Counterfeit Detection and Prevention in Additive Manufacturing Based on Unique Identification of Optical Fingerprints of Printed Structures

Printed Electronics (PE) based on additive manufacturing has a rapidly growing market. Due to large feature sizes and reduced complexity of PE applications compared to silicon counterparts, they are more prone to counterfeiting. Common solutions to detect counterfeiting insert watermarks or extract unique fingerprints based on (irreproducible) process variations of valid components. Commonly, such fingerprints have been extracted through electrical methods, similar to those of physically unclonable functions (PUFs). Hence, they introduce overhead to the production resulting in additional costs. While such costs may be negligible for application domains targeted by silicon-based technologies, they are detrimental to the ultra-low-cost PE applications. In this paper, we propose an optical unique identification, by extracting fingerprints from the optically visible variations of printed inks in the PE components. The images can be obtained from optical cameras, such as cell phones, thanks to large feature sizes of PE, by trusted parties, such as an end user wanting to verify the authenticity of a particular product. Since this approach does not require any additional circuitry, the fingerprint production cost consists of merely acquisition, processing and saving an image of the circuit components, matching the requirements of ultra-low-cost applications of PE. To further decrease the storage costs for the unique fingerprints, we utilize image downscaling resulting in a compression rate between 83&#x2013;<inline-formula> <tex-math notation="LaTeX">$188\times $ </tex-math></inline-formula>, while preserving the reliability and uniqueness of the fingerprints. The proposed fingerprint extraction methodology is applied to four datasets and the results show that the optical variation printed inks is suitable to prevent counterfeiting in PE.

Abstract licence: CC BY-NC-ND

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.