VU SoS Lablet Quarterly Executive Summary - APR 2022

A. Fundamental Research

The Science of Security for Cyber-Physical Systems (CPS) Lablet focuses on (1) Foundations of CPS Resilience, (2) Analytics for CPS Cybersecurity, (3) Development of a Multi-model Testbed for Simulation–based Evaluation of Resilience, (4) Mixed Initiative and Collaborative Learning in Adversarial Environments, and (5) Cyber Makerspace for CPS security. 

• In the area of foundations of CPS resilience, we focus on multi-agent reinforcement learning. A network of reinforcement learning (RL) agents that cooperate with each other by sharing information can improve learning performance of control and coordination tasks when compared to non-cooperative agents. However, networked Multi-agent Reinforcement Learning (MARL) is vulnerable to adversarial agents that can compromise some agents and send malicious information to the network. In this work, we consider the problem of resilient MARL in the presence of adversarial agents that aim to compromise the learning algorithm. First, we develop an attack model which aims to degrade the performance of a target agent by modifying the parameters shared by an attacked agent. In order to improve resilience, we develop aggregation methods using medoid and soft-medoid functions. Our analysis shows that the medoid-based MARL algorithms converge to an optimal solution given standard assumptions and improve the overall learning performance and robustness. Simulation results show the effectiveness of the aggregation methods compared with average and median-based aggregation.
• Learned models and policies can generalize effectively when evaluated within the distribution of the training data but can produce unpredictable and erroneous outputs on out-of-distribution inputs. In order to avoid distribution shift when deploying learning-based control algorithms, we seek a mechanism to constrain the agent to states and actions that resemble those that it was trained on. In control theory, Lyapunov stability and control-invariant sets allow us to make guarantees about controllers that stabilize the system around specific states, while in machine learning, density models allow us to estimate the training data distribution. Can we combine these two concepts, producing learning-based control algorithms that constrain the system to in-distribution states using only in-distribution actions? In this work, we propose to do this by combining concepts from Lyapunov stability and density estimation, introducing Lyapunov density models: a generalization of control Lyapunov functions and density models that provides guarantees on an agent's ability to stay in-distribution over its entire trajectory. 
• In the multi-model testbed effort, we investigate simulation-based recovery planning. Despite the rapid development of cybersecurity methods, recovery of the operation of an impacted CPS after a cyber-attack, as a core element of cyber resilience, is often left to human decision-makers. There is a high demand for an autonomous intelligent cyber defense agents (AICAs) for planning a rapid recovery. In this work, we introduce and demonstrate a system for recovery planning using simulation-based predictive monitoring to recover the system from attacks (cyber, physical, or hardware) and disruptions automatically. The recovery planning system first evaluates the impact of system degradation and generates courses of actions (COAs) for efficient recovery. Then, it evaluates these COAs through integrated heterogeneous simulations that accounts for unavoidable uncertainty. By formalizing security and safety requirements, it formally verifies recovery COAs with confidence guarantees, and obtains the optimal recovery COAs. We use two recovery scenarios in smart cities to demonstrate the effectiveness of our recovery planning system.

B. Community Engagement(s)

• PI Xenofon Koutsoukos was guest editor in a special issue on Artificial Intelligence/Machine Learning Enabled Reconfigurable Wireless Networks appeared in the IEEE Transactions of Network Science and Engineering, 9(1), 2022.

C. Educational Advances