Research On Robust DOA Estimation Algorithm Under The Condition Of Array Manifold Perturbation

With the rapid development of array signal processing in recent years,direction-of-arrival(DOA)estimation has become a research hotspot,and playes an increasingly vital role in many fields,such as wireless sensor array network,unmanned driving,electronic countermeasure and so on.It is difficult for classical subspace algorithms to achieve accurate DOA estimation in low signal-to-noise-ratio(SNR),coherent targets and other harsh environments.Subsequently,sparse signal recovery(SSR)algorithms came into being to provide a solution to the above problem.However,due to the imperfect calibration of the array,there are many factors between antennas that destroy the ideal array manifold structure,resulting in the estimation performance degradation or even failure of the algorithm.As a result,based on the background of array manifold perturbation,this paper mainly studies the robust DOA estimation algorithms for noncircular signals under the condition of mutual coupling error,and joint angle estimation and array calibration algorithm for monostatic MIMO radar under the condition of gain-phase error.Firstly,the array received signal model and mathematical theory foundation are given.At the same time,the principle and implementation steps of classical subspace DOA estimation algorithms are detailedly described.Then the sparse recovery principle based on compressed sensing(CS)theory and its representative algorithm are systematically introduced.Finally,the angle estimation performance of the above algorithms is compared and analyzed through simulation experiments and results.Secondly,the mutual coupling error is mainly considered based on the background of array manifold perturbation.Meanwihile,combining with the widespread existence and superiority of noncircular signals,two DOA estimation algorithms for noncircular signals are effectively proposed to slove the sparse recovery problem of noncircular signals under the condithion of mutual coupling error.At first,a noncircular block extended data model in the data domain is constructed based on noncircularity and the parametric decoupling idea,then a joint reweighted block sparse recovery algorithm based on l₁-SVD principle is proposed.Next different from the singular value decomposition technique in the data domain,the joint reweighted block sparse recovery method based on WSF principle is re-proposed via untlizing the eigenvalue decomposition technology in the covariance domain.In the end,a series of simulation experiments are carried out to demonstrate the robustness and superiority of the two proposed noncircular sparse direction finding algorithms under the condition of mutual coupling error.Eventually,the gain-phase error is mainly considered based on the background of array manifold perturbation,and a joint parameter estimation algorithm based on eigenspace is proposed to achieve angle estimation and array calibration of monostatic MIMO radar.Firstly,the signal model of monostatic MIMO radar is constructed under the condition of gain-phase error.Then the estimation of signal parameters via rotational invariance techniques(ESPRIT)principle is utlized to obtain the DOA rough estimation based on the decoupled signal subspace.Next the modified multiple signal classification(MUSIC)principle is adopted to achieve DOA precise estimation through a local spectrum peak searching based on the decoupled noise subspace.Finally,the gain-phase error is successfully estimated based on DOA accurate value.From the two perspectives of direction finding and gain-phase error calibration,the proposed algorithm not only outperforms the traditional subspace algorithms,especially for close-spaced targets,but also is suitable for non-uniform linear array(Non-ULA).More importantly,local searching greatly reduces the computational cost of the proposed algorithm.At last,the simualation comparison results of various algorithms are fully analyzed,and the robust advantage of the proposed algorithm gets strongly proved.