Energy Restoration in a Linear Sensor Network

Abstract

The coverage provided by a sensor network degrades over time as the batteries powering the sensors become exhausted. A common approach to energy restoration in sensor networks powered by batteries is to use a robot acting as a mobile battery charger/changer. The goal is to constantly minimize the number of coverage holes and their duration. In this paper, we focus on decentralized on-line strategies for energy restoration by a robot in linear sensor networks, i.e. whose topology is modelled as a line. We consider a standard On-Demand strategy, where a sensor in need of charge sends a request in the direction of the robot, and the robot moves to serve the requests as they arrive. We examine also a simpler variant of this strategy, Straight, in which the direction of movement of the robot along the line cannot be changed until it reaches the end of the line. We finally consider the simplest possible on-line strategy, Blind, where no requests are sent and the robot automatically and continuously moves from one end of the line to the other, servicing any sensor found needing recharging. We experimentally study the efficiency of these strategies, and we make the counter-intuitive discovery that the simpler the strategy, the better its efficiency. In particular, Blind which does not require any communication nor memory nor computation, is at least as efficient as the other two. We also provide strong analytical support to these experimental findings. In fact we prove that, starting with initially empty batteries, the Blind strategy has better coverage performance that the other two strategies for almost all network sizes. Indeed for some network sizes no other strategy, even if centralized and offline, can do better.