Influence of communication graph structures on pheromone-based approaches in the context of a partitioning task problem

  • Authors:

  • Thomas Kemmerich

  • Affiliations:

  • International Graduate School of Dynamic Intelligent Systems, Knowledge Based Systems, University of Paderborn, Germany

  • Venue:
  • COIN'09 Proceedings of the 5th international conference on Coordination, organizations, institutions, and norms in agent systems
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper introduces and formalizes a multi-objective agent coordination problem, called General Online Partitioning Problem (GOPP). The goal is to find a cost-optimal, distance minimizing, and uniform partitioning of an agent set to a set of targets in a 2-dimensional world. Agents build a communication graph based on local neighborhood relations. We propose a message-based Ant Colony Optimization (ACO) algorithm that disposes target-specific pheromones in this communication graph to solve the GOPP. It is analyzed why different typed and newly arriving pheromone traces are unable to grow into once established pheromone structures. Furthermore, an example is presented in which pheromone-based approaches working on communication graphs are unable to find optimal solutions. We present experimental results and compare the new approach to existing ones. Besides the proved nonoptimality of our novel approach, the evaluation shows that the algorithm produces high quality solutions on average.