View article

The problem of initiation and tracking of targets in heavy clutter is considered from the standpoint of over-the-horizon radar. Fundamental modelling assumptions are identi ed and a statement of the track initiation problem is given. Current automatic track initiation approaches are surveyed including Colegrove's modi ed probabilistic data association (PDA) lter, integrated probabilistic data association and interactive multiple-model PDA. Extensions to cater for target-clutter merging due to nite measurement resolution and adaptive enhancements are described. The latter includes a description of PDA-based schemes incorporating adaptive waveform selection to improve tracking performance, on-line clutter modelling and the use of signal-to-noise ratio information to assist detection. In addition, we detail the incorporation of non-uniform clutter models in PDA-based trackers. We formulate track initiation as a problem in hidden Markov modelling. Analytical tools are proposed as an aid in performance evaluation. These fall into several categories including performance bounds obtainable through trellis-based analyses, and results obtained by Monte Carlo simulation rendered more e cient through the use of techniques from importance sampling. Promising areas for further study are singled out for the continuation of the research.