Maximum likelihood approach to HF radar performance characterization

As part of NURC maritime surveillance research activities, there is a need to characterize the performance of HF radar vessel-detection processing, in support of subsequent automatic tracking as well as fusion with other available sensor data. In particular, HF radar detection and localization performance for a given contact SNR threshold must be determined. This paper examines two approaches to do so: one based on the maximum-likelihood criterion, the other using a localization threshold on target-originated contacts. The differences in and merits of the two approach are discussed with simulated and real maritime HF radar data. We introduce a multi-stage track-extract-track approach to HF radar tracking that is effective in a heavily-cluttered environment. Finally, we introduce scalar information metric for sensor characterization and discuss its relevance to sensor selection and detection threshold determination.