomings, for example, it is sensitive to the errors of element position and multi-path. This paper focuses on the category of feature subspace and convex opitimazition algorithm. In view of the existing error in the practical application we propose three robust beamforming algorithms.The main content and innovation points are as follows:1. For the output SINR loss problem of the multi-constrained linearly constrained minimum variance algorithm, we proposed a robust linearly constrained minimum variance beamforming based on complex constraints. In the algorithm, we use complex constraints, which are variables, the optimal constraint values can be solved by using the criterion of maximizing signal to noise ratio, and the detailed derivation of the algorithm is given in the paper. The computer simulation results show the effectiveness of our proposed algorithm, the algorithm has a good output performance and has an equal computational complexity compared to the conventional beamforming algorithm.2. The estimation of covariance matrix is one of the key issues to be resolved in beamforming. Under the condition of small snapshots, the estimation of covariance is poor, to deal with this problem we propose a knowledge aided robust beamforming. Firstly, the algorithm obtains the covariance matrix which does not contain the desired signal by the method of the reconstruction method, and then uses the sampling data covariance matrix to estimate the optimal covariance matrix jointly. Comparative experiments show that the proposed algorithm has better performance under scenario of small snapshots.3. Currently, some reasearchers have proposed an iterative robust beamforming algorithm, however, when the interference to noise ratio of interference is larger than the input SNR of the desired signal, the algorithm may converges to the direction of the interference. In the paper we proposed an algorithm by using a new signal subspace method, which does not need to know the number of soures, and can work in low SNR. This method can be used to obtain the projection matrix of interferences plus noise, by using the property that the desired signal is orthogonal to the interference plus noise subspace, we can avoid the algorithm converging to the direction of interference. In the iterative process, by using a relaxed constraint to ensure that the desired signal is in the constrainted subspace, we can get the optimal solution. Finally, the simulation experiment demonstrates the effectiveness of the proposed algorithm, and the algorithm is analyzed and summarized. |
