Evaluating the Applicability of Multi-agent Software for Implementing Distributed Industrial Data Management Approaches

Abstract

Distributed approaches to industrial control or information management problems are often tackled using Multi-agent methods. Multi-Agent systems—solutions resulting from taking a Multi-agent based approaches—often come with a certain amount of “overhead” such as communication systems, but can provide a helpful tool with the design and implementation. In this paper, a distributed data management problem is addressed with both a bespoke approach developed specifically for this problem and a more general Multi-agent approach. The two approaches are compared using architecture and software metrics. The software metric results show similar results, although overall the bespoke approach was more appropriate for the particular application examined. The architectural analysis indicates that the main reason for this difference is the communication and computation overhead associated with the agent-based system. It was not within the scope of this study to compare the two approaches under multiple application scenarios.