%0 Thesis %A Dachowicz, Adam %D 2020 %T Tailored Traceability and Provenance Determination in Manufacturing %U https://hammer.purdue.edu/articles/thesis/Tailored_Traceability_and_Provenance_Determination_in_Manufacturing/12735854 %R 10.25394/PGS.12735854.v1 %2 https://hammer.purdue.edu/ndownloader/files/24106172 %K manufacturing %K traceability %K anti-counterfeiting %K physically unclonable functions %K Mechanical Engineering %X

Anti-counterfeiting and provenance determination are serious concerns in many industries, including automotive, aerospace, and defense. These concerns are addressed by ensuring traceability during manufacturing, transport, and use of goods. In increasingly globalized manufacturing contexts, one-size-fits-all traceability solutions are not always appropriate. Manufacturers may not have the means to re-tool production to meet marking, tagging, or other traceability requirements. This is especially true when manufacturers require high processing flexibility to produce specialized parts, as is increasingly the case in modern supply chains. Counterfeiters and saboteurs, meanwhile, have a growing attack surface over which to interfere with existing supply chains, and have a leg up when implementation details of traceability methods are widely known. There is a growing need to provide solutions to traceability that i) are particularized to specific industrial contexts with heterogeneous security and robustness requirements, and ii) reliably transmit information needed for traceability throughout the product life cycle.


This dissertation presents investigations into tailorable traceability schemes for modern manufacturing, with a focus on applications in additive manufacturing. The primary contributions of this dissertation are frameworks for designing traceability schemes that i) achieve traceability through recovery of manufacturer-specified signals, from simple identity information to more detailed strings of provenance data, and ii) are tuned to maximize information carrying capacity subject to the available data and intended use cases faced by the manufacturer.


In the vein of physically unclonable function (PUF) literature, these frameworks leverage the intrinsic information present in material structure, such as phase or grain statistics. These structures, being functions of largely random and uncontrollable physical and chemical processes, are by their nature uncontrollable by a manufacturer. According to the frameworks proposed in this dissertation, anti-counterfeiting and traceability schemes are designed by extracting large libraries of features from these properties, and designing methods for identifying parts based on a subset of the extracted features that demonstrate good utility for the present use case. Such schemes are customized to handle specific material systems, metrology, expected part damage, and other concerns raised by a manufacturer or other supply chain stakeholders.


First, this dissertation presents a framework that leverages this intrinsic information, and models for damage that may occur during use, for designing schemes for genuinity determination. Such schemes are useful in contexts like anti-counterfeiting and part tracing. Once this framework is established, it is then extended to design schemes for dynamically and securely embedding manufacturer-specified messages during the manufacturing process, with a focus on implementation in additive manufacturing. Such schemes leverage both the intrinsic information inherent to the material / manufacturing process and extrinsically introduced information. This extrinsic information may include cryptographic keys, message information, and specifications regarding how an authorized user may read the embedded message. The resulting embedding schemes are formalized as "malleable PUFs.''


The outcomes of these investigations are frameworks for designing, evaluating, and implementing traceability schemes that can be used by manufacturers, academics, and other stakeholders seeking to implement secure and informative traceability schemes subject to their own unique constraints. Importantly, these frameworks can be adapted for a range of industrial contexts, and can be readily extended as new methods for in-situ measurement and control in additive manufacturing are developed.

%I Purdue University Graduate School