The Study Of Matched Field Localization Based On Compressive Sensing And Subspace Technology

Source localization by matched field is one of the classical problems in underwater acoustics, and it is also one of the means to locate underwater targets Conventional matched field processing suffers from low resolution and high side lobe level. The existing matched field methods with high resolution suffer from poor tolerance, and they also greatly depends on the number of snapshots and coherent sources can be resolved by them. In order to overcome these defects of the existing matched field processing methods, matched field processing methods based on compressive sensing and subspace technology are provided in this paper. By utilizing the space sparsity of the targets in the ocean waveguide, a sparse mathematical model is constructed, which is also solved by spasity reconstruction methods in compressive sensing and subspace technology, and the goal of locate underwater targets is achieved.The main contents of this paper includes:Firstly, the feasibility of achieving the goal of matched field positioning by using compressive sensing and subspace theory is studied. When reconstructing a sparse signal by using compressive sensing, the restricted isometry property must be satisfied. In the third chapter we will prove that this property is easily satisfied if the number of hydrophones is more than twice the number of targets. It is the same as the subspace method, which demand that the spark of the measurement matrix is more than some value, which is also easily satisfied.Secondly, matched field positioning based on sparsity reconstruction is studied. Prepositioning based on KMeans methods is provide in this paper. The Prepositioning process is as following:calculate the cost function value of each grid point by linear matched field processer, which has strong tolerance, low time complexity, and then achieve the goal of prepositioning by KMeans clustering method. When prepositioning is finished, the candidate position is used to position by the sparsity. reconstruction methods based on compressive sensing and subspace method, which leads to a reduced time complexity. At the same time the normalization of the green function vectors in the measurement matrix is necessary to reduce the impact of the range between the targets and the hydrophones in the matched field positioning.Thirdly, the noise reduction method based on singular value decomposition is studied, and its principle is also proved in this paper.Fourthly, the impact of rough seabed in matched field positioning is studied. The scattering field is calculated by perturbation method which is implemented by OASES-3D software developed by MIT.Fifthly, the impact of snapshot nubmer, signal to noise ratio, the number of hydrophones, coherent sources and environmental mismatch to the matched field method is studied.Sixthly, based on the sparsity mathematical model of wide band matched field positioning, SBOMP (Simultaneous Block Orthogonal Matching Pursuit) method is provided to solve the sparsity mathematical model, which achieve the goal of wideband matched field positioning.Lastly, in the field of sparsity reconstruction theory, we modified the existing sensing dictionary design method. The original sensing dictionary design method based on alternative projection is only applicable to real domain, and we generalized it to complex domain.