Evolutionary Design of Attack Strategies

Abstract

The motivation for this presentation came from our work on trying evolutionary approaches on building protocols for security. We eventually came to an understanding that evolutionary approaches help us in finding problems that we would not be aware of, and that generated the idea of trying evolutionary approaches not to generate new protocols, but to generate attacks.

The basic motivation is that there is a fundamental symmetry between the attacker and the defender, in the sense that an attacker needs to only find one attack path to successfully compromise the system, where a defender must be able to secure all of them to have a secure system. So by this analogy, the brute force search over the space of all possible attack paths is more suitable for a defender, whereas some kind of informed search where not all possibilities in the search space are examined is more suitable for an attacker.