Research On DOA Estimation Of Array Signals Based On Compressed Sensing

Direction of arrival?DOA?estimation is one of the important research directions in array signal processing.It is widely used in radar,electronic reconnaissance system,communication and other fields.Traditional DOA estimation algorithms are susceptible to noise,so there are many limitations in application.Compressed sensing theory observes and recovers signals in a new way,and it has good estimation performance in DOA estimation,which has attracted wide attention of scholars.In this context,the DOA estimation algorithm based on compressed sensing is studied.The main contents are as follows:?1?The array signal model is introduced,and the basic concept of compressed sensing and the reconstruction method of sparse signal are studied.On this basis,the similarities between DOA estimation and compressed sensing theory are analyzed,and the DOA estimation model based on compressed sensing is established.At the same time,the differences between DOA estimation and compressed sensing are pointed out,which indicates the direction for future research.?2?Compressed sensing theory is applied to DOA estimation of quasi-stationary signals,and a DOA estimation algorithm of quasi-stationary signals using sparse signal reconstruction is proposed.The algorithm transforms the covariance matrix of the received data by KR product,and DOA estimation is processed as a virtual array signal model after dimensionality reduction,denoising and realization.Based on the l1-SVD algorithm,the orthogonality of subspace is used to determine weights,which enhance the sparsity of solution vectors.The weighted l1-SVD algorithm is used to reconstruct the sparse signal to obtain DOA estimation of the source.The simulation results show that,compared with the subspace-based DOA estimation algorithm,the spatial spectrum of the proposed algorithm has sharper spectral peaks,the proposed algorithm has higher resolution for near-distance sources,and it also has better performance in the condition of low signal-to-noise ratio.?3?On the basis of the above algorithms,a real-valued DOA estimation algorithm using approximate l₀ norm?AL0-DOA?is proposed for general random signals,which makes the approximate l₀ norm algorithm applicable to DOA estimation with multiple snapshots,noises and complex operations,and has a fast computation speed.By using smoothing function to approximate l₀ norm,the problem of l₀ norm that cannot be solved directly is transformed into the optimization problem of smoothing function,which can be solved quickly by modified Newton algorithm.Simulation results verify the effectiveness of AL0-DOA algorithm,which has fast calculation speed and high accuracy.