Research On Calibration Method For Array Gain And Phase Uncertainties In Doa Estimation

High resolution direction-of-arrival(DOA) estimation method is one of the main research direction in array signal processing, which is widely applied to military and civilian areas such as radar, sonar, biomedicine, astronomical observation and so on. In order to guarantee the high accuracy of the parameter estimation, DOA estimation methods usually require that the array manifold is precisely known. However, in practical applications the array errors(e.g., gain and phase errors, mutual coupling, sensor location errors and so on) occur, which leads to the departure of the true array manifold from the ideal one. As a result, the DOA estimation methods may degrade dramatically or even become invalid in the presence of array errors. Therefore, the existence of array errors is a bottleneck which impedes the practicability of the high resolution DOA estimation methods. In this paper, we studied the calibration method for array gain and phase errors, the main research contents are as follows:Firstly, we constructed array gain and phase errors mathematical model based on the high resolution MUSIC algorithm, and analyzed the effect of array gain and phase errors on the accuracy of DOA estimation through simulation experiments.Secondly, for the calibration method using sources in known locations, we propesed a new calibration method using one source in known location based on received data vector. Compared with calibration method using multiple sources in known locations, our method is easier to achieve in engineering, and can estimate gain and phase errors more accurately. Then studied the property of the two order covariance matrix of received data, we proposed an ISM based on covariance matrix(R-ISM) correction method. It need only two auxiliary sensors, and can calibrate gain and phase errors in multi sources case, compared with ISM method, R-ISM method doesn’t have the problem that the number of auxiliary sensor is higher than the number of the source number and has good estimate performance. Aiming at the classical self-calibration method, analyzed the estimation performance from two aspects of the iterative threshold and the initial value of the iterative, and studied classical self-calibration improved method based on initial value preprocessing.Lastly, we proposed a Newton iterative self-calibration method based on maximum likelihood criterion(N-ML), the method firstly utilized the cost function of maximum likelihood criterion, deduced the first derivative and Hessian matrix of the cost function regarding array gain and phase errors. Then, utilized Newton iterative to estimate the array gain and phase errors and DOA parameters. Simulation experiments show that N-ML method is better than the maximum likelihood criterion self-calibration method in the computation complexity and estimation precision, and N-ML method obtained better parameter estimation effect under the condition of source coherence.