Reaction-Diffusion Based Topology Self-Organization for Periodic Data Gathering in Wireless Sensor Networks

Considering characteristics of wireless sensor networks,control mechanisms must be scalable, adaptive, robust, fully distributed, and self-organizing. In this paper, we focus on topology control for periodic data gathering, which certainly is one of typical applications of WSNs. We propose a novel mechanism based on a biological self-organization mechanism, that is, a reaction-diffusion model, to organize the best topology in a self-organizing and autonomous way. A reaction-diffusion model is a mathematical model for pattern generation on the surface of body of fishes and mammals. Nodes generate spatially distributed spot patterns through mutual interaction among neighboring nodes. The node which has a peak of activator concentration is elected as cluster head, and other nodes send their data following the gradient of activator concentration to the cluster head. Through simulation experiments, it is shown that organized topology accomplishes as small energy consumption and delay as the best topology optimally derived.