Cai, Feiyang, Li, Jiani, Koutsoukos, Xenofon.
2020.

Detecting Adversarial Examples in Learning-Enabled Cyber-Physical Systems using Variational Autoencoder for Regression. 2020 IEEE Security and Privacy Workshops (SPW). :208–214.

Learning-enabled components (LECs) are widely used in cyber-physical systems (CPS) since they can handle the uncertainty and variability of the environment and increase the level of autonomy. However, it has been shown that LECs such as deep neural networks (DNN) are not robust and adversarial examples can cause the model to make a false prediction. The paper considers the problem of efficiently detecting adversarial examples in LECs used for regression in CPS. The proposed approach is based on inductive conformal prediction and uses a regression model based on variational autoencoder. The architecture allows to take into consideration both the input and the neural network prediction for detecting adversarial, and more generally, out-of-distribution examples. We demonstrate the method using an advanced emergency braking system implemented in an open source simulator for self-driving cars where a DNN is used to estimate the distance to an obstacle. The simulation results show that the method can effectively detect adversarial examples with a short detection delay.

Zhou, Xingyu, Li, Yi, Barreto, Carlos A., Li, Jiani, Volgyesi, Peter, Neema, Himanshu, Koutsoukos, Xenofon.
2019.

Evaluating Resilience of Grid Load Predictions under Stealthy Adversarial Attacks. 2019 Resilience Week (RWS). 1:206–212.

Recent advances in machine learning enable wider applications of prediction models in cyber-physical systems. Smart grids are increasingly using distributed sensor settings for distributed sensor fusion and information processing. Load forecasting systems use these sensors to predict future loads to incorporate into dynamic pricing of power and grid maintenance. However, these inference predictors are highly complex and thus vulnerable to adversarial attacks. Moreover, the adversarial attacks are synthetic norm-bounded modifications to a limited number of sensors that can greatly affect the accuracy of the overall predictor. It can be much cheaper and effective to incorporate elements of security and resilience at the earliest stages of design. In this paper, we demonstrate how to analyze the security and resilience of learning-based prediction models in power distribution networks by utilizing a domain-specific deep-learning and testing framework. This framework is developed using DeepForge and enables rapid design and analysis of attack scenarios against distributed smart meters in a power distribution network. It runs the attack simulations in the cloud backend. In addition to the predictor model, we have integrated an anomaly detector to detect adversarial attacks targeting the predictor. We formulate the stealthy adversarial attacks as an optimization problem to maximize prediction loss while minimizing the required perturbations. Under the worst-case setting, where the attacker has full knowledge of both the predictor and the detector, an iterative attack method has been developed to solve for the adversarial perturbation. We demonstrate the framework capabilities using a GridLAB-D based power distribution network model and show how stealthy adversarial attacks can affect smart grid prediction systems even with a partial control of network.

Shabbir, Mudassir, Li, Jiani, Abbas, Waseem, Koutsoukos, Xenofon.
2020.

Resilient Vector Consensus in Multi-Agent Networks Using Centerpoints. 2020 American Control Conference (ACC). :4387–4392.

In this paper, we study the resilient vector consensus problem in multi-agent networks and improve resilience guarantees of existing algorithms. In resilient vector consensus, agents update their states, which are vectors in ℝd, by locally interacting with other agents some of which might be adversarial. The main objective is to ensure that normal (non-adversarial) agents converge at a common state that lies in the convex hull of their initial states. Currently, resilient vector consensus algorithms, such as approximate distributed robust convergence (ADRC) are based on the idea that to update states in each time step, every normal node needs to compute a point that lies in the convex hull of its normal neighbors' states. To compute such a point, the idea of Tverberg partition is typically used, which is computationally hard. Approximation algorithms for Tverberg partition negatively impact the resilience guarantees of consensus algorithm. To deal with this issue, we propose to use the idea of centerpoint, which is an extension of median in higher dimensions, instead of Tverberg partition. We show that the resilience of such algorithms to adversarial nodes is improved if we use the notion of centerpoint. Furthermore, using centerpoint provides a better characterization of the necessary and sufficient conditions guaranteeing resilient vector consensus. We analyze these conditions in two, three, and higher dimensions separately. We also numerically evaluate the performance of our approach.