부산대학교 도서관

Tracking with bistatic radar measurements is a challenging problem due to the nonlinear relationship between the radar measurements and the Cartesian coordinates, especially for long distances. This nonlinearity leads, for 3-D bistatic radar, to a nonellipsoidal measurement uncertainty region in Cartesian coordinates, similar to a thin contact lens, that causes consistency problems for a tracking filter. The recently developed conversion of the bistatic radar measurements into Cartesian coordinates enables to maintain consistency by using a converted measurement Kalman filter. However, such a filter suffers from a poor bistatic range accuracy, limiting the multitarget tracking performance in a dense environment of targets or clutter. A solution is suggested by using a filter in the measurement coordinate system and converting its results into Cartesian coordinates. Consistent Cartesian estimation is obtained together with significantly improved filtered bistatic range accuracy. The latter is important in data association, resulting in a measurement-to-track association gates that are over an order of magnitude smaller compared to the Cartesian filters. It is shown that if the data association is performed in sensor coordinates, the (Cartesian) volume of the contact lens association region is over 20 times smaller than if it is performed in Cartesian coordinates because the ellipsoid in the latter case is extremely conservative.