High-Energy-First (HEF) Heuristic for Energy-Efficient Target Coverage Problem

TL;DR

This paper introduces a new energy-aware heuristic for the target coverage problem in wireless sensor networks, demonstrating improved solution quality and practical efficiency over existing algorithms through comprehensive empirical evaluation.

Contribution

A unified interpretation of existing algorithms is provided, and a novel greedy heuristic prioritizing residual battery life is proposed, outperforming previous methods.

Findings

The new heuristic outperforms existing algorithms in solution quality.

A naive greedy approach achieves solutions within 10% of optimal.

The problem is NP-complete, motivating heuristic solutions.

Abstract

Target coverage problem in wireless sensor networks is concerned with maximizing the lifetime of the network while continuously monitoring a set of targets. A sensor covers targets which are within the sensing range. For a set of sensors and a set of targets, the sensor-target coverage relationship is assumed to be known. A sensor cover is a set of sensors that covers all the targets. The target coverage problem is to determine a set of sensor covers with maximum aggregated lifetime while constraining the life of each sensor by its initial battery life. The problem is proved to be NP-complete and heuristic algorithms to solve this problem are proposed. In the present study, we give a unified interpretation of earlier algorithms and propose a new and efficient algorithm. We show that all known algorithms are based on a common reasoning though they seem to be derived from different algorithmic paradigms. We also show that though some algorithms guarantee bound on the quality of the solution, this bound is not meaningful and not practical too. Our interpretation provides a better insight to the solution techniques. We propose a new greedy heuristic which prioritizes sensors on residual battery life. We show empirically that the proposed algorithm outperforms all other heuristics in terms of quality of solution. Our experimental study over a large set of randomly generated problem instances also reveals that a very na\"ive greedy approach yields solutions which is reasonably (appx. 10%) close to the actual optimal solutions.