Game theoretical model for information transmission in structure‐free wireless sensor networks

Abstract

We consider, as a game, the competition between sensors' transmissions in a structure-free wireless sensor network. They contemplate two nodes at the same level strategically optimising their decisions over a finite set of strategies. Specifically, they consider two strategies are available to each node. Nodes can choose to transmit to keep the network updated, but at the risk of not sensing during the transmission time and the potential of sending repeated information if other nodes have already transmitted that information. Nodes can choose to sense the environment to detect new events, but this strategy leaves the network with outdated information if all nodes adopt it. As a result, nodes have to strategically make their decisions depending on the nodes' parameters, the importance of information, and the locations of the head nodes. They tackle this decision-making problem using two game theoretic models: a non-cooperative two-player game model and a potential game model. They derive the Nash equilibria (NE) and highlight their existence conditions. Finally, they use learning algorithms that use local information at each node to reach the NE. These algorithms are the fictitious play (FP) algorithm and a modified FP that is inspired by the cumulative proportional reinforcement algorithm.