Physical signal-based intrusion detection for cyber physical systems
Bayens, Christian Johannes
MetadataShow full item record
In recent years, a significant emerging target of cyber-based attacks by nation-states and other advanced groups is cyber-physical systems (CPS). These attacks target major utility, manufacturing, or public service infrastructure in order to collect ransom or intellectual property. A major feature of these attacks is the ability to spoof network traffic to indicate normal activity to a user while malicious instructions are sent to the physical machinery. This thesis investigates methods by which physical sensing and signal analysis may be used as intrusion detection in such a scenario. For the sector of additive manufacturing (AM), we use audio classification and motion detection to identify malicious prints. For the power utility sector, we use a specialized low-frequency radio receiver to detect switching events which can then be compared to network traffic to detect the presence of malicious activity.