McEneaney, William (Univ. of California, San Diego), Singh, Rajdeep (Lockheed Martin)

Deception-Enabled Control Theory for Stochastic Games with One-Sided Imperfect Information

Scheduled for presentation during the Regular Session "Stochastic systems" (FrP08), Friday, July 28, 2006, 15:45−16:10, Room I

17th International Symposium on Mathematical Theory of Networks and Systems, July 24-28, 2006, Kyoto, Japan

This information is tentative and subject to change. Compiled on April 24, 2024

Abstract

The class of discrete, imperfect information, stochastic games, where one player has perfect information, and the second player has only observation-based information, has recently become of interest. A deception-robust controller was previously obtained for the player with imperfect information. Also, a deception-enabled controller was obtained for the player with perfect information, under the assumption that that player also has a (perfect) model of the opponent's controller. Here, it is first demonstrated that knowledge of the opponent's controller is not necessary to the optimality of a potentially deceptive control for the full information player. Then the appropriate game information states and value definitions are given, and the optimal deception-enabled controller for the full-information player is obtained.