An Intelligent Agent for Testing Distributed Agent-based Systems
Christopher Rouff, Suk Bae Lee · 2008
Often a Catch-22 situation arises when developers start to test agent-based and other distributed systems. They can not test their agents until other agents are available. Developers need other agents in a community to send messages to their agents to trigger events. Therefore, a sizeable set of agents, none of which may have been developed or tested, may be needed to test a single agent. To get around this problem, developers write dummy agents that do not take any input and just send out predefined messages. This creates extra work and time for the developer and if requirements change these agents will need to be updated. For agent-based systems that learn, a large battery of testing is needed to test that the learning performs correctly. The test and monitoring agent described in this paper is an intelligent test agent that has automatic and semi-automatic features that allows complete testing to be done in less time and provide testing of the system in parts or as a whole. This will provide developers and testers a single system to do both testing and monitoring and will also reduce the overhead in learning multiple systems and provide savings through the reduced number of tools to learn and maintain.