Direction Of Arrival Estimation Algorithms Based On Covariance Matrix Sparse Reconstruction

In recent years, with the rapid development of information technologies, traditional DOA estimation algorithm requires a lot of poor sampling data, coherent signal processing effects, require a higher signal to noise ratio and other reasons, in the emerging field of application has been greatly restricted. Therefore, how to use a limited observation data to achieve high resolution signal parameter estimation, has become an important development direction. With the rise and development of the theory of compressed sensing, it for DOA estimation technology development open up a broader space. The theory points out that under the conditions allow, sparse or compressible signal can be high precision recovered from a limited sample data. In order to overcome the defects of traditional algorithms and effectively using the airspace sparse feature of the signal, this paper uses the theory of compressed sensing to estimate the parameters of the signal.And now, the main content and the finished work of this dissertation are listed as follows:1. Starting from the development history of the theory, the related background knowledge of the theory of the DOA estimation algorithms and compressive sensing are summarized.2.Briefly analyzes the basic theory of array signal processing, and gives the basic mathematical model of uniform line array(ULA), and an introduction to traditional narrow-band signal DOA estimation algorithms, including several classic coherent and non-coherent narrowband signal processing algorithms.3.Focuses on the three core elements of the theory of compressed sensing emphatically, the conditions of sparse reconstruction RIP and MIP properties did a simple analysis. Meanwhile, in view of the airspace signal sparse features, this paper sets up the reconstructed model and introduces the classic L1_SVD algorithm based on the theory of compression sensing.4.By L1_SRACV algorithm which is the classic sparse reconstruction based on covariance matrix, this paper leads to the improved measures of such algorithms, including using the methods of KR product transformation and forms of covariance matrix transformation to obtain new observation vector and redundant dictionaries. And, using a large number of simulation experiments to explain the advantages and disadvantages of the improved algorithms.