A multi-objective genetic algorithm-based adaptive weighted clustering protocol in VANET

• Published in: 2015 IEEE Congress on Evolutionary Computation (CEC) pp. 994 - 1002
• Main Authors: Hadded, Mohamed, Zagrouba, Rachid, Laouiti, Anis, Muhlethaler, Paul, Saidane, Leila Azouz
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.05.2015
• Subjects: Ad hoc Networks; Cluster Protocol; Clustering algorithms; MODE; MOPSO; Multi-Objective Optimization; Nominations and elections; NSGA-II; Optimization; Pareto Front; Protocols; Road transportation; VANET; Vehicles; Vehicular ad hoc networks
• ISSN: 1089-778X; 1941-0026
• DOI: 10.1109/CEC.2015.7256998

Vehicular Ad hoc NETworks (VANETs) are a major component recently used in the development of Intelligent Transportation Systems (ITSs). VANETs have a highly dynamic and portioned network topology due to the constant and rapid movement of vehicles. Currently, clustering algorithms are widely used as the control schemes to make VANET topology less dynamic for Medium Access Control (MAC), routing and security protocols. An efficient clustering algorithm must take into account all the necessary information related to node mobility. In this paper, we propose an Adaptive Weighted Clustering Protocol (AWCP), specially designed for vehicular networks, which takes the highway ID, direction of vehicles, position, speed and the number of neighboring vehicles into account in order to enhance the stability of the network topology. However, the multiple control parameters of our AWCP, make parameter tuning a nontrivial problem. In order to optimize the protocol, we define a multi-objective problem whose inputs are the AWCP's parameters and whose objectives are: providing stable cluster structures, maximizing data delivery rate, and reducing the clustering overhead. We address this multi-objective problem with the Non-dominated Sorted Genetic Algorithm version 2 (NSGA-II). We evaluate and compare its performance with other multi-objective optimization techniques: Multi-objective Particle Swarm Optimization (MOPSO) and Multi-objective Differential Evolution (MODE). The experiments reveal that NSGA-II improves the results of MOPSO and MODE in terms of spacing, spread, ratio of non-dominated solutions, and inverse generational distance, which are the performance metrics used for comparison.