Researchers from the University at Buffalo have developed a means of tracking the origin of any FDM 3D printed part based only on the extruder used to create them.
According to the team, each 3D printer’s hot-end features distinct thermodynamic properties that impact on the way in which parts are built. By compiling a database of these unique fingerprints or ‘Thermotags,’ the scientists say that in future, it could be possible to better help Intellectual Property (IP) holders prevent the replication of their novel designs.
“What would be the best way to protect our IP from someone else printing the same design using their own printer?” said Zhanpeng Jin, an Associate Professor at the University at Buffalo. “We wanted to find something internal. This ThermoTag behaves like the fingerprint of the 3D printer, and when you print out a new product, you can use watermarking [to prevent patent infringement].”
The researchers’ hot-end setup on a Prusa i3 3D printer. Image via the Transactions on Information Forensics and Security journal.
Introducing thermal IP protection
Over the last decade, the increased affordability of desktop 3D printers has made in-house manufacturing, rapid prototyping and enhanced part customization more accessible than ever before. However, the technology’s open nature also makes it relatively easy to download CAD files, and reproduce patented designs without the owner’s consent.
In the past, patent holders have resorted to the inclusion of Physically Unclonable Functions (PUFs) within their products, in order to make their IP physically identifiable. Such PUFs, including textures, patterns and taggants are very difficult to replicate, but they also add layers of complexity to the manufacturing process, potentially increasing part cost.
Adding watermarks is a similarly complicated task, and given that the process is often performed manually, it has limited efficacy within large batch production. What’s more, if malignant forces were to decode the system behind a firm’s watermark, they could even use it to decrypt those of other products, raising the risk of a potential security breach.
To overcome these drawbacks, the Buffalo team has now developed a new approach, in which the unique heat signature behind each system’s extruder can be identified during pre-heating and incorporated into a 3D model. Once printed, the resulting part then features a hidden watermark which enables the detection of any counterfeits, that can’t be replicated without access to the original system.
The experimental hot-ends used to develop their 3D printing ‘fingerprinting’ method. Image via the Transactions on Information Forensics and Security journal.
Finding 3D printing’s ‘fingerprint’
In order to calibrate their fingerprinting process, the scientists’ measured the rate of temperature change and thermal diffusivity within 45 different extruders during pre-heating. Given that factors such as hardware, materials and imperfections can lead to a unique thermal conduction process, the team found that each hot-end exhibited distinct heat characteristics, at an accuracy rate of around 92%.
Using a Lulzbot Mini, Prusa i3 HICTOP and Ultimaker 2 Go, the team then showed the impact of a machine’s model on its hot-end’s thermal properties, by producing five sets of identical parts with unique thermal variances. Having proven the efficacy of their approach, the researchers proceeded to program an auto-encoder, which automatically extracted the key features from the temperature curves of each machine.
By assimilating vast amounts of thermal data, the team’s algorithm was eventually able to distinguish between the pre-heating behaviors of different hot-end configurations, and generate one-off identifiable watermarks. During one experiment, the team were even able to add binary numbers to a part, by locally modifying the thickness of each printed layer, to spell out an ISBN10 checksum.
Although the scientists were ultimately able to embed and decode their self-styled ‘Thermotags’ into printed parts, they conceded that their approach remains vulnerable to hackers. As a result, they’re now developing a blind watermarking process, through which it could be possible to create a more obscure 3D printing fingerprint, that’s deployable as an ultra-secure anti-counterfeiting measure.
The IP debate rumbles on
Discussion over 3D printing patent enforcement has intensified in the last couple of years, with many firms seeking to better protect their IP. Many of these concerns were crystallized in a European Commission report published last year, which identified the industry’s IP challenges, and suggested changes that could make the sector more competitive.
Similarly, in January 2019, the The UK Intellectual Property Office (IPO) conducted research into “3D printing and intellectual property.” In the project dubbed ‘3DPIP Futures,’ the group found no urgent need for IP regulatory reform, but did urge public bodies to communicate more clearly with firms, to clarify how existing laws apply to the sector.
Elsewhere, researchers from the University of Exeter and Durham University have warned that in future, 3D printed watermarks could become a threat to people’s privacy. In their paper, which was contested by InfraTrac’s Sharon Flank, the team found that VR interfacing may provide hostile actors with a means of surveilling people’s movements.
The researchers’ findings are detailed in their paper titled “ThermoTag: A Hidden ID of 3D Printers for Fingerprinting and Watermarking.” The research was co-authored by Yang Gao, Wei Wang, Yincheng Jin, Chi Zhou, Wenyao Xu and Zhanpeng Jin.
Editor's Note: This article is originally appeared on 3DPrintingIndustry.com by Paul Hanaphy