Research On Subspace Estimation And Beamforming Of Spatial Weak Targets

In the process of deep space exploration of a radio telescope,the beamforming performance of phased array feed is affected by steering vector uncertainty and strong interferences.To solve this problem,the robust adaptive beamforming algorithm is thoroughly studied and a recursive Bayesian beamformer based on subspace projection is proposed in this paper.Meanwhile,in order to realize the accurate construction of vertical projection matrix of signal subspace,the methods of subspace estimation are extensively explored and two Bayesian subspace estimation algorithms are proposed.The main research work and innovation points of this paper are summarized as follows:(1)With a simple derivation of the signal and array model,the hypothesis and statistical characteristics of narrowband signal processing in a uniform linear array are given.After that,basic concepts of subspace and fundamental principles of beamforming are both introduced.At the same time,the performance of adaptive beamforming algorithms based on different optimization criteria is analyzed.(2)The robust adaptive beamforming algorithm is thoroughly investigated and a recursive Bayesian beamformer based on subspace projection is proposed.Subspace projection can effectively eliminates the strong interferences,while the recursive Bayesian beamformer shows good robustness to uncertain steering vectors.The results of numerical simulations show that the output performance of the proposed beamformer is similar to optimal Max-SINR algorithm,and the proposed beamformer can work well even at very low signal-to-noise ratio(SNR),which can effectively solve the problem of steering vector uncertainty and strong interferences during the beamforming process.(3)In order to accomplish the accurate construction of vertical projection matrix of interference subspace,the fast approximated power iteration(FAPI)subspace estimation algorithm is reserached.FAPI can accurately track a set of arbitrary orthogonal basis of the primary subspace,which is the optimal one of the orthogonal iterative subspace estimation algorithms.The results of simulations show that,compared with other typical fast subspace algorithms,FAPI has higher estimation accuracy and better orthogonal performance.(4)Aiming to solve the problem that it is difficult to estimate the signal subspace accurately under a very low SNR,two Bayesian methods of signal subspace estimation are presented.Firstly,a Bayesian subspace method based on signal deflation is proposed to accurately estimate the steering vectors of multiple signals at very low SNR.Then a Bayesian subspace method based on combination of multiple signals is proposed,which improves the cumbersome deflation process and realizes orthogonal iteration of the subspace tracking.The results of numerical simulations show that,compared with eigenvalue decomposition and FAPI,the proposed two Bayesian methods have better subspace estimation performance,especially when weak signals are considered.