Research On Direction Of Arrival Estimation With Super Resolution

With the rapid development of microelectronic technology, digital signal processing technology and parallel processing technology, array signal processing theory and practical application have become more and more active. DOA estimation theory and technology have become more sophisticated, but there are still many fields worth researching. Based on the array signal processing methods, this paper will devote to the research in the field of super-resolution algorithm on DOA estimation, including estimation spatial resolution and the evaluation criteria, false peak and angle ambiguity, precise high-speed parallel realization of DOA estimation based on cyclic high order cumulants and subspace estimation algorithm based on eigendecomposition and other related issues. Specific studies are as follows:1. Based on the basic MUSIC algorithm and the traditional evaluation methods in DOA estimation, using the basic probability theory as a tool, the concept of confidence interval in random variables is introduced to the super-resolution array system. Then detailedly study some important parameters and properties affecting direction finding performance when using array antenna to estimate DOA, and give a new, simpler and more intuitionistic evaluation method of the super-resolution performance. Reveal the qualitative relations and quantitative mathematical expressions between the spatial resolution of direction finding algorithm and parameters such as the number of antenna array elements, antenna array element spacing, the number of sampling snapshots, SNR of received signal, angles of DOA. Via promotion, the method can be applied not only to antenna array with different manifolds, such as ULA, UCL, but also to the different algorithms, such as ESPRIT algorithm, maximum likelihood algorithm. Therefore, the evaluation method is a universal application.2. Using antenna array spacing as a parameter, analyze the impact of elements spacing to finding angle ambiguity, quantitatively analyze the problems brought by false peaks in direction finding system, reach a number of practical conclusions, meanwhile give practical method to pick out real peaks from false peaks in direction finding system, and verify the correctness of the methods through a lot of simulation experiment results.3. Detailed and systematic introduce the definition and basic properties of high order cumulants and cyclic stationary theory. Analyze and compare the differences and relationship between the modem signal processing method based on cyclic high order cumulants and traditional signal processing method based on the Gaussian process. On this basis, bring forward TLS-ESPRIT algorithm based on cyclic high order cumulants, verify the excellent performance of the method on resisting Gaussian noise and interference with different cyclic frequency through simulation experiments and the improved algorithm has lower complexity than the traditional direction finding algorithm based on common high order cumulants, together with good real-time performance, which has greatly enhanced its practical value.4. Based on currently the most widely used algorithm MUSIC,theoretically divide the algorithm into specific steps to realize: calculation of correlation matrix of receiving signals from the antenna array, eigendecomposition of matrix, construction and search for spatial spectrum. After studying the amount of time of these steps in practical computation, bring out different hardware schemes when real-time realization of algorithm, introduce pipeline technology and data interleaving processing technology, which greatly accelerates the computation speed.