Abstract: A key challenge with the distributed digital nature of additive manufacturing is validating both part quality and the security of the associated digital data. Monitoring physical side-channel emissions is one approach that can be used to achieve this goal. To improve security, this monitoring system can be digitally disconnected, but this eliminates digital methods of communicating information about the part and build to the monitoring system. This limitation can be mitigated by adding features, such as barcodes or QR codes, to a build to physically transmit information instead. The drawback of adding these features is increased material consumption, build time, and the possibility of affecting part properties. However, by inserting metadata into the toolpath in a non-intrusive way, physical side-channel emissions can be transmitted without having a significant impact on the part being fabricated. The metadata can contain part/build IDs, quality information, and/or tamper resistant security features. This work discusses the requirements, benefits, and challenges of inserting, transmitting, and receiving this metadata and demonstrates two approaches, one using g-code for a material extrusion system and an acoustic monitoring system, and the second using a laser powder bed fusion system and a laser position monitoring system. In the first approach, stepper motor speed is varied to generate a series of tones containing the metadata, in the second, a QR code is used to generate the toolpath, but modified process parameters are used such that only an accelerated scanning motion occurs without activating the laser. Both approaches demonstrate the ability to embed and receive metadata without having a significant effect on finished part properties.
Bio: Logan Sturm is a Weinberg fellow at Oak Ridge National Laboratory in the Embedded Systems Security group, part of the Critical Resilient Infrastructure (CRI) section. He received his Ph.D. in mechanical engineering from Virginia Tech in 2020 where he researched cyber security for additive manufacturing (AM) systems. His research at ORNL is in cyber security for manufacturing systems including AM, CNC, and hybrid systems, with a focus on using physical sensing to detect anomalous behavior. Current work includes embedding metadata into part toolpaths to automatically communicate to monitoring systems over an airgap using physical emissions, evaluating computer vision-based quality control systems against adversarial attacks, and exploring approaches for training and education to mitigate cyber security risks. Broad research interests include in situ monitoring, data analytics, machine learning, and human factors in a security context. Logan is also currently serving on an ASTM committee to develop standards to address the unique cyber security challenges in additive manufacturing.