Optimizing the receiver maneuvers for bearings-only tracking (Q1301372)

The optimization of an observer trajectory (a planar one) for the target motion analysis is discussed. The problem is put into the form of a discrete state space vector equation. The space state vector \(X\) is the difference between two subspace states: \(X_s\), which includes the coordinates and velocity components of the source, and \(X_{\text{rec}}\), which contains the analog quantities for the observer. The input vector \(u\) includes the controls applied by the observer. It is assumed that between two consecutive moments \(t_k, t_{k+1}\) the velocities are constant and rectilinear. The state vector must be such that a functional over the Fisher Information Matrix (FIM) is maximized. The essential features of the discussed problem are: the parameters of the trajectories are only partially observed and the FIM possesses all required properties except monotonicity. These facts make the problem a nonclassical one. A large part of the paper is devoted to the analysis of the FIM properties. It is shown that under the hypothesis of bounded controls, these belong to the class of bang-bang ones. The merit of the paper consists in providing a general framework for the optimization of the observer strategy based only on observations (bearings as the authors call them). We think we are faced with a basic paper in the domain of target motion analysis.