Date of Original Version



Technical Report

Rights Management

All Rights Reserved

Abstract or Description

Abstract: "When large Multi Agent Systems operate over open platforms such as the Internet, there is the possibility that some agents will endeavor to disrupt the MAS with the aim of achieving their own agendas at the cost of the others. For the protection of the agents in it's [sic] domain, each society of agents should make available certain services that allow agents to detect and avoid such attempts by malicious third parties. One of these is a rigorous method of identifying agents, so that each agent's actions can be referred to, recorded and evaluated. A robust identification mechanism can allow the construction of higher level-functions such as authentication, non-repudiation and secure communication. Agents can then use these basic functions provided by the MAS infrastructure along with their beliefs, current state of the MAS and the current goals to make decisions about trust. In this paper we present one architecture that allows agents to be named robustly, even in open environments, and build upon this scheme to provide MAS services that can allow agents to authenticate each other, digitally sign communication tokens and encrypt messages and data that is exchanged. We explore possible scenarios of the agent life in an open agent society and give an implementation independent architecture to secure agents and their environment."

Included in

Robotics Commons