Beyond CAGE: Investigating Generalization of Learned Autonomous Network Defense Policies

Advancements in reinforcement learning (RL) have inspired new directions in intelligent automation of network defense. However, many of these advancements have either outpaced their application to network security or have not considered the challenges associated with implementing them in the real-world. To understand these problems, this work evaluates several RL approaches implemented in the second edition of the CAGE Challenge, a public competition to build an autonomous network defender agent in a high-fidelity network simulator. Our approaches all build on the Proximal Policy Optimization (PPO) family of algorithms, and include hierarchical RL, action masking, custom training, and ensemble RL. We find that the ensemble RL technique performs strongest, outperforming our other models and taking second place in the competition. To understand applicability to real environments we evaluate each method's ability to generalize to unseen networks and against an unknown attack strategy. In unseen environments, all of our approaches perform worse, with degradation varied based on the type of environmental change. Against an unknown attacker strategy, we found that our models had reduced overall performance even though the new strategy was less efficient than the ones our models trained on. Together, these results highlight promising research directions for autonomous network defense in the real world.

Towards Causal Models for Adversary Distractions

Automated adversary emulation is becoming an indispensable tool of network security operators in testing and evaluating their cyber defenses. At the same time, it has exposed how quickly adversaries can propagate through the network. While research has greatly progressed on quality decoy generation to fool human adversaries, we may need different strategies to slow computer agents. In this paper, we show that decoy generation can slow an automated agent's decision process, but that the degree to which it is inhibited is greatly dependent on the types of objects used. This points to the need to explicitly evaluate decoy generation and placement strategies against fast moving, automated adversaries.