Cooperating intelligent agents for distributed satellite systems

摘要

An approach has been developed that models individual satellites as intelligent agents and allows them to cooperate in system-level problem solving. ^As an initial test, agents were programmed to solve the well-understood n-queens problem (place n queens on a nxn chessboard such that none can attack the others). ^The intelligent agent approach produced encouraging results compared to conventional approaches. ^Intelligent agents solved a 34x34 board in 77 moves compared to over two billion moves required by conventional means. ^This research is being extended to the (more challenging) minimum queens problem (find the fewest queens that provide full coverage on an nxn board). ^This better reflects one problem faced by TechSat 21 (find the fewest number of satellites that provide coverage over a specified area). ^The results will be evaluated for application to constellation design, cluster pattern management, and several mission aspects including cooperative signal processing, target detection, and target tracking. ^In addition, intelligent agents may be applied to selected nondeterministic polynomial complete (NP-Complete) problems to evaluate their effectiveness on tasks such as scheduling. ^(Author)