sensor Fusion Architectures for ballistic missile Defense
Donald E. Maurer, Robert W. Schirmer, Michael Kalandros, Joseph S. J. Peri · Johns Hopkins APL technical digest · 2006
he project described in this article is developing sensor fusion techniques, within the framework of bayesian belief networks, that will be feasible for operational missile defense systems. As full bayesian nets may be too computationally intensive for the engagement timeline, we investigated techniques that combine outputs of several small nets operating in parallel. An important application is the handover problem wherein a ground-based radar transmits its track picture to the intercepting missile to be combined with the inter ceptor’s track picture to maximize the likelihood of selecting the true target from among other objects. We discuss net architectures addressing this problem. since Dempstershafer algorithms have also been proposed as a solution, we compare our results with a Dempster-shafer approach to understand the advantages and disadvantages of each.