Configuring requirements-compliant multi-agent systems

Abstract

This paper describes Proteus, a tool set that allows users to build distributed multi-agent systems by selecting system properties, ranging from availability of critical components, performance characteristics, desired level of communication protection (encrypt, signature, specific algorithms), policies for society objects (e.g., messages, black-board objects, files, databases) and certificate authorities. Proteus then analyzes selected properties for consistency and tries to resolve them autonomously using previously provided resolution rules. The user is notified of any conflicts Proteus could not resolve for guidance. Once properties are deemed to be resolved, Proteus can generate system policies as well as the configuration files (e.g. XML files) that can be used to deploy the system.