Title: Analysis of mesh router placement in wireless mesh networks using Friedman test considering different meta-heuristics

Authors: Tetsuya Oda; Yi Liu; Shinji Sakamoto; Donald Elmazi; Leonard Barolli; Fatos Xhafa

Addresses: Graduate School of Engineering, Fukuoka Institute of Technology (FIT), 3-30-1 Wajiro-Higashi, Higashi-Ku, Fukuoka 811-0295, Japan ' Graduate School of Engineering, Fukuoka Institute of Technology (FIT), 3-30-1 Wajiro-Higashi, Higashi-Ku, Fukuoka 811-0295, Japan ' Graduate School of Engineering, Fukuoka Institute of Technology (FIT), 3-30-1 Wajiro-Higashi, Higashi-Ku, Fukuoka 811-0295, Japan ' Faculty of Information Technologies, Polytechnic University of Tirana, Bul. 'Dëshmorët e Kombit', 'Mother Theresa' Square, Nr. 4, Tirana, Albania ' Department of Information and Communication Engineering, Fukuoka Institute of Technology (FIT), 3-30-1 Wajiro-Higashi, Higashi-Ku, Fukuoka 811-0295, Japan ' Department of Languages and Informatics Systems, Technical University of Catalonia, C/Jordi Girona 1-3, 08034 Barcelona, Spain

Abstract: In this paper, we deal with connectivity and coverage problem in wireless mesh networks (WMNs). We used Friedman test to compare genetic algorithm (GA), tabu search (TS), hill climbing (HC) and simulated annealing (SA). We found out that GA, TS, HC and SA have differences in their performance. Then, we used the implemented systems WMN-GA, WMN-TS, WMN-HC and WMN-SA to evaluate and compare the performance of the systems for different distributions of mesh clients in terms of size of giant component (SGC) and number of covered mesh clients (NCMC). The simulation results show that for uniform distribution the WMN-HC and WMN-SA perform better than WMNGA and WMN-TS. However, for small radius of communication distance, the SGC of WMN-TS is better than other systems. For normal distribution, for big radius of communication distance, the WMN-GA has the best performance. For exponential distribution, the WMN-HC and WMN-SA perform better than WMN-GA for all communication distances. For Weibull distribution, the WMN-TS has a good performance for small radius of communication distance, but for big radius of communication distances the WMN-GA, WMN-HC and WMN-SA perform better.

Keywords: wireless mesh networks; WMNs; Friedman test; genetic algorithms; tabu search; hill climbing; simulated annealing; giant component; user coverage; mesh router placement; metaheuristics; network connectivity; network coverage; simulation.

DOI: 10.1504/IJCNDS.2015.070289

International Journal of Communication Networks and Distributed Systems, 2015 Vol.15 No.1, pp.84 - 106

Received: 19 Feb 2015
Accepted: 07 Mar 2015

Published online: 01 Jul 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article