Defending Side Channel Attacks in Cyber-Physical Additive Layer Manufacturing Systems

Additive layer manufacturing has been termed as one of the proponents of the fourth industrial revolution. However, due to the presence of cyber and physical domain components in additive manufacturing, they are prone to physical-to-cyber domain attacks. One of the example of such attacks is side-channel attacks, where an attacker can steal valuable intellectual property of the 3D objects being printed by the additive manufacturing system. To understand and defend these attacks, we have to analyze all the possible side-channels and minimize the amount of leakage in each of them. The first objective of this project is to demonstrate the novel side-channel attack models. To achieve this, we have modelled an acoustic attack model where utilizing acoustics emanated by the 3D printer through the side channel, we have successfully reconstructed the 3D-test objects and its corresponding G-code (cyber domain data) with high axis prediction accuracy of 86.00% and length prediction error of just 11.11%. Besides this, we have also incorporated the acoustic side-channel to detect firmware modification attacks by estimating a data-driven behavioral model of the 3D-Printer system using acoustic. With our estimated acoustic model of the system, the average detection accuracy for range of errors introduced in the cyber-domain is 77.45%. Moreover, our current work also incorporates multiple side-channel analysis, and understanding of the cyber-domain for designing leakage aware cyber-domain security solutions.