Adversarial dynamics are a critical facet within the cyber security domain,in which there exists a co-evolution between attackers and defenders in anygiven threat scenario. While defenders leverage capabilities to minimize thepotential impact of an attack, the adversary is simultaneously developingcountermeasures to the observed defenses. In this study, we develop a set oftools to model the adaptive strategy formulation of an intelligent actoragainst an active cyber defensive system. We encode strategies as binarychromosomes representing finite state machines that evolve according toHolland's genetic algorithm. We study the strategic considerations includingoverall actor reward balanced against the complexity of the determinedstrategies. We present a series of simulation results demonstrating the abilityto automatically search a large strategy space for optimal resultant fitnessagainst a variety of counter-strategies.
展开▼
