Date of Award
Master of Science
Department of Operational Sciences
Darryl K. Ahner, PhD.
With the advancing capabilities of Intelligence, Surveillance, and Reconnaissance (ISR) assets and sensors, effective utilization of these resources continues to pose a challenge to military decision makers. The methodology developed explores allocation of ISR assets while balancing detection of new targets versus surveillance of already detected targets using entropy as a Measure of Effectiveness (MOE). Scenarios with an unknown number of static and moving targets in a bounded geographical region are considered. A baseline model was built to examine four different search algorithms: random, raster, greedy, and a rollout algorithm based on dynamic programming. A space-filling Nearly Orthogonal Latin Hypercube experimental design was applied to generate data to examine four MOEs: step entropy, average entropy, number of targets found, and time steps to completion. Based on statistical analysis and time series plots, the rollout algorithm's performance dominated others algorithms considered. In addition to minimizing uncertainty in the first 100 time steps of the run, the rollout algorithm also produced the highest number of targets found within the fixed time step scenario, and, for the exhaustive target detection scenario, discovered all of the targets within the region in less time steps. Based on these results, the rollout algorithm provides superior performance in the allocation of ISR assets while balancing detection of new targets versus surveillance of already detected targets.
DTIC Accession Number
Dismukes, Tamilyn S., "Surveillance Versus Reconnaissance: An Entropy Based Model" (2012). Theses and Dissertations. 1205.