The Brueckner network: an immobile sorting swarm

In many industrial applications, the dynamic control of queuing and routing presents difficult challenges. We describe a novel ant colony control system for a multiobjective sorting problem using an Emergent Sorting Network (ESN) designed by Sven Brueckner. Here, an immobile population of extremely simple agents reside at fixed vertices of a network, passing parts through the network, and as a result sorting a stream of colored parts. We explore effects of network size, and the effect of task difficulty (number of colors sorted) on timing and sorting performance. We demonstrate an unexpected regime shift in the swarm's collective behavior caused by network filling effects, and show evidence that this effect is due to the creation of ad hoc buffer regions: transient task specialties arising among the homogeneous agents. Categories and Subject Descriptors I.211 [Artificial Intelligence]: Distributed Artifical Intelligence--multiagent systems, coherence and coordination General Term...

• Type: Conference
• Year: 2006
• Where: GECCO
• Authors: William A. Tozier, Michael R. Chesher, Tejinderpal S. Devgan
• Keywords: Emergent Sorting Network, GECCO 2006, Multiobjective Sorting Problem, Optimization, Sorting