A research team has developed a security system for additive manufacturing processes to help protect them against possible cyberattacks, as well as to ensure the overall integrity of the components produced by 3D-printing methods.
Scientists at the Georgia Institute of Technology and Rutgers University have developed a three-layer method combining acoustic, spatial, and material techniques that “enable real-time and post-production detection of a defective print,” Raheem Beyah, professor and associate chair in Georgia Tech’s School of Electrical and Computer Engineering, told Design News .
The system—the first of its kind—uses a validation technique that’s independent of printer firmware or software in the controlling computer to ensure that components produced using additive manufacturing have not been compromised.
Raheem Beyah, a professor and associate chair in Georgia Tech’s School of Electrical and Computer Engineering, is shown in a 3D printing lab at the Woodruff School of Mechanical Engineering. Beyah led a team that developed a three-layered security system for 3D printing to ensure components of this type of fabrication process are not compromised. (Credit: Christopher Moore, Georgia Tech)
Dedicated security processes are commonplace among business IT systems due to the high risk of intrusion, data breaches, and cyber attacks. However, until now they haven’t specifically secured additive manufacturing systems present on computers and hardware attached to those networks. These systems are becoming more prevalent to replace conventional fabrication processes in areas ranging from aerospace components to medical implants, making the software that controls them also vulnerable.
Each of the three layers of the system Beyah and his team developed has its specific form and function, and for each researcher’s require a “golden, or trusted” copy of the print and the corresponding signature, he said. For the acoustic layer, the team used an inexpensive microphone to record the sounds from various actuators and fans, which generate an acoustic pattern.
“This pattern should be identical for every identical object that is printed,” he explained. “Thus, we can compare the audio from the golden copy to that from subsequent prints. Deviations indicate some sort of problem—either malicious or benign.”
The spatial layer is comprised of a linear potentiometer and gyroscopic sensor attached to each other to construct a set of spherical coordinates to describe the printer’s motion, Beyah said. “Similar to the acoustic layer approach, we obtain data—this time from the linear potentiometer and gyroscopic sensor— from the printing of the golden copy and compare it with the data generated from subsequent prints,” he said.
The materials layer uses contrast agents embedded into the printed object during the printing process to act as signature markers for particular prints without compromising the structural integrity of the original model, Beyah said.
“We embed nanoparticles [gold nanorods] at different points in the printed model to generate a pattern specific to the model,” he said. “This can be accomplished by using a machine that has dual extruders with one containing filament with embedded nanorods. This pattern is only known to us—thus if the print is modified the nanorods will be out of place or missing.”
The researchers tested their technique on three different types of 3D printers and a computer