A linear least squares algorithm for bearings-only target motion analysis

Traditional bearings-only target motion analysis (TMA) statistical models assume a priori that measurements are independent when conditioned on the target. This paper presents a novel track-before-detect “empirical” maximum a posteriori (EMAP) approach in which measurements are assumed independent prior to the detection decision. The EMAP estimators proposed here are joint detection/estimation methods whose intended use is target tracking. A limiting case of the EMAP formulation is shown to be equivalent to the traditional maximum likelihood (ML) formulation. Triangulation and constant velocity target examples are presented. The EMAP algorithm is an iteratively re-weighted linear least squares algorithm for these problems, and has significantly lower computational complexity than the standard ML estimator