The paper describes what we have learned from implementing a multi-agent architecture used to support enterprise notions and principles for intelligent systems design. In the last couple of years, agent-based management systems have been widely used to monitor and control business design processes. In this paper, multi-agent architecture is proposed and applied to the workflow management system in ASE1 (Advance Semiconductor Electronic) Inc., which is the world’s well-known largest provider of independent semiconductor for manufacturing services in assembly and test. The proposed FIPA-OS autonomous workflow management system uses a workflow co-ordination mechanism and an agent integration mechanism to enable the routine daily jobs error handle. Our vision for extending the FIPA-OS architectural elements to cover the development and implementation of generic web-Centric collaborative applications concludes the paper.
Workflow management systems had a significant impact on organizations in the past decade (Cardoso, Sheth, Miller, Arnold, & Kochut, 2004), and are a promising technology that automates business processes to improve the efficiency and manageability of an enterprise’s daily teamwork (Wang, Wang, & Xu, 2005). How to rapidly and cost-effectively satisfy the various dynamic demands of the market adapt to an unpredictable business environment and increasing rates of new production methods are the major challenges for manufacturing enterprises. Achieving the combination of internal and external flows on heterogeneous platforms in a distributed environment is an important research area (Wang, Shen, & Hao, 2006). ASE Inc. is the world’s largest provider of independent semiconductor manufacturing services in assembly and test. As a global leader geared towards meeting the industry’s ever growing needs for faster, smaller and higher performance chips, ASE Group develops and offers a wide portfolio of technology and solutions, including IC test program design, front-end engineering testing, wafer probe, wafer bump, substrate design and supply, wafer level packaging, flip chip, system-in-package, final testing and design manufacturing services through Universal Scientific Industrial Co. Ltd. In this paper, a distributed multi-agent architecture is proposed and applied to the workflow management system in ASE. The remainder of this paper is organized as follows: in Section 2, we review the related literature in the areas of agent technology, workflow management and the FIPA-OS platform. Section 3 illustrates our autonomous workflow management system and agent communication model using the FIPA-OS framework. In Section 4, we describe what we have learned from implementing a multi-agent application used to support enterprise workflow management system. Finally, we summarize the contributions of this paper and provide suggestions for future work.
In this research, we propose a FIPA-OS based multi-agent autonomous workflow management system for ASE Inc. Using the FIPA-OS framework to establish communications in distributed environments, this agent platform successfully increases interoperability and information transparency in ASE enterprise environments. The major contribution in this paper is the development of a novel multi-agent system architecture for enterprise and its application to a collaborative workflow management system in ASE, an instance of inter-enterprise collaboration with a proof-of-concept prototype to simulate the order entry. The proposed approach is particularly suitable for integrating the FIPA-OS agent framework and enterprise software existing within an organization or among collaborative units.
Finally, this paper presented an efficient integration of multi-agent workflow management, namely the FIPA-OS agent framework with real-time behaviour in an enterprise environment. A multi-agent system that includes deliberative and pure reactive processes has thus been implemented using the FIPA-OS platform. The results are promising for improving semantic integration approaches to address complex scenarios and developing practical applications with enterprise partners.
