Secure Intelligent Agents Based on Formal Description Techniques

Abstract

This paper describes a practical solution for the incorporation of security services in agents. From a set of basic user requirements, the agents will be able to find out the best way to implement a security policy to accomplish the communication. A tipical scenario could be the following: one user wants to provide his communications with security, but he only knows a set of basic user requirements, and he has not any knowledge about cryptography, security protocols, security keys, etc. Configuring the agent correctly, this would be able to find out the optimal security protocol which best fit with those requirements. Once the best protocol is found, our agents will interpret it dynamically, for which formal description techniques will be used. The main aim of this work consists of finding out this security protocol. To do it, a global optimization method will be used, in particular, genetic algorithms. These algorithms are capable of finding out the optimal value in a function, within a domain. Our domain will be the security protocols which could be optimal, and the function to optimize would be a function capable of evaluating those protocols, and the optimal value will be the protocol that best fit with the security requirements. In short, our agents will be capable of finding out and interpreting dynamically the ad-hoc security protocol from a set of basic user requirements.

This work is supported by Science and Technology Ministry of Spain under the proyect “A Multi-Agent System Architecture for the automatic and dynamic implementation of security protocols (TIC2001-3376).”