Research On Algorithms Of DOA Estimation And Tracking Based On Compressed Sensing Theory

The requirement of DOA estimation and tracking technology is getting higher and higher with the application of them in many fields like radar detection and sonar localization being more and more wide and deep.In recent years,some distinct advantages,such as the low requirement of snapshots and the decoherence in nature,promote the research of DOA estimation in array signal as the developing of compressed sensing.More further,the DOA tracking technology based on compressed sensing also starts gradually with the formation and improvement of theories like dynamic compressed sensing.In this paper,we study the DOA estimation and tracking algorithms based on compressed sensing in the above context and the main contents of the research are summarized as follows:1)A DOA estimation algorithm based on improved smoothed LO norm is proposed by approximating the LO norm with smoothed function of better performance after the research of smoothed LO norm.This algorithm is easy to accomplish and accurate.Compared with the OMP algorithm and the primitive algorithm of smoothed LO norm,our algorithm has better performance and can achieve better DOA estimation just in one snapshot.2)A DOA estimation algorithm based on re weighted smoothed LO norm is proposed by giving the improved smoothed LO norm a new weight.This algorithm is also easy to accomplish.Compared with the improved smoothed LO norm,this algorithm has higher accuracy and can achieve better DOA estimation just in one snapshot.3)A DOA estimation algorithm based on reweighted smoothed LO norm with multiple snapshots is proposed by extending the reweighted smoothed LO norm to MMV model because the DOA estimation technology is usually under the MMV model in practical application.This algorithm can achieve high precision estimation of DOA just in a few snapshots.4)In order to track the DOAs of moving signal sources,a DOA tracking algorithm based on dynamic compressed sensing in single snapshot is put forward by applying dynamic compressed sensing to dynamic DOA and establishing the sparse probability model of dynamic DOA to obtain the weighted L1 norm and minimizing the L1 norm with linear weights.This algorithm can achieve high precision tracking of DO A and has better performance than PASTd algorithm and particle filter algorithm in certain signal to noise ratio.5)In order to promoting the noise suppression capability of DOA tracking algorithm based on dynamic compressed sensing,this algorithm is extended to the MMV model.Hence a DOA tracking algorithm based on dynamic compressed sensing with multiple snapshots is proposed after singular value decomposition for the received signals to reduce calculation.This algorithm can achieve high precision tracking of DOA just in a few snapshots.