Use of multiple hypotheses in radar tracking
W. G. Bath · 1992
There has been a great deal of work on the mathematical theory of multiple hypothesis tracking. The published work encompasses 'optimum' solutions in which the entire history of detections is used either to maintain the full multimodal probability distribution of track state or, equivalently, to propagate all possible trial tracks . It also encompasses 'instantaneous' solutions (such as probabilistic data association and its variants) in which only a single track hypothesis is propagated in time and measurements are averaged prior to updating the track. The published techniques are generally evaluated for slow maneuvers (which last for many measurements) and for high probabilities of target detection. These are important cases. However, the radar system designer frequently needs to know the minimum update rate and minimum detection probability for which maneuvers may be tracked without seduction. This topic is covered by the author.