An exploration of bugs and debugging in multi-agent systems
David Poutakidis, Lin Padgham, Michael Winikoff · 2003
Abstract. Debugging multi-agent systems, which are concurrent, distributed,and consist of complex components, is difficult, yet crucial. In earlier work we have proposed mechanisms whereby protocol specifications available from thedesign process can be used for monitoring the execution of the multi-agent system they describe. Protocol specifications can be used at run-time for reportingany discrepancies in interactions compared to that which was specified. In this paper we describe and categorise a range of bugs found in actual multi-agent sys-tems developed by students in an Agent Oriented Programming and Design class. We then indicate how these bugs manifest to the debugging agent and what in-formation it is able to provide to the user to assist in locating and diagnosing the problem. 1 Introduction Based on foundational work in AI and Philosophy, agent technology has significant ap-plications in a wide range of domains [5]. Agent based software engineering methodologies [1] have emerged to facilitate the development of agent based applications. Theseapplications are often developed as multi-agent systems and are notoriously difficult to debug since they are distributed, concurrent, and complex. Under these circumstancesgood debugging tools are essential.