Solving Travelling Salesman Problem Using an Improved Ant Colony Optimization Algorithm

Authors

  • RUFAI, K. I Computer Science Department, Tai Solarin University of Education, Ijagun, Ogun State, Nigeria
  • USMAN, O. L. Research Centre for Cyber Security, Faculty of Information Science and Technology, University Kebangsaan Malaysia, Bangi, Selangor, Malaysia.
  • OLUSANYA, O. O. Computer Science Department, Tai Solarin University of Education, Ijagun, Ogun State, Nigeria.
  • ADEDEJI, O. B. Computer Science Department, Tai Solarin University of Education, Ijagun, Ogun State, Nigeria.

Keywords:

Travelling Salesman Problem, Ant Colony Optimization, Convergence, Nigria52 dataset, Combinatorial optimization

Abstract

Travelling Salesman Problem (TSP) is a well-known and mostly researched problem in the field of combinatorial
optimization. In this study, an attempt was made to model an improved Ant Colony Optimization (ACO) algorithm
that has a fast convergence speed and very good global optimization ability when solving TSP based on Nigeria52
dataset. The proposed improved ACO algorithm was tested using the MATLAB (R2018a) software, with TSP
conceived as a complete weighted Hamiltonian cycle of 52 cities, and with Abuja as the starting point of the
salesman's tour. The results of an improved ACO simulation show that the algorithm was able to improve
convergence by avoiding local minima with an optimal path length (cost) of 5866.9249 when the value of
exploitation (?) is less than that of exploration (?). The results also considerable improvement compared against
tradition ACO algorithm with 98% precision.

Downloads

Published

2021-09-14