A study on the application of different two-objective evolutionary algorithms to the node localization problem in wireless sensor networks

A number of applications of wireless sensor networks require to know the location of the sensor nodes. Typically, however, mainly due to costs and limited capacity of the batteries powering the sensor nodes, only a few nodes of the network, denoted anchor nodes in the literature, are endowed with their exact positions. Thus, given a number of anchor nodes, the problem of estimating the locations of all the nodes of a wireless sensor network has attracted a large interest in the last years. The localization task is based on the estimated distances between pairs of nodes in range of each other and is particularly hard in the most appealing scenario, that is, when the network connectivity is quite low. In a recent paper, we have proposed to tackle the localization problem as a two-objective optimization task with the localization accuracy and the number of connectivity constraints that are not satisfied by the candidate geometry as the two objectives. In this paper, we aim to evaluate the behavior of five state-of-the-art multi-objective evolutionary algorithms (MOEAs) in solving the localization problem on different network topologies. We show that one of these MOEAs, namely PAES, statistically outperforms the others in terms of localization error.