Bearings-only constant velocity target maneuver detection via expected likelihood

This work addresses the problem of target maneuver detection using passive bearing measurements when the target motion dynamics can be described by a discrete constant velocity (CV) model. The proposed expected likelihood (EL) maneuver detector (ELMD) extends the EL approach from a direction of arrival estimation framework to maneuver detection of targets governed by a CV model. As a result of its array signal processing nature, and in contrast to measurement residual-based maneuver detectors, ELMD does not require statistical assumptions of the bearing measurements, and its maneuver detection performance improves with increasing signal-to-noise ratio and sample support. Maneuver detection performance of ELMD and two conventional maneuver detectors for the passive bearings-only target tracking problem are evaluated in this work via simulations using mean detection delay, detection delay standard deviation, and probability of false alarm criteria. ELMD is shown to outperform the other two evaluated algorithms in all tested practical scenarios.