Direction Estimation Under Array Error Condition

Array error is almost unavoidable in the direction finding based on sensor array in practice. The performance of the DOA(Direction of Arrival) algorithms decreases badly in array error condition for the array manifold is different from the ideal model. It's one of the main reasons for many DOA methods' applications to real systems have been very limited. Therefore it be of current significance to research robust DOA algorithms. In this dissertation, the self calibration based on sparse decomposition and eigenstructure are studied to calibrate the channel gain and phase error and array element location error and mutual coupling in array as well as estimate the DOA. The main work of the thesis are listed as follows:1. An analysis of effect of sensor gain and phase uncertainty on MUSIC algorithm is presented. The performance of the self calibration based on subspace method and the calibration based on single auxiliary source is tested by computer simulations. Then we discuss the feasibility of the self calibration based on MUSIC apply to estimation of the 2 dimensional (2-D) DOA and sensor gain and phase error with uniform circle array.2. Self calibrating methods for direction finding with sensor gain and phase uncertainties and sensor location errors are designed. The methods which combined signal sparsely decomposition and genetic algorithm provide high accuracy of DOA estimation in the prensence of array error condition as well as calibration of each sensor. In addition the approachs have good performance in the case of serious array error.3. Since the mutual coupling model of uniform linear array (ULA) and uniform circle array (UCA) are special, self calibrating approachs based on transformation of matrix with sensor mutual coupling uncertainty of UCA and with sensor mutual coupling and gain and phase uncertainty of ULA are presented. The new approachs provide higher accuracy of initial DOA estimation without the imformation of mutual coupling. The iterating process before convergence of the algorithms get shorter. At last a algorithm which apply the mutual coupling calibrating method to the estimation of DOA for wideband LFM signal is presented. The algorithm based on time-frequency analysis provides robust estimation of DOA in the presence of sensor mutual coupling, and mass calcutions are needless.