Research On The Algorithm Of Off-grid DOA Estimation Via Sparse Data Model

The Direction of Arrival(DOA)estimation is an important part of the passive direction finding system and is widely used in radar,sonar,and location detection communication systems.The sampling frequency of the traditional direction finding algorithm is limited by the Nyquist sampling law.The higher sampling frequency has higher requirements on the hardware device,and the higher direction-finding performance can be obtained only if the direction finding environment is ideal.Since the signal is sparse in the entire space,Compressed Sensing(CS)theory can be applied to the estimation of the direction of arrival.This theory can overcome the above problems and has certain advantages.Compressive sensing theory provides a completely new perspective on the direction of arrival estimation,turning the direction of arrival estimation into a sparse reconstruction problem,and then using a sparse solution to solve the problem.However,the DOA estimation problems in the CS frame has the problems of large amount of computation and long computation time,which hinders its engineering application.This paper studied DOA estimation algorithm for signal departure from the sparse model establishment in sparse reconstruction of data domain,and proposed an improved method to improve the algorithm's direction finding accuracy and reduce the computational complexity.First of all,the thesis introduces the research background of the topic,outlines the research progress and research status of DOA estimation and compressed sensing theory.This paper studies the theoretical basis of DOA estimation under the CS framework,and briefly describes them from the establishment of the signal model and the introduction of DOA estimation algorithms.For the CS frame structure,the three aspect of signal sparse representation,the design of the measurement matrix and the signal reconstruction algorithm are introduced.The DOA estimation mathematical model under the CS framework is established.Secondly,by constructing real sparse Bayesian models with second order Taylor expansion to improved sparse Bayesian escape algorithm based on real value effectively reduced model errors.Then the complex problem can be transformed into a real-valued problem through the unitary transformation,and the singular value decomposition of the measured value can reduce the dimension of the measured value matrix and reduce the computational complexity.At last,a real-valued sparse Bayesian model was established.By using a priori information and a series of iterative optimization processes to solve posterior density function realized DOA estimation.Through computer simulation experiments,the effectiveness of the algorithm was verified and the performance of the algorithm was analyzed accordingly.Finally,aiming at the problem of large model error in the approximation model based on the first-order partial derivatives,the establishment of a steering vector model based on trigonometric function approximation is proposed to effectively reduce the model error and improve the direction-finding accuracy.Then sparse Bayesian model is constructed,using prior information,iteratively seeking to solve the posterior density function.When solving the outlier parameters,two different solving methods are selected.Through computer simulation experiments,the direction-finding performance under these two methods is analyzed...