Target localization methods for distributed MIMO radar systems and co-prime array systems

Abstract

In the first part of this thesis, we focus on the methods for target localization in multiple-input multiple-output (MIMO) radar with widely separated antennas (WSA). Chapter \ref{chapter2} addresses an effective algorithm for static target localization in MIMO radar systems with widely separated antennas. The algorithm derived uses time-of-arrival (TOA) measurements from multiple transmitter-receiver pairs and is based on a hyperbolic method suitable for radiation source localization in passive sensor networks. It does not have the local convergence problem as the conventional iterative method. Some combinations of the derived algorithm and the conventional iterative method are presented. In a numerical example, it is shown that the proposed methods can achieve the Cramer-Rao lower bound (CRLB) in the range of moderate processed measurement noise and obtain the better localization performance as the number of transmitters and receivers increases. Furthermore, some remarks are made on the robustness of the proposed methods. In Chapter 3 we propose an improved method for moving target localization with a noncoherent MIMO radar system having widely separated antennas. The method is based on the well-known two-stage weighted least squares (2SWLS) method as an extension of the method in Chapter 2, but in contrast to the recently proposed Group-2SWLS, it requires only one reference transmitter (or receiver) without grouping and combining. This change allows us to obtain a closed-form solution which is less likely to be degraded by bias. Furthermore, the proposed method can easily utilize not only TOA and frequency-of-arrival (FOA) data but also time-difference-of-arrival (TDOA) and frequency-difference-of-arrival (FDOA) data. We also introduce new auxiliary variables for the purpose of numerical stability

our method using the auxiliary variables is shown to be numerically more stable than the Group-2SWLS, while attaining the CRLB at higher noise levels. The adoption of auxiliary variables requires no additional computations in contrast to the adoption of the concept of Turbo-2SWLS for the same purpose. In the second part of this thesis, we deal with direction-of-arrival (DOA) estimation. In Chapter 4, two cyclic coordinate descent (CCD) algorithms for nonnegative gridless compressive sensing suitable to solving the problem of line spectral estimation are proposed. The methods perform atom update in subdomains of the continuous space relying on atom merging and atom activating, exhibit fast convergence and have practically feasible computational complexity. In an application to the direction-of-arrival (DOA) estimation in co-prime arrays, which is an instance of line spectral estimation, we demonstrate that the proposed methods are superior to the joint sparsity reconstruction method (JLASSO) and the MUSIC method with spatial smoothing (SS-MUSIC) in terms of several criteria, and analyze their convergence properties. In Chapter 5, under some practical assumptions, we design the optimal steering angles for direction-finding systems using two beams in the sense of minimizing a derived CRLB given a DOA. It is proved that an acceptable approximation of the optimal steering angles, referred to as the best mirror steering angle, can be obtained by solving a simplified design problem. Analyzing the CRLB and the best steering angle as functions of beam phase center spacing (BPC spacing), we shed light on a connection between monopulse squint angle and the best mirror steering angle. Furthermore, via a sensitivity analysis, we investigate how sensitive the designed steering angles are to the deviations of the DOA, beamwidth, and BPC spacing. Computer simulations are performed to validate our analysis and results.