Advanced Algorithm Development

"Information Extraction & Transport (IET), Inc., proposes a new approach to multiple-object tracking and discrimination: Hybrid Stochastic Clustering (HSC) for ballistic missile defense (BMD) exo- and endo-atmospheric object tracking and discrimination inhigh object density, time-varying sensor resolution, and object spawning environments. The hybrid stochastic clustering concept is a direct extension of the clustering method on which Reid's multiple hypothesis tracking (MHT) algorithm is based, and wasdeveloped as part of the Project Hercules FY01 efforts. As an MDA Phase I SBIR effort, we propose to conduct feasibility studies and concept-proving demonstrations using prototype software. A stochastic cluster is an a posteriori independent component ofa probability distribution of the finite random set of the states of the objects being tracked. A stochastic clustering method maintains the probability distribution in terms of as many clusters as possible so that the overall efficiency of computationresource usage is maximized. We propose a distributed hybrid stochastic clustering approach using a different tracking and discrimination algorithm for each stochastic cluster based on its complexity. The set of applicable algorithms includes multiplehypothesis tracking, Janossy density function (JDF), and Poisson point process approximation (P3A) algorithms, all based on the theory of random finite sets. The result of this Phase I ef