An Intelligent, Multi-Agent Environment for Concurrent and Collaborative Configuration of Personal Computers

Abstract

Although Concurrent and Collaborative Engineering (CCE) has enjoyed widespread acceptance in industry, many implementation problems remain. With the advent of more powerful artificial intelligence techniques, CCE can be further improved. This paper demonstrates how intelligent software agents may be deployed to facilitate concurrent, collaborative engineering. A system architecture, Java Agent Alive!, is presented as a multi-agent environment. A case study of configuring a personal computer (PC) from its processor, memory and hard disk drive is discussed to highlight the power of software agents in negotiating for the PC configuration with the best price and performance. A software agent is created and assigned to each of the PC components. These agents attend two levels of agent conferences, viz. the bidding conference and the PC component vendor's conference. At both conferences, each agent strives to offer components with the best performance and the lowest price. The agents were ascribed artificial intelligence through the Java Expert System Shell (JESS). At the end of the negotiations, five PC configurations were finalised that met the expectations of the user, who is informed of the outcome via e-mail. The strengths and limitations of the system architecture and the domain application of PC assembly, as well as means to enhance security, are also discussed. Some recommendations to further improve the limitations of Java Agent Alive! and the PC Assembly application are made.