Modeling and Analyzing Periodic Sensor-Reading Modification Attacks in Cyber-Physical Systems

Wireless networks in a cyber-physical system can be vulnerable to periodic attacks that are commonly employed by energy-constrained intruders to compromise a system. In this work, we investigate a class of periodic attacks, termed periodic sensor-reading modification attacks, where events transmitted over sensor channels can be maliciously altered. First, we construct models for such attacks and analyze the system behavior under attacks. Then, we specifically study periodic sensor-reading modification attacks in a fault diagnosis scenario. A case study involving an unmanned aerial vehicle system illustrates how such attacks interfere with fault diagnosis accuracy. Furthermore, we design an algorithm to derive stealthy and successful strategies to prevent operators from identifying a fault in finite observations.