Research On Direction Finding With Partly Calibrated Arrays

Array signal processing is an important branch of signal processing in many applications.In practical applications,errors of array can not avoid,which include senors position errors,gain/phase errors,etc.All of them are impact effect on the direction of arrival(DOA)estimation.In practice,the calibration methods for array errors are so complex that the calibration techniques are the bottleneck for the deep application of the array signal processing.Thus,exploring the automatic calibration of array errors is very important for the radar,sonar,communication and so on.In this dissertation,we focus on the problem to calibrate gain and phase errors and DOA estimation with partly calibrated cells of the array.The key content of the dissertation include:1.Based on ESPRIT techniques,we do some research on the DOA estimation and the calibration of gain and phase errors.We transform the problem to a quadratic minimum estimation with calibration constraints.The algorithm realizes the joint estimation of DOA and the errors.Furthermore,we explore the modified algorithm for the joint estimation.The modified algorithm uses multi-variable analysis for solving the minimum of the cost function.After several iterations,we can obtain more precise estimation with a little more computation load than that of the first algorithm because the whole apertures of the array are used.2.Deeply study two methods of array calibration and DOA estimation with partly calibrated cells in the condition of Gaussian color noise.In the case of uncorrelated signals,the array can be calibrated according to the specific structure of the data covariance matrix.After the gain/phase cancellation,the non-uniformity of array noise can be transform to uniformity noises by reformulating the calibrated data covariance matrix.For correlated signals,the non-uniformity of senor noises is mitigated by solving a least squares minimization problem.Using factor analysis method,we can obtain the signal subspace and color noise covariance matrix.Then,using ESPRIT technique,we can determine the DOA.The effectiveness of the proposed methods is illuminated by the simulation experiments.