A Novel Parallel and Distributed Algorithm Based on Crossbar Composite Spring Nets

Abstract

This paper presents a spring net approach to multi-agent systems (MAS) and computer networks (CN), which transforms the parallel optimization problem in MAS and CN into the evolutionary deformation process of a class of crossbar composite spring nets (CCSN). The proposed spring net is essentially different from the elastic net developed by Durbin and Willshaw and subsequently extended byClaudia, Burr, Stone, et al. The mathematical and physical models of the CCSN used for parallel optimization in MAS and CN are discussed. The CCSN approach outperforms over the elastic net and other optimization methods currently used in MAS and CN in terms of the ability to deal with a variety of complicated social interactions and autonomous behaviors that may occur in MAS and CN.