Implementation of a JPDAF tracker for an electronically scanned radar

Abstract

Reliable tracking in a high clutter, dense target environment requires the utilization of a flexible tracking system capable of solving difficult track associations with minimized computational burden. To study this problem we have implemented, through simulation, a Joint Probabilistic Data Association Filter with an Electronically Scanned Array. This application is capable of handling large numbers of targets illuminated at arbitrarily spaced time intervals. We have developed efficient techniques for the implementation of the JPDAF algorithms, and have begun studying optimization techniques for data utilization and shared workload in order to realize the full potential of the ESAIJPDAF symbiosis. Since the actual JPDAF is not capable of track initiation, a separate adaptive, gated, line-fitting algorithm has been developed for this purpose.

Multi-object filtering with stochastic populations

“…In the situation illustrated in Figure1, for example, (φ, φ, φ, z 3 ) and (φ, z 1 , φ, φ) are compatible, but (φ, φ, φ, z 3 ) and (φ, z 1 , φ, z 3 ) are not since they share measurement z 3 5. The proof is given in Section VII-D.…”

“…However, not every subset of I • t forms a valid characterization of the individuals X • t , since any two individuals whose observation paths are incompatible, i.e., who share a non-empty observation 4 , may not exist simultaneously without violating Assumption (M7). The subsets of targets in I • t made of pairwise-compatible targets are called hypotheses and form a set H t maintained by the DISP filter 5 . Following Eq.…”

“…They rely on an intuitive approach and heuristics in order to represent the uncertainty on the number of targets in the population of interest, including ad-hoc mechanisms for track creation and deletion. Examples of exploitations of these track-based approaches in sensor scheduling problems can be found in [5], [6], or in [7, chap. 14].…”