Poster: Physics-Based Attack Detection for an Insider Threat Model in a Cyber-Physical System
Lists
Only you see this
Abstract
To ensure the proper functioning of critical systems, it is important to design secure Cyber Physical Systems (CPS). Since CPS are connected systems, most studies consider external adversaries as a threat model, which might not be able to cater for an insider threat with the physical access to the system. In this article, we proposed an attack detection mechanism for an insider who has physical access to a CPS. The proposed method exploits the dynamics of the system and detects an attack based on the laws of Physics. Based on the mass flow equations, we analyze the rate of change in the plant's process and create a feature vector based on the process dynamics. The model has been trained by passing rate of change in system's state as input to Support Vector Machine (SVM), to detect the abnormal behavior in the system. Based on the proposed framework, experiments are performed on a real water treatment testbed, to validate our model and to measure the efficiency of the plant in normal and under attack scenarios. The detection result shows that proposed scheme can detect attacks with accuracy as high as $96%$.
BibTeX
@inproceedings{agrawal2018poster,
author = {Anand Agrawal and Chuadhry Mujeeb Ahmed and Ee-Chien Chang},
title = {{Poster: Physics-Based Attack Detection for an Insider Threat Model in a Cyber-Physical System}},
booktitle = {AsiaCCS},
year = {2018},
doi = {10.1145/3196494.3201587},
}Loading related content...