The initial portion of this thesis examines the problem of tracking a maneuvering target in the 2-dimensional (X,Z) plane, vertical to the ocean floor, using passive time-delay measurements. The target is free to maneuver in velocity and make depth changes at times unknown to the observer. In the past, tracking systems have used Extended Kalman Filters to process the nonlinear measurements, but these have inherent divergence problems. To overcome this, a nonlinear prefilter is added to linearize the measurements and thus allow the use of a conventional Kalman Filter which makes the tracking system more 'robust' and also decouples the depth estimator from the polar range estimator. The depth estimator is discussed in detail here.

The latter part of this thesis introduces tracking in the 2-dimensional horizontal (X, Y) plane, parallel to the ocean floor, to observe polar range and target bearing angle. The approach of using a nonlinear prefilter and a standard Kalman Filter is similar to the one described above. Subsequently, the analysis is extended to a Kalman Filter which is not 'matched', i.e. it does not possess any knowledge of the deterministic inputs which cause target motion. This necessitates the use of a bank of Kalman Filters and an adaptive weighting scheme. Test results are included to show that all source maneuvers can be tracked with a relatively high degree of accuracy.