Non-Linear, non-recursive estimation for passive localization and tracking

Present passive ranging algorithms are almost exclusively based on the extended Kalman filter with ad hoc variations tacked on in an effort to improve stability and predictability. For passive sonar applications, classical nonrecursive, non-linear estimation gives a stable and predictable algorithm, at the expense of an apparently larger but still acceptable computing load. A description and error analysis of one such "batch" algorithm is given, and its tracking performance compared against theory, for a single sensor obtaining bearing and doppler measurement sequences.