Abstract of Talk
An Evolutionary Approach to Optimal Path Planning Problem in Sensor Networks
Abstract: Path planning is an important task in sensor networks and robotics and has received considerable attention in the past. This paper describes an evolutionary approach to find optimal path for a mobile node in a grid of stationary sensors. The aim of the work is to study the behavior of an evolutionary algorithm that will discover the best possible path within the required constraints. We assume that sensors are uniformly deployed in a predefined deployment area. A mobile node walks through the area, communicating with the sensors in the range at regular intervals while keeping a bound on maximum distance traveled. We then compare our result using Cramer Rao Bound (CRB) for unbiased evaluation. Multiple genetic operators including mutation, splicing, selection and cross-over are used to create new paths which are evaluated for optimality. The paper also introduces some modified operators including removing an area from path, adding a new area to path, greedy shortes t path and others. Extensive simulations are performed in order to evaluate the methodology. If the new generation contains a better path, it is saved and used in subsequent generations while discarding the least optimal path. However, if no better solution is found, the entire generation goes through the process again. The process repeats until the improvements with each generation become insignificant for a considerable period of time thus resulting in an optimal path.