Scalability of evolving networks of trading agents

Article type: Research Article

Authors: Wang, Juan^{a; *} | De Wilde, Philippe^b

Affiliations: [a] College of Information Engineering, Shenzhen University, Shenzhen 518060, People's Republic of China | [b] School of Mathematical and Computer Sciences, Heriot Watt University, Edinburgh EH14 4AS, United Kingdom

Correspondence: [*] Corresponding author. E-mail: juanwang@szu.edu.cn

Abstract: Scalability is a vital issue for multi-agent systems. In this paper, we present a trading model of a real-world scenario. It is modelled as networks of interacting agents that can evolve automatically over time, based on behaviors of agents and system configurations. In this model, the agents have resources, and receive a payoff for selling the resources. Agents are regarded as players in a game. We describe an artigicial multi-agent ecosystem where the Darwin's theory of natural selection were applied. as evolutionary dynamics on the population of agents. Through discrete event simulations, the scalability of this model under various system configurations is analyzed and scaling laws are derived.

Keywords: Multi-agent systems, scalability, ecosystem, decision making

DOI: 10.3233/IDT-2007-1403

Journal: Intelligent Decision Technologies, vol. 1, no. 4, pp. 197-204, 2007