A hybrid genetic algorithm for generating optimal synthetic aperture radar target servicing strategies

The purpose of this research was to develop a software tool for generating optimal target servicing strategies for imaging fixed ground targets with a spaceborne SAR. Given a list of targets and their corresponding geographic locations and relative priorities, this tool generates a target servicing strategy that maximizes the overall collection utility based on the number of targets successfully imaged weighted by their relative priorities. This tool is specifically designed to maximize sensor utility in the case of a target-rich environment. For small numbers of targets, a target servicing strategy is unnecessary, and the targets may be imaged in any order without paying any particular attention to geographic proximity or target priority. However, for large, geographically diverse target decks, the order in which targets are serviced is of great importance. The target servicing problem is shown to be of the class NP-hard, and thus cannot be solved to optimality in polynomial time. Therefore, global search techniques such as genetic algorithms are called for. A unique hybrid algorithm that combines genetic algorithms with simulated annealing has been developed to generate optimized target servicing strategies. The performance of this hybrid algorithm was compared against that of three different greedy algorithms in a series of 20 test cases. Preliminary results indicate consistent performance improvements over greedy algorithms for target-rich environments. Over the course of 20 trials, the hybrid optimizing algorithm produced weighted collection scores that were on average 10% higher than the best greedy algorithm.