Research On The Method Of DOA Estimation Based On Compressed Sensing Theory

Underwater array signal processing as an important part of modern signal processing,is widely used in underwater communication,sonar detection,ship reconnaissance,ship navigation,underwater topography detection,underwater image processing and many other fields in recent decades.Compressed sensing?CS?theory as a new signal processing theory,has gradually applied to the underwater array signal processing,and forms a kind of DOA estimation algorithm.This paper first establishes the spatial spectrum estimation model based on the CS theory,including the establishment of spatial spectrum model,the analysis of the CS theory and the applicability of CS theory in spatial spectrum estimation.Then this paper discusses three kinds of typical algorithms:₁l-SVD algorithm,Orthogonal Matching Pursuit?OMP?algorithm,Sparse Bayesian Learning?SBL?algorithm,and compares the algorithm simulation results with subspace narrowband signal DOA estimation algorithm.On this basis,the multi-level search method is proposed,so that the CS-DOA algorithm can reduce the computational complexity in the case of estimation accuracy.Then the spatial sparsity of signal in time domain is extended to frequency domain,obtains a wideband CS-DOA algorithm based on the compressed sensing theory.The article respectively discusses multiple dictionary algorithm,frequency domain focusing algorithm and spatial slicing algorithm through the simulation analysis,then compares them with signal subspace algorithm.At the end of this paper,the CS-DOA algorithm is used to process the data of water pool and lake.It can be seen from the results that the performance of CS-DOA algorithm in dealing with the real underwater acoustic signal is basically consistent with simulation results,and can be effectively applied to underwater acoustic signal passive array signal processing.It can be seen from the simulation results and the experimental analysis:the narrowband signal CS-DOA algorithm does not need to consider the relationship between wave signal,it has excellent ability to estimate DOA signal of different degree of correlation.Compared with the subspace algorithm,the CS-DOA algorithm generally has sharp spatial spectrum peak,and also has higher performance in small snapshot situation,but the computation of the algorithm is larger than the narrowband space based algorithm.The estimation performance of the wideband CS-DOA algorithm is independent of the type and correlation of wideband signal,so the performance of algorithm is better than that of the subspace based algorithm when dealing with strong correlation signal.The spatial spectrum of wideband CS-DOA algorithm is more acute in the direction of the signal,and the algorithm can still have a better estimation accuracy in the case of small number of array elements or large signal bandwidth.However,the algorithm has a large amount of computation as well as the narrowband CS-DOA algorithm,which can reduce the computation complexity by use of multi-level search methods.