Research On Source Detection And Estimation Based On Sparse Bayesian Learning

Direction of Arrival(DOA)estimation,as an important research content of array signal processing,is widely used in radar,sonar,and communication systems,so it has attracted widespread attention from researchers.The electromagnetic environment of modern battlefield is deteriorating day by day,which makes it more difficult to observe the target signal.Therefore,higher requirements are put forward for DOA estimation algorithm,especially in the condition of small snapshot and low signal-to-noise ratio,traditional algorithms are difficult to meet the needs.In recent years,sparse representation theory is widely used in the field of signal processing.Its basic idea is to represent signals with linear combination of as few atoms as possible under the condition of given redundant dictionary.To a certain extent,this theory can overcome the shortcomings of traditional algorithm.In the actual direction finding environment,the target signal satisfies the sparsity in the spatial domain,so sparse representation theory can be used in the DOA estimation of array signal,this paper will study the DOA estimation algorithm under sparse representation model based on sparse bayesian learning theory,in order to improve the estimation performance of direction finding algorithm,the main contents are as follows:Firstly,the theoretical basis of DOA estimation is studied,including array signal receiving models and classic subspace DOA estimation algorithms.In addition,the theoretical basis of sparse representation is studied,and the rationality of using it for DOA estimation is analyzed.Based on this,the basic theory of sparse Bayesian learning is studied,and the advantages compared with other sparse reconstruction algorithms are analyzed.Then,to solve the problem of poor performance of DOA estimation under the condition of a uniform linear array,a sparse Bayesian learning DOA estimation algorithm based on coprime array is studied.By vectorizing the autocorrelation matrix of data received by array,the aperture of array is expanded.Based on this,a sparse bayesian model is established,and the noise variance is regarded as a part of the signal,the maximum posterior estimation is used to reconstruct sparse signal.The simulation results show that the algorithm is effective.Finally,aiming at the model error caused by grid mismatch,the discrete grid is regarded as a variable,which is jointly estimated with sparse signal,and then the off grid DOA estimation model is established.In order to improve the performance of DOA estimation,instead of directly processing the data received by the array,the sparse model of the equivalent signal is constructed by calculating the weighted subspace of the received data,which has higher estimation accuracy.Then,the sparse signal is reconstructed in the block sparse Bayesian framework.The effectiveness of the proposed algorithm is verified by comparing other mainstream algorithms through simulation experiments.