Black widow optimization for reducing the target uncertainties in localization wireless sensor networks
Rubén Ferrero-Guillén,José-Manuel Alija-Pérez,Alberto Martínez-Gutiérrez,Rubén Álvarez,Paula Verde,Javier Díez-González
DOI: https://doi.org/10.1093/jigpal/jzae032
2024-03-24
Abstract:Abstract Localization Wireless Sensor Networks (WSN) represent a research topic with increasing interest due to their numerous applications. However, the viability of these systems is compromised by the attained localization uncertainties once implemented, since the network performance is highly dependent on the sensors location. The Node Location Problem (NLP) aims to obtain the optimal distribution of sensors for a particular environment, a problem already categorized as NP-Hard. Furthermore, localization WSN usually perform a sensor selection for determining which nodes are to be utilized for maximizing the achieved accuracy. This problem, defined as the Sensor Selection Problem (SSP), has also been categorized as NP-Hard. While different metaheuristics have been proposed for attaining a near optimal solution in both problems, no approach has considered the two problems simultaneously, thus resulting in suboptimal solutions since the SSP is biased by the actual node distribution once deployed. In this paper, a combined approach of both problems simultaneously is proposed, thus considering the SSP within the NLP. Furthermore, a novel metaheuristic combining the Black Widow Optimization (BWO) algorithm and the Variable Neighbourhood Descent Chains (VND-Chains) local search, denominated as BWO-VND-Chains, is particularly devised for the first time in the author’s best knowledge for the NLP, resulting in a more efficient and robust optimization technique. Finally, a comparison of different metaheuristic algorithms is proposed over an actual urban scenario, considering different sensor selection criteria in order to attain the best methodology and selection technique. Results show that the newly devised algorithm with the SSP criteria optimization achieves mean localization uncertainties up to 19.66 % lower than traditional methodologies.
mathematics, applied,logic