Randomization and super-heuristics in choosing sensor sets for target tracking applications

Surveillance systems tracking multiple targets often do not have the sensing or computational resources to apply all sensors to all targets in the allocated time intervals. Hence, sensor management schemes have recently been proposed to reduce the tracking demands on these systems while minimizing the loss of tracking performance by selecting only enough sensing resources to maintain a desired covariance level for each target. The sensor manager algorithm itself, however, incurs a computational burden and needs to be implemented efficiently. This paper explores the use of randomization and super-heuristics to develop computationally efficient methods for implementing sensor manager algorithms.