On Formal Modeling of Agent Computations.

Authors: Tadashi Araragi, Paul Attie, Idit Keidar, Kiyoshi Kogure, Victor Luchangco, Nancy Lynch. and Ken Mano.

In the 1st NASA Workshop on Formal Approaches to Agent-Based System (FAABS), pages 48-62, April, 2000. Springer Verlag Lecture Notes in Artificial Intelligence 1871.

Abstract:

This paper describes a comparative study of three formal methods for modeling and validating agent systems. The study is part of a joint project by researchers in MIT's Theory of Distributed Systems research group and NTT's Cooperative Computing research group. Our goal is to establish a mathematical and linguistic foundation for describing and reasoning about agent-style systems.

Our project examines the power of the following three formal methods in studying different agent applications: (1) Erdos is a knowledge-based environment for agent programming; it supports dynamic creation of agents, dynamic loading of programs, and migration. Erdos also supports automatic model checking using CTL. (2) Nepi2 is a programming language for agent systems, based on the pi-calculus. It extends the pi-calculus with data types and a facility for communication with the environment. (3) I/O automata is a mathematical framework extensively used for modeling and reasoning about systems with interacting components. We have extended the I/O automata formalism to account for issues that arise in agent systems, such as dynamic creation and destruction of agents, mobility, and naming. We model a simple e-commerce system using the three formalisms.

Download paper: ps, pdf, ps.gz.


Last modified: Mon Jul 1 14:23:41 EDT 2002