Renewal models for maneuvering targets

Most model-based tracking algorithms represent the temporal dynamics of a maneuver acceleration with a Markov process. The sample paths so generated may not be plausible. A renewal process model is shown to be more realistic, but a tracker based upon it is seen to be more complicated to implement. The response of the renewal-model tracker is compared with a simpler estimator based upon a Markov model both for the case in which the tracker utilizes an imager, and the case where it does not. It is shown in the latter case that performance improvement is not commensurate with algorithmic complexity. For an image-enhanced tracker, a sophisticated model of target dynamics promises significant performance improvement.<
>