The Research Of DOA Estimation Of Coherent Sources In The Presence Of Impulsive Noise

The estimation of direction-of-arrival is an important area in array signal processing. The methods based on array signal processing can estimate DOA of several sources in different directions simultaneously. Because of the special nature in coherent signals, the DOA estimation of coherent signals is a research hot and difficulty spot of this research area, which is widely used in radar, communication and some etc.Conventional DOA estimation algorithms are mainly proposed for the Gaussian noise. But in reality, there is a large number of non-gaussian distribution noise, such as atmospheric noise, sea clutter noise, man-made noise, they are impulsive noise. So DOA estimation in the impulsive noise environment becomes the important research topic. Because of the special nature in a stable distribution, the impulsive noise can be modeled asαstable distribution.In the thesis, The author proposed Some modified algorithm after analyzing the algorithm of coherent signals in a stable distribution and the questions of existing in them. The main work of the thesis is in several aspects as follows.(1) First of all, The conventional algorithm of DOA estimation are analyzed and three de-correlation Algorithm are introduced, mainly analyzed the nature of Singular Value Decomposition. Some simulation results are given.(2)Then, Singular Value Decomposition is based on fractional lower order moments matrix is proposed for one-dimensional coherent signal sources in the presence of impulsive noise. The method is not only successfully apply matrix rearrangement to DOA estimation in non-gaussian distribution noise, but also inherit the inhibition for impulsive noise of fractional lower order matrix. Simulation results show the method has good capability of spectral peak searching and lower probability of error.(3) At last, make use of mode space transform, the method proposed in this paper, singular Value Decomposition based on fractional lower order moments matrix, is successfully used in two-dimensional uniform circular array model and have the advantage of directional and Non-ambiguity direction measurement, and good capability of spectral peak searching. Simulation results show Singular Value Decomposition is based on fractional lower order moments matrix can successfully used in two-dimensional DOA estimation for coherent sources.