Abstract: Ant Colony Optimization (ACO) is a computer emergence model to solve problems by swarm intelligence. The aim of this paper is to provide ant colonies with a global memory structure (GMS) so that the time needed to resolve a problem decreases drastically. In the past few years ACO have become a strong alternative to classic algorithms. However, the biggest disadvantage of ACO is the lack of structures which can provide every individual of the population with simple memory mechanisms. The GMS presented in this paper is applied to cleaner robots that must search a gallery looking for piles of marks and clean them. It is based on the variation of the common map by deleting all the superfluous nodes which appears as the resolution of the problem progresses. A node is considered as superfluous when it is useless for every ant and it delays every route of the colony. All the robots must share and update the GMS when they find any superfluous node. The extra process charge needed for executing the memory in parallel to ant’s activities is absorbed by the time saved in the resolution of the problem which depends on the characteristics of the map and its abstracted graph.

Keywords: Ant Colony Optimization, Swarm Intelligence, Memory.

ACM Classification Keywords: I.6. Simulation and Modelling, B.7.1 Advanced Technologies, I.2.8 Problem Solving, I.2.11 Distributed Artificial Intelligence

Link:

GLOBAL MEMORY STRUCTURE FOR ANT COLONY OPTIMIZATION ALGORITHMS

Ángel Goñi, Paula Cordero

http://foibg.com/ibs_isc/ibs-11/ibs-11-p16.pdf