Verifying Protocol Conformance for Logic-Based Communicating Agents ⋆ Matteo Baldoni, Cristina Baroglio, Alberto Martelli, Viviana Patti, and Claudio Schifanella Dipartimento di Informatica — Universit`a degli Studi di Torino, C.so Svizzera, 185 — I-10149 Torino (Italy) {baldoni,baroglio,mrt,patti,schi}@di.unito.it Abstract. Communication plays a fundamental role in multi-agents systems. One of the main issues in the design of agent interaction pro- tocols is the verification that a given protocol implementation is “con- formant” w.r.t. the abstract specification of it. In this work we tackle those aspects of the conformance verification issue, that regard the de- pendence/independence of conformance from the agent private state in the case of logic, individual agents, set in a multi-agent framework. We do this by working on a specific agent programming language, DyLOG, and by focussing on interaction protocol specifications described by AUML sequence diagrams. By showing how AUML sequence diagrams can be translated into regular grammars and, then, by interpreting the problem of conformance as a problem of language inclusion, we describe a method for automatically verifying a form of “structural” conformance; such a process is shown to be decidable and an upper bound of its complexity is given. We also give a set of properties that describes the influence of the agent private information on the conformance of its communication policies to protocol specifications. 1 Introduction Multi-agent systems (MASs) often comprise heterogeneous components, different in the way they represent knowledge about the world and about other agents, as well as in the mechanisms used for reasoning about it. Notwithstanding hetero- geneity, agents must cooperate, to execute a common task or compete for shared resources; interoperation is, normally, ruled by a set of shared interaction pro- tocols. The design and implementation of interaction protocols are crucial steps in the development of a MAS. Following the development process, described in [21], for interaction protocol engineering, two different kinds of test are to be executed. The first consists in verifying the consistency of an abstract protocol ⋆ This research is partially supported by MIUR Cofin 2003 “Logic-based development and verification of multi-agent systems (MASSiVE)” national project and by the European Commission and by the Swiss Federal Office for Education and Science within the 6th Framework Programme project REWERSE number 506779. J. Leite and P. Torroni (Eds.): CLIMA V, LNAI 3487, pp. 196–212, 2005. c  Springer-Verlag Berlin Heidelberg 2005