DOA Estimation And Tracking Algorithm Based On Evolution Strategy

A number of algorithm can be used for DOA estimation,but the basic principles and general steps are the same,through the observation of a large number of data snapshots,and then estimate the signal covariance matrix. The signal subspace and noise subspace can be isolated through a variety of matrix decomposition such as QR decomposition,eigenvalue decomposition, singular value decomposition. Pseudo-spectral can be estimated by the orthogonal characteristics of noise subspace and signal subspace. Different methods can be used for estimate the pseudo-spectral, such as one by one searching and the way of polynomial rooting,There are many methods can be used for the DOA tracking problems,such as using the batch method,recursive methods,evolutionary algorithm-by-snapshot search,and DOA tracking and Kalman algorithm combined with another idea that is also DOA tracking. On the basis of these problems this paper made some research:(1) This paper elaborate the basic principles and mathematical theory of DOA estimation. There kinds of commonly used formation such as array,planar array and circular array,and the quantum genetic algorithm, clonal immune evolutionary algorithm and improved particle swarm optimization.(2) This paper combine quantum genetic algorithm, cloning immune evolutionary algorithm with the multi-dimensional DOA estimation,thus able to estimate the signal's azimuth and elevation,improved anti-noise performance of the multi-dimensional parameter estimation, numerical stability and run-time. The validity of the algorithm are proved by the computer simulation.(3) This paper combine the particle swarm optimization and DOA tracking,using an improved particle swarm algorithm for a particular fitness function to search the global optimal value, and greatly improves the estimation accuracy and the run time. The is also a new research direction of the DOA tracking.(4) Aiming at the problem of the DOA(Direction-of-Arrival)tracking algorithm performance will be serious decline when the signal source coherent, analyzes the coherent covariance matrix rank losses' reason and the effect of covariance matrix rank losses on tracking algorithm. In the covariance matrix updating, take the recovery for the covariance matrix's rank of every new data quickly shoot by two times of matrix multiplication, and then gives the optimal solution of cost function recursive form. This algorithm can overcome the influence of signal source coherent, and do not need take an average of a large number of data quickly shoot to make covariance matrix and decompose matrix characteristics, so it save a lot of operation time and improve the tracking algorithm's real-time ability and the tracking performance in case of coherent, the computer simulation proves the efficiency of the method.