Target tracking by time difference of arrival using recursive smoothing

The paper presents a simple recursive solution to passive tracking of maneuvering targets using time difference of arrival (TDOA) measurements. Firstly, an iterative Gauss-Newton algorithm is developed for stationary target localization based on a constrained weighted least-squares (CWLS) criterion. The advantages of the CWLS estimate are its inherent stability due to the absence of local minima at infinity and its capability to match the performance of the maximum-likelihood (ML) estimate. To track maneuvering targets, a computationally efficient recursive least-squares (RLS) algorithm is developed, which smoothes successive stationary target location estimates obtained from the ML or CWLS solution using a constant-acceleration motion model. In simulation studies, the proposed recursive tracking algorithm is compared with a Kalman tracking algorithm that estimates the target track directly from the TDOA measurements, and is shown to be capable of outperforming the Kalman tracker.