The Research On Key Problems Of Array Signal Processing

Adaptive beamforming and high resolution direction of arrival(DOA) estimation are two important research directions of array signal processing, which have important military and civilian applications in communication, sonar, radar, radio, seismic exploration and other fields. This dissertaion systematically and deeply carry out theorectical and experimental studies on adaptive beamforming and estimate method of DOA under small samples. The main topics of this dissertation are listed as follows:1. The small samples and the channel random response errors will deteriorate the performance of the linearly constrained minimum variance(LCMV) beamforming, a new method based on iterative subspace tracking and structural constraint is presented. The approach is performed in two stages. First, Clearing operation at training data set was used to calculate each principal eigenvector in sequentially, then, the adaptive beamforming algorithm based on subspace projection can be obtained by constraining its weight vector to a specific form. The high output signal to interference plus noise ration of subspace processing is remained and the robustness against small sample support and in the presence of channel mismatch is improved. Aiming at the problem of signal subspace dimension determining in small sample, the beamforming algorithm based on iterative adaptive weighted fusion sample covariance matrix and the prior covariance matrix is proposed. In the estimation of covariance matrix, weighted coefficient is calculated based on the minimum mean square error criterion, and the priori covariance matrix is updated using the iterative adaptive approach. The proposed method can alleviate effectively the prior knowledge and the current data model mismatch problem, improve the estimation accuracy of the covariance matrix in the small sample and avoid the determination of the subspace dimension.2. Considering the issue that full-dimensional adaptive beamforming usually has very high computation cost and need the large number of sampling data which is independent and identically distributed, a new adaptive beamforming algorithm based on subspace reconstructing and cross terms adaptive removing is proposed. The proposed approach is illuminated by the multi-dimensional array data having the characteristics of reconstruction of fractal-dimension. Firstly, the fractal-dimensional signal subspace of the array data is estimated using training samples; Then, the full-dimensional signal subspace is reconstructed by adopting the tensor product operation and the cross terms is removed adaptively; Finally, the beamforming weight vector is deduced by the subspace projection algorithm. Theoretical analysis and simulation results show that the method has lower computational complexity and can effectively improve the output signal-to-interference plus noise ration(SINR) with the small sample support. In order to avoid the problem of subspace dimension determination in the reducing the computation complexity, a knowledge-aided reduced-rank beamforming algorithm based on joint iterative optimization is presented. The method firstly realize adaptive dimension reduction by joint optimization for reduced-rank matrix and beamforming weighted vector; and then online update the priori covariance matrix of array data using the spatial spectra reconstruction technology; finally, do weighted processing for the update covariance matrix and sample covariance matrix to improve the estimation accuracy of the array covariance matrix. The experimental results show that the algorithm can obtain a faster convergence rate and has robustness for the signal subspace dimension by the weighted fusion processing and joint iterative optimization of the reduced dimension matrix and weight vector on the basis of online updating for the prior covariance matrix.3. The performance of DOA will deteriorate under the small sample and adjacent strong signal and weak signal, a modified MUSIC approach with weighted pseudo-noise subspace projection and a modified Capon approach with adaptive weighted for discriminating strong signal and weak signal are proposed. The different weighting factors are applied to traditional MUSIC algorithm and Capon spectrum estimation method, the more sharp spatial spectrum can be geted, which will help to improve the accuracy of DOA estimation of the adjacent signal. The high-resolution of subspace processing is remained and the robustness against small sample support and in the presence of strong signal and weak signal is improved. The theoretical analysis and simulation experiments show that the proposed method can effectively increase the resolution probability of adjacent strong signal and weak signal in conditions of small samples and low SNR.4. Aiming at the problem of coherent signal subspace method(CSSM) depending on the estimated accuracy of signal subspace in estimation of direction of arrival(DOA) for wideband signal, a new estimated method of DOA for wideband signal based on iterative adaptive spectral reconstruction is proposed. Firstly, the wideband signal is divided into several narrowband signals by discrete Fourier transformation(DFT); Then, the signal matched power spectral in referenced sub-band is computed, which can form the initial data self-correlative matrix; Finally, the linear restrained minimum variance spectral(Capon spectral) of signals in other sub-bands are reconstructed using sequential iterative means, the DOA can be estimated by locations of spectral peaks. The proposed method can realize the fully dimensional focusing of different sub-band data and does not need to accurate estimate the signal subspace under the small sample and low signal to noise ratio(SNR), and it also does not need to estimate pre-estimate angle, which can significantly improve the accuracy of DOA estimation. The simulation experiments verify the effectiveness of the proposed method.