Underwater Target DOA Estimation Based On Sparse Covariance Matrix

The estimation of the target’s direction of arrival is an important prerequisite for positioning,navigation,interference,and imaging techniques.With the rapid development of Compressed Sensing theory,signal sparsity theory is applied to array direction finding.The sparse representation model based on covariance matrix can transform the array direction finding problem into the solution norm minimization problem.In recent years,it has received a lot of attention.This paper studies the representative Sparse Iterative Covariance-based Estimation SPICE.In this paper,the echo signal form and array model of the existing sonar detection target are summarized,and the classical spatial spectrum estimation method based on covariance matrix is introduced and theoretically derived.The effects of data length,angular interval and signal-to-noise ratio on conventional beam forming,minimum variance distortion-free response algorithm and multi-signal classification algorithm for direction of arrival estimation are analyzed.The results show that the larger the data length and angular interval,the higher the signal-to-noise ratio,the better the performance of the algorithm,and the multi-signal classification algorithm has the best performance.In this paper,the three core contents of the sparse representation,observation matrix and reconstruction algorithm of the Compressed Sensing theory are introduced.The greedy algorithm,convex relaxation algorithm and iterative hard threshold method are theoretically derived.The effects of sparsity and observed signal length on the signal reconstruction performance of orthogonal matching pursuit algorithm and base tracking denoising algorithm are analyzed.The simulation results show that the sparser the signal and the larger the length of the observed signal,the better the reconstruction effect.This paper focuses on the introduction and theoretical derivation of the SPICE method.SPICE has many advantages such as less prior knowledge required,but there are disadvantages such as a large amount of calculation.In this paper,a Redistribution Sparse Iterative Covariance-based Estimation RSPICE algorithm is proposed.The improved algorithm uses the sparseness of the signal in the spatial domain to transform the azimuth estimation problem of the signal into the angular vector sparse solution problem.Through the meshing process,the location of the angle estimation value and the update position of the power estimation value in the iterative process are avoided.The operation of the non-signal direction angle vector is carried out,and the direction of the incoming signal is accurately estimated by the spatial domain fine division,which reduces the operation of the signal direction related parameters in the iterative process,thereby improving the angular resolution and shortening the operation time.The paper analyzes the effects of angular interval,data length,signal-to-noise ratio,grid width and target number on the performance of RSPICE algorithm and compares it with SPICE algorithm and multiple signal classification algorithm.The simulation results show that the RSPICE algorithm has higher angular resolution,which can accurately estimate the signal azimuth when the signal-to-noise ratio is low and the data length is small,and is less affected by noise and other factors.The RSPICE method operation time increases as the grid width decreases and the number of sources increases,but it is still lower than the SPICE method.At the end of this paper,the lake test data is processed.The experimental data processing results show that the proposed RSPICE method has better angular resolution than the conventional beamforming method and SPICE method.Compared with the SPICE method,it has shorter calculations time and better real-time.