Track Segment Association with a Distributed Field of Sensors

A performance prediction procedure is developed for geographical localization of targets traversing through a distributed field of sensors. The analysis is based on a tracking scheme where targets are detected and tracked locally by as few as two sensors. Target estimates from adjacent sensor sets are extrapolated to an estimated time of intersection, and the hypothesis is made that the two segments have a common origin (i.e., that the two tracks have emanated from a single target). Through a statistical test based on a maximum likelihood ratio, this hypothesis is either accepted or rejected. It is then shown that associated target track segments can be reestimated using the combined set of measurements. The theoretical results have been corroborated by applying a maximum likelihood tracking algorithm to simulated data and observing the results for a series of Monte Carlo simulations. The actual and expected statistics on the probability of correctly associating target segments, along with the localization ellipses, are displayed.