A gravitational search algorithm for solving the relay node placement problem in wireless sensor networks

Summary Nowadays, the use of wireless sensor networks (WSNs) is increasing in many fields of application, such as industrial monitoring, home automation, and intensive agriculture. However, this technology presents an important shortcoming, which has not been solved yet: the energy efficiency. This factor involves a critical economic disadvantage, affecting quality of service. This situation led us to tackle the relay node placement problem, that is, the addition of relay nodes to traditional WSNs as a way to optimize such networks, assuming two important factors: energy consumption and average coverage. To achieve this goal, three multi‐objective (MO) evolutionary algorithms are considered (non‐dominated sorting genetic algoritm II, strength Pareto evolutionary algorithm 2, and MO gravitational search algorithm), assuming a freely available data set and different stop criteria to analyze the behavior of the algorithms. All the results obtained are studied through a widely accepted statistical methodology and two MO metrics (hypervolume and set coverage), concluding that the novel swarm intelligence algorithm MO gravitational search algorithm provides the best performance on average. Moreover, we study the advantages provided by the addition of relay nodes to traditional WSNs. Finally, we compare our proposal with another author approach, assuming a heuristic. Copyright © 2015 John Wiley & Sons, Ltd. This paper deals with the addition of relay nodes to traditional wireless sensor networks, assuming two relevant factors: energy consumption and average coverage. To this end, we consider three multi‐objective algorithms from evolutionary computation, the two standard non‐dominated sorting genetic algorithm II and strength Pareto evolutionary algorithm 2 and a novel multi‐objective approach of the swarm intelligence multi‐objective gravitational search algorithm. The metaheuristics are applied to solve a freely available data set, concluding that the new multi‐objective gravitational search algorithm provides the best performance on average.