Consideration is given to recursive filtering algorithms for passive sonar tracking of a maneuvering target. An essential ingredient of such algorithms is a stochastic model describing the target motion. The basic problem posed by a maneuvering target is that there exists a mismatch between the modeled target dynamics and the actual target dynamics, especially when the target undergoes sudden pilot-induced maneuvers. Traditionally, diffusion or random walk models that are easy to handle but of questionable validity have been used. A more realistic, linear piecewise deterministic model (LPD) is proposed, consisting of straight-line tracks with random changes of course. The model is then a piecewise deterministic Markov process. A general method which gives a simple functional approximation to the solution of the forward equation associated with the LPD model is introduced. This functional approximation can be used in conjunction with appropriate global approximation methods to produce a practical nonlinear filter for target tracking.<<ETX>>
[1]
Harold W. Sorenson,et al.
On the development of practical nonlinear filters
,
1974,
Inf. Sci..
[2]
R. Bucy,et al.
Digital synthesis of non-linear filters
,
1971
.
[3]
Harold W. Sorenson,et al.
Recursive Bayesian estimation using piece-wise constant approximations
,
1988,
Autom..
[4]
Mark H. A. Davis.
Piecewise‐Deterministic Markov Processes: A General Class of Non‐Diffusion Stochastic Models
,
1984
.
[5]
H. Sorenson,et al.
Recursive bayesian estimation using gaussian sums
,
1971
.