Matrix Completion Algorithm And Its Application In DOA Estimation

Most DOA(Direction Of Arrival)estimation algorithms are based on the assumption that noise is Gaussian white noise,the element in array has no damage and the incident source is independent irrelevant.However in a real world applications,due to the external environment of array or the difference of array element itself,sometimes the data we get from array is incomplete.In most cases,the noise can be colored noise.Therefore,we discusses the DOA estimation under the condition that data is incomplete and the noise is colored noise.Above all,the main contents of this paper are as follows:Firstly,analyze the advantages and the disadvantages of most classic DOA estimation algorithms;then analyze the model of colored noise.Secondly,consider the condition that data received from array is incomplete due to some damage to array.A new method is proposed to estimate the DOAs of multiple spatial sources using a uniform linear array assuming that some of the sensors happen to be out-of-order.Firstly,a Matrix Completion(MC)algorithm is applied to recover the observed incomplete data,and then an DOA estimation algorithm using the recovered data is proposed to obtain the correct parameter estimation.Finally,computer simulation results show that the proposed algorithm has a great performance improvement compared to those based on incomplete data in terms of Signal-to-Noise Ratio(SNR)and the sample rate of sensors.Thirdly,a new algorithm is proposed to deal with a most common colored noise,then analyze the efficiency of the algorithm.We consider the situation that the background noise is nonuniform colored noise in the arrays.Traditional DOA(Direction of Arrival)estimation algorithms based on subspace such as MUSIC(Multiple Signal Classification)and ESPRIT(Estimation of Signal Parameters via Rotational Invariance Technique)cloud be failed due to the nonuniform colored noise.In this paper,we propose a new method to deal with the unknown nonuniform colored noise.After simulation,the results show that our method has a great performance when SNR(Signal-to-Noise Ratio)is very low.