Bearing Line Tracking and Bearing-Only Target Motion Analysis

The framework of this paper is a passive sonar system, more precisely at the core of bearing estimation and bearings-only target motion analysis (BO-TMA). A cosine of relative bearing estimation and bearings v.s. time image is first obtained by a conventional frequency-domain beam-former [D.H Johnson et al, 1993]. Computing the first two moments of each line of this image, we evaluate the likelihood of the cosine of relative bearing of the target. Then, the cosine of relative bearing and its derivative are considered as the two components of a state vector of a linear dynamic system. This state vector is then estimated (or extracted) by a classical algorithm of the hidden Markov model (HMM) arsenal, whose parameters are adjusted according to the statistical assumptions. Unlike, we consider the presence of one sole target, but the beginning and the end of the line are unknown and must be estimated too. The extracted track is used as measurements set of the BO-TMA. Finally, the confrontation of the TMA results allow us to evaluate the performance of the triplet (beam-former, bearing extraction, BO-TMA).