Design of active middleware protocols for coordination of distributed resources
Abstract
In this research, collaboration-based information exchange among autonomous agents in a distributed environment is investigated. This research explains a conceptual model of networked enterprises' active middleware and addresses the significance of coordination in everyday business activities. Active middleware provides fundamental functions for integrating high-level services in large, distributed organizations with local autonomy, and individual system platforms and processes. The focus of this research is on a coordination protocol that is a major component in active middleware. The coordination protocol is critical to enable rational collaboration among the users who rely on a network of independent information systems, e.g., multi-ERP systems, extended supply networks, and virtual enterprises. The scope of this research is to develop a methodology to model, quantify, and analyze the impact of a coordination protocol in distributed manufacturing systems in order to establish a principle for coordination protocol selection. The focus problem in this research is the allocation of distributed resources solved through associating agents with resources and activities. Agents associated with activities (task agents) request agents at resources (resource agents) for the resource they require. The task agents and resource agents communicate using coordination protocols to exchange information and decisions. To improve coordination network performance, e.g., multi-agent systems where centralized authority is not available, the application of agent's viability measure involved in the agent's decision mechanism during the coordination process is proposed. This research also develops a simulation tool using a parallel computer for coordination protocol evaluation, called Teamwork Integration Evaluation/Protocol (TIE/Protocol). It accommodates the independent interactions and decision behaviors of independent individual entities. The experimental results show that the coordination protocol, which is designed based on the coordination principle, performs better than a general-purpose protocol. In addition, with viability of agents and system monitored during the coordination process, a protocol considering the viability performs better than a protocol ignoring the viability.
Degree
Ph.D.
Advisors
Nof, Purdue University.
Subject Area
Industrial engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.