Particle Filtering for the Joint Tracking and Identification of Objects

"Toyon Research Corporation proposes to develop a system to track and identify ballistic objects. The system will use radar measurements containing kinematic information (e.g., range, range rate, azimuth) as well as feature information (e.g., HRR profiles,RCS, electrical length). We propose to develop a particle filtering algorithm which directly estimates the joint PDF of an object's state and class type given a scan of radar measurements. The particle filter has many benefits including the ability tohandle nonlinear measurement and dynamic models, non-Gaussian statistics, a mixed continuous/discrete state and constrained dynamic problems. The particle filter's unique attributes allow the maximum extraction of information from the sensor measurementsand transfer of information across the different phases of the missile trajectory. The short time frame for a successful missile engagement demands that an algorithm fully utilize all available sensor measurements to locate and identify potential threats.The particle filter has the potential to achieve this goal. We will implement our algorithm in Phase I and demonstrate its performance in non-real-time operation on a simple problem of interest to the MDA. We will investigate the possibility for real-timeoperation in Phase II. The successful completion of this research will provide the Missile Defense Agency with a system capable of tracking and identifying ballistic objects using h