An efficient hybrid multi-objective memetic algorithm for the frequency assignment problem. (January 2020)

• Record Type: Journal Article
• Authors: Kiouche, Abd Errahmane; Bessedik, Malika; Benbouzid-SiTayeb, Fatima; Keddar, Mohamed Reda
• Is Part Of: Engineering applications of artificial intelligence. Volume 87(2020)
• Publication Date: 2020-01
• Subjects: Multi-objective genetic algorithm -- Memetic algorithm -- Local search -- Clonal selection -- Receptor editing -- Frequency assignment problem; Engineering -- Data processing -- Periodicals; Artificial intelligence -- Periodicals; Expert systems (Computer science) -- Periodicals; Ingénierie -- Informatique -- Périodiques; Intelligence artificielle -- Périodiques; Systèmes experts (Informatique) -- Périodiques; Artificial intelligence; Engineering -- Data processing; Expert systems (Computer science); Periodicals; 620.00285
• DOI: 10.1016/j.engappai.2019.103265
• Languages: English
• ISSNs: 0952-1976

Abstract

Abstract: This paper investigates hybridization of multi-objective memetic algorithm and artificial immune system (AIS) for the Frequency Assignment Problem (FAP) in cellular mobile networks. The considered objectives to minimize are the total interference, the maximal interference, and the number of used frequencies. The proposed approach integrates FAP-specific local search into the evolutionary process to overcome the shortcoming of the multi-objective genetic algorithm, as well as clonal selection and receptor editing, which aims to improve the algorithm exploration and exploitation abilities. Based on the hypervolume metric, the proposed hybrid multi-objective algorithm produces high quality solutions as proved by the tests performed over COST259 instances and corroborated by the comparisons with the most frequently referred algorithms in the related literature. Furthermore, the effect and the behaviour of the main parameters of our algorithm and the interaction between them are analysed using the Design of Experiment (DOE). Graphical abstract: Highlights: The proposed approach is a multi-objective memetic algorithm that integrates immune operators in its evolutionary process to improve the algorithm exploration and exploitation abilities. As application, we deal with the Frequency Assignment Problem (FAP) in cellular networks considering three objectives to minimize: the total interference, the maximal interference, and the number of used frequencies. The proposed approach integrates FAP-specific local search into the evolutionary process as well as clonal selection and receptor editing inherited from AIS. The behaviour of the main algorithm factors and the interaction between them are analysed using the ANOVA statistical test. The performances of the newly proposed algorithm are measured in terms of hypervolume on COST259 instances.