📝 SummarXiv

Abstract

Passive bearing-only tracking (BOT) estimates the target states by utilising noisy bearing measurements captured by a sensor array. The sensor array is often towed behind the ship, using a long flexible cable to reduce interference from the own-ship's inherent noises. This forms a towed cable sensor-array system (TCSAS). During BOT, the tow-ship has to perform a manoeuvre to make the tracking system observable. Such a manoeuvre destabilises the TCSAS, thus making its exact location unknown \emph{w.r.t.} tow-ship. However, it is very crucial to know the exact location of the towed sensor-array to perform efficient and reliable target state estimation. The existing BOT approaches perform TMA during own-ship manoeuvre either by pausing the measurement updation step of the estimation algorithm or assuming a fixed aft position for the towed sensor-array. These assumptions lead to unreliable state estimation. To address this, we propose a dynamic model for TCSAS, using a lumped mass approach, which will provide the location of the sensor array during the own-ship manoeuvre. This location will be fed to the state estimation algorithm. The dynamic of TCSAS in 3D space is obtained by solving the equations obtained from the moment balance condition and quasi-static equilibrium condition at the lumped mass points. Moreover, the bearing data captured by the towed sensor-array is corrupted with non-Gaussian noise. It is handled using the maximum correntropy criterion based Kalman filter with a kernel bandwidth selection technique, proposed in this paper. The proposed sensor-array dynamic model is verified for a real-world BOT engagement scenario.