Title

Practical issues in target tracking

Date of Completion

January 2001

Keywords

Engineering, Aerospace|Engineering, Electronics and Electrical|Remote Sensing

Degree

Ph.D.

Abstract

This dissertation is on three topics in target tracking: the effect of radar/sonar waveforms on tracking performance, a modified probabilistic data association filter (PDAF) based on a Bayes detector and sufficient conditions for the matrix Cramér-Rao lower bound (CRLB) scaling due to measurements of uncertain origin. ^ In the first part (chapter 2), the effect of radar/sonar waveforms on tracking performance is investigated. The traditional approach for tracking system design is to treat the detection and tracking subsystems as independent units. However, the two subsystems can be designed jointly to obtain better tracking performance. Different waveforms, sidelobe-reduction techniques and measurement-extraction schemes are explored from the detection-tracking system point of view via the hybrid conditional averaging (HYCA) technique. ^ The second part (chapter 3) is on a modified PDAF based on a Bayes detector. Existing detection systems generally operate using a fixed threshold, optimized to the Neyman-Pearson criterion. An alternative is Bayes detection, in which the threshold varies according to the ratio of prior probabilities. In a recursive target tracker such as the PDAF such priors are available in the form of a predicted location and associated covariance. A new PDAF with a Bayes detector is developed. ^ The third part (chapter 4) deals with the matrix CRLB scaling due to measurements of uncertain origin. In many target tracking situations measurements are of uncertain origin. That is, at each scan a number of measurements are obtained, and it is not known which, if any, of these is target-originated. In several earlier papers the surprising observation was made that the CRLB for the estimation of a fixed parameter vector that characterizes the target motion, for the special case of measurements in the presence of additive white Gaussian noise, is simply a multiple of that for the case with no uncertainty. This is particularly useful as it allows comparison in terms of a scalar. This result is explored to determine how wide the class of such problems is. It includes many non-Gaussian situations. ^