Title: Balanced centroids (BC) k-means clustering algorithm to transform MTSP to TSP

Authors: P. Venkateswara Reddy, A.C.S. Kumar, M.S. Bhat, R. Dhanalakshmi, P. Parthiban

Addresses: Department of Mechanical Engineering, Jawaharlal Nehru Techological University Hyderabad, Kukatpally, Hyderabad – 500085, Andra Pradesh, India. ' Department of Mechanical Engineering, Jawaharlal Nehru Techological University Hyderabad, Kukatpally, Hyderabad – 500085, Andra Pradesh, India. ' School of Management Studies, Jawaharlal Nehru Techological University Hyderabad, Kukatpally, Hyderabad – 500085, Andra Pradesh, India. ' Cisco Systems India Pvt. Ltd., No. 148 Asv Suntech Park, Oggium, Old Mahabalipuram Road, Thoraipakkam, Chennai – 600096, Tamil Nadu, India. ' Department of Production Engineering, National Institute of Technology, Tiruchirappalli – 620 015, India

Abstract: Solving a multiple travelling salesperson (mTSP) is more difficult than solving a travelling salesperson. Although the TSP has received a great deal of attention, the research on the mTSP is limited. In the mTSP, the N cities must be partitioned into m tours, with each tour resulting in a TSP for one salesperson. The purpose of this paper is to review in brief the existing literature on the mTSP and to transform the most |NP complete| problem to an ordinary TSP. We propose a technique to transform mTSP to TSP using BC k-means clustering. The methodology we mentioned here will be suitable for any type of Euclidean mTSP. The result shows that the number of iterations (NI) required to get the solution are less in BC k-means clustering in most of the cases.

Keywords: multiple travelling salesperson; k-means; clustering algorithms; balanced centroids; tours; Euclidean mTSP; iterations; logistics management; economics.

DOI: 10.1504/IJLEG.2010.036299

International Journal of Logistics Economics and Globalisation, 2010 Vol.2 No.3, pp.187 - 197

Published online: 01 Nov 2010 *

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