A game theoretic model for management of mobile sensors

Abstract

This paper suggests a framework for multi- sensor multi-target tracking with mobile sensors. Sensors negotiate over which targets to track (possibly sharing tar- gets to benejit from data fusion technology) using a game theory based algorithm. Sensors' preferences over negotia- tion offers are articulated with individual utility functions which encompass both information gain and directional derivative. An approach to consider terrain effects on mo- bile sensors is also explained. Simulation results show that the negotiation algorithm has interesting advantages com- pared to a greedy algorithm that seeks to optimise informa- tion gain without consideration to derivatives. We notice that the negotiation procedure forces sensors to share tar- gets, while improving robustness to sensor failure. Sensors also tend to proactively reconsider their target assignments for long-term improved information gain.