Self-Organizing Topology Control based on Reaction-Diffusion Model for Data Gathering in Wireless Sensor Networks

A wireless sensor network (WSN) consists of a large number of small, low-cost, and fragile sensing devices, called sensor nodes, communicating with each other through unreliable and unstable wireless communication. Therefore, control mechanisms for WSNs must be scalable to the number of nodes, adaptive to dynamically changing communication environment, and robust to failures. In addition, due to difficulty in managing sensor nodes in a centralized manner for limited energy and network resources, control mechanisms must be fully distributed and selforganizing. Considering these requirements, it is not surprising that much attention has been paid to application of biological mechanisms in these days, ranging from signal transduction network to swarm intelligence. A variety of biologically-inspired control mechanisms have been proposed for routing, clustering, scheduling, and topology control of WSNs. In this thesis, we focus on topology control for energy efficient and low delay data gathering, which certainly is one of the most important issues of WSNs. Independently of applications, unless all sensor nodes operate with electric power supply, energy efficiency is a strong requirement for control mechanisms. In addition, it is necessary for sensor nodes to deliver up-to-date sensor information to a sink node of sensor data, e.g. a gateway server, as fast as possible for prompt reaction to detected events and conditions. The amount of consumed energy and the delay in data gathering heavily depend on the topology of a WSN. A variety of topologies have been considered for energy-efficient and low-delay data gathering in a WSN, such as, direct transmission from all sensor nodes to a sink node, a tree topology rooted at a sink node, and multi-tier topology organized by clustering sensor nodes into groups. However, there has never been comprehensive and complete comparison among them so far, to the best