Autonomous and Adaptive Cyber-Defense: Application of Game Theory in 5G Security Architectures

5G network development paves the way for new services to the customer, but also brings new and complex cyber-attacks, and opportunities for attackers to exploit pre-existing attack strategies. Against such a threat environment, regular static defense systems may either not implement all the tools to be effective against all attacks, or be too costly. In this paper, we propose an adaptive cyber-defense solution that leverages on game theory algorithms to determine the best defensive strategy at any given time in a 5G architecture. This solution takes into account the current threat environment, the expected and measured effectiveness of available defense mechanisms, as well as their costs. For demonstration purposes, we focused on a specific innovative defense mechanism: the Moving Target Defense (MTD). MTD is used to dynamically move the attacker's target in order to block the attack, or at least to make it too costly for the expected result. Different variations of this strategy are implemented in our system, displaying different levels of effectiveness and costs, and the algorithm constantly chooses which one to use. The defense mechanisms and the algorithm are embedded into a global autonomous security architecture. Our experiments measure the time necessary to determine the best defense strategy, as well as the global effectiveness and cost of our solution compared to non-adaptive ones.