Self spreading nodes using potential games and genetic algorithms

Dynamically changing topology, lack of centralized authority, nodes´ selfishness, and unknown deployment terrain present difficulties in self spreading of nodes in mobile ad hoc networks (MANETs). In these settings, game theory (GT) and genetic algorithms (GAs) are promising tools to improve the area coverage with reduced computational overhead. We present our node spreading potential game (NSPG) using a GA for MANET nodes to position themselves in an unknown terrain with obstacles. NSPG is a distributed and scalable game participated by nodes autonomously. The decisions about node movements are solely based on localized data where the best next location to move is selected by a GA. Our approach is suitable for real-life MANET applications since it requires only a limited synchronization among players´ closest neighbors without a priori knowledge of an environment. We prove that NSPG converges to a stable state. Simulation results show that NSPG performs well with respect to convergence speed and adaptability to adverse terrain conditions such us arbitrarily placed obstacles.