Target Association Using Detection Methods

A residual-based scheme is presented for solving the radar track-to-track association problem using bearings-only measurements. To accomplish track association between two stations, the residuals of a bank of nonlinear filters called modified gain extended Kalman filters are analyzed. Once tracks have been associated between two stations, tracks from additional stations may be associated with tracks from the first two stations by checking algebraic parity equations. Traditional track association methods rely on the local stations' estimated target positions and error variances. These local estimates may be quite inaccurate or even divergent when using bearings-only measurements. Our method bypasses this difficulty because our filters use raw data from multiple stations. An example demonstrates that our methods yield results superior to those of standard methods.