Distributed Multi-Ant Colony System Algorithm using Raspberry Pi Cluster for Travelling Salesman Problem

Mustafa Muwafak  Alobaedy; Ali A.  Khalaf; Yousef  Fazea

doi:10.24996/ijs.2022.63.9.35

Authors

Mustafa Muwafak Alobaedy School of Information & Communication Technology, HELP University, Kuala Lumpur, Malaysia
Ali A. Khalaf Computer Science Department, College of Science, University of Baghdad, Baghdad, Iraq https://orcid.org/0000-0002-3631-5341
Dr. Yousef Department of Computer & Information Technology, Marshall University, 1 John Marshall Drive, Huntington, WV 25755, USA https://orcid.org/0000-0003-3544-2434

DOI:

https://doi.org/10.24996/ijs.2022.63.9.35

Keywords:

Metaheuristic Algorithm, Single-Board Computing Cluster, Combinatorial Optimization problem, Distributed Computing, Distributed Algorithm

Abstract

The traveling salesman problem is addressed in this paper by introducing a distributed multi-ant colony algorithm that is implemented on a Raspberry Pi cluster. The implementation of a master and eight workers, each running on Raspberry Pi nodes, is the central component of this novel technique. Each worker is responsible for managing their own colony of ants, while the master coordinates communications among workers’ nodes and assesses the most optimal approach. To put the newly built cluster through its paces, several datasets of traveling salesman problem are used to test the created cluster. The findings of the experiment indicate that a single board computer cluster, which makes use of multi-ant colony algorithm, is a viable alternative for distributed computing. This approach's extensibility options are extensively discussed as well.