Research On Localization Algorithms For Moving Target Based On Multiple Parameters

In the past decades, due to the important role in many military and civilian fields such as radar, sonar, navigation, electronic countermeasure, emergency rescue and wireless communication, target localization technology has drawn considerable attention and been widely applied. In a common localization system, the target states determine the multiple parameters of the received signals. How to utilize these parameters to locate the target is the key point of the researches. Especially in the case that the target or the receiver is moving, it is a difficulty to localize a target by using the highly non-linear relationship between the Doppler shift measurements and the target states. This thesis studies on several target localization problems. The main contents and contributions are summarized as follows.1. The passive localization problem for moving target using time difference of arrival(TDOA) and frequency difference of arrival(FDOA) is studied. Based on the signal receive process in practice, the received signals are modeled as involving TDOA and FDOA as well as phase difference. According to the signal model, the maximum likelihood(ML) estimation is proposed to directly estimate the target position and velocity without first estimating the TDOA and FDOA, and the Cram′er-Rao low bound(CRLB) is derived. Theoretical analysis and simulation results corroborate that the proposed method performs better than the conventional two-steps methods.2. The active localization problem for moving target based on time delay and Doppler shift is researched, especially for the multiple-input multiple-output(MIMO)radars with widely separated antennas. Based on time delay and Doppler shift estimates,a group two-steps weighted least squares(2WLS) method is proposed. By utilizing the time delay and the Doppler shift measurements together, the proposed method can estimate the target position and velocity simultaneously. The proposed method has explicit solution and need not initialization. Both the position and velocity estimation of the proposed method can achieve the CRLB accuracy at low noise conditions.3. The localization problem for moving target using only Doppler shift is studied.The Doppler shift measurement models in an active localization problem and a passive localization problem are considered. After normalizing, both of the two Doppler shift measurement models can be represented as the same form. For the localization problem based on only Doppler shift measurements, the existing feasibles solution are only grid search method and local optimization method. However, the grid search method requires high computational complexity, and the local optimization method requires good initialization and cannot guarantee convergence. To overcome this difficulty, the convex optimization technology is applied to solve the localization problem using only Doppler shift measurements for the first time. A direct SDP method and a differential SDP method are proposed to approximately solve the ML estimation in the case that the receiver is stationary and moving, respectively. The proposed method can estimate the target position and velocity simultaneously without initialization, and has no divergence problem.4. The stationary target localization problem using only Doppler shift measured by moving receivers is researched. Again, the existing feasible solution are only grid search method and local optimization method. To handle this problem, a 2WLS method, an indirect SDP method and a direct SDP method are proposed, all of which need not initialization and have no divergence problem. After that, for a stationary target, researches show that the Doppler shift measurement and the direction of arrival(DOA) measurement of linear array(LA) can be transformed into each other. In the two-dimensional(2D) plane,the Doppler shift measurement can be transformed into LA-DOA measurement, which can be used to estimate the target position by employing 2D cross-location method. In the three-dimensional(3D) space, the Doppler shift measurement and the LA-DOA measurement can be combined to form the azimuth and elevation angle measurements. At last, in the case that each receiver in 3D space can obtain only one LA-DOA measurement, the traditional cross-location method cannot work. However, researches show that the LADOA measurement can be transformed into Doppler shift measurement, and hence can be applied to estimate the target position by using the proposed Doppler shift based localization methods. Simulation results corroborate the theoretical results and the promising performance of the proposed methods.5. Passive source localization problem based on TDOA and gain ratios of arrival(GROA) is studied. Based on the TDOA and GROA measurement models, the constrained weighted least squares(CWLS) estimator is presented. After that, a SDP method is proposed to approximately solve the CWLS problem, and a root-find(ROOT) method is proposed to directly solve it. Owing to the special feature of the CWLS problem, both the SDP method and the ROOT method can usually find its optimal solution. Therefore,the proposed methods can provide more accurate location estimates than the existing2 WLS method, which approximately solve the CWLS problem.