Impact of Migration Topologies on Performance of Teams of Agents

The chapter sums up the impact of some parameters defining the process of migration of data between asynchronous team of agents (A-Team) working in parallel in the architecture designed for solving difficult combinatorial optimization problems. A-Teams cooperate through exchange of intermediary computation results. The process of forwarding a result from one A-Team to another is called migration. Several known migration models have been compared, with different topologies and frequencies of migration. Also, an original model of communication has been proposed. The model, called Randomized, has no predefined migration topology, each A-Team sends data to another A-Team that is chosen at random. In this model migration is non-periodic and triggered only after an A-Team failed to improve its best current solution within a predefined time. All considered models, differing in migration topologies and frequencies, have been tested on instances of the Euclidean planar traveling salesman problem. The proposed model, Randomized, outperforms all other models under investigation, producing significantly better results.