*********************************
There is now a CONTENT FREEZE for Mercury while we switch to a new platform. It began on Friday, March 10 at 6pm and will end on Wednesday, March 15 at noon. No new content can be created during this time, but all material in the system as of the beginning of the freeze will be migrated to the new platform, including users and groups. Functionally the new site is identical to the old one. webteam@gatech.edu
*********************************
Title: Securing Additive Manufacturing Systems from Cyber and Intellectual Property Attacks
Committee:
Dr. Raheem Beyah, ECE, Chair, Advisor
Dr. Fatih Sarioglu, ECE
Dr. Aaron Stebner, MSE
Dr. Saman Zonouz, Rutgers
Dr. Nikhil Gupta, NYU
Abstract: Additive Manufacturing (AM), also known as 3D printing, is gaining popularity in critical industry sectors such as aerospace, automobile, health care, and construction. Since AM heavily replies on computer systems to operate, there is an increased risk of cyberattacks on AM systems. For example, researchers have shown the possibility to maliciously modify the G-code instructions or the firmware of a 3D printer such that the printed objects become defective. When the defective objects are used in a system, they can break in operation, causing damage to the system and loss of lives. To protect AM systems from cyberattacks, a series of Intrusion Detection Systems (IDSs) leveraging side channels in AM systems have been proposed in the literature. These IDSs function by comparing an observed side-channel signal against a reference side-channel signal in real time, point by point or window by window. However, we found that the existing IDSs do not function properly for real printing processes due to a lack of dynamic synchronization. We found the existence of time noise in AM processes. When the same G-code file is executed on the same printer multiple times, the timing of the printing processes varies slightly. In fact, there is randomness in the duration for each G-code instruction and there are random gaps between G-code instructions. If not compensated for, time noise can cause existing IDSs to fail, especially over a long period of time. To mitigate this problem, we propose to use dynamic synchronization to determine the timing relationship between two signals to be compared. We found that Dynamic Time Warping (DTW), which is a well-established method to perform dynamic synchronization, does not work well for side-channel signals in AM systems. Instead, we propose a novel method, called Dynamic Window Matching (DWM) to automatically find the timing relationship between two signals. In addition to IDSs for AM systems, this thesis also touches on the topic of intellectual property protection for AM systems. Specially, we studied the optical side-channel attack to steal intellectual property in AM systems and proposed to use the optical noise injection method to defend against the optical side-channel attack.
