The growing accessibility of 3D printers and downloadable models online has made it easier than ever to produce products and printed parts from home. Millions of digital design files are available for sale or for free online. Intellectual property (IP) owners have a strong interest in protecting their work and preventing the counterfeit manufacturing of their products. As one example, Honda demanded that all Honda-related files available on Printables.com be taken down just last year [1]. Unfortunately, copyright ownership can be difficult to prove in the world of digital files and 3D printed objects. Presented with a 3D printed object, how could one prove that it originated from a specific manufacturer or printer, and wasn’t printed by someone who made a replica based on the original product?
One possible solution is the use of digital watermarking. By embedding identification information or other metadata into the 3D model or 3D object, it’s possible to show that a model or printed object came from a specific company or printer. In this form of watermarking, the identification data can either be inserted directly into the 3D model before being sent to a 3D printer, or added to the G-code after the slicing phase.
In one study, researchers developed a blind watermarking algorithm designed to be applied to 3D printed objects using fused deposition modeling (FDM) 3D printers [2]. The term “blind watermarking” is used to describe the ability to locate and read the watermark data from the object without requiring the original, non-watermarked object for comparison. In this technique, the researchers were able to embed predetermined watermark data into the object by varying the thickness of the printed layers. Another study introduced a method that can generate unique fingerprints of 3D printers by using the thermodynamic properties of the printer’s extruder hot-end to build a watermark [3]. This technique allows manufacturers to insert digital watermarks into their 3D products that can indicate both where the object was printed (i.e., the company or manufacturer), and which specific printer and extruder hot-end were used to print the object in the manufacturing facility.
As digital watermarking techniques become increasingly advanced and robust against attacks, it’s important to recognize their limitations. One common assumption is that instructions sent to a 3D printer are executed without modification. In this way, as long as the printer is configured correctly and doesn’t malfunction, the watermark data will be properly transferred to the object being printed. However, manufacturers who use watermarking techniques should be aware of the risk of instructions being manipulated, corrupted, or dropped. If left unchecked, a malicious actor could inject themselves into the pipeline and cause minor or major changes to the G-code, resulting in a printed product that resembles the model in appearance, but is just different enough to remove or obfuscate any watermark that should have been added. A skilled attacker could even add their own watermark to the data, calling into question the authenticity and true IP owner of the products once placed onto the market.
To detect these types of attacks, BreakPoint Labs has developed BISON, which monitors the G-code executed by an additive-manufacturing machine and compares it to the intended G-code. In doing so, every command is verified. If an unusual or unintended command is executed, an alert is sent to the operator.
If you are interested in learning more about securing additive manufacturing or a demonstration of BreakPoint Lab’s BISON AM solution capability, please contact us at info@breakpoint-labs.com.
[2] https://ieeexplore.ieee.org/document/8943311
[3] https://ieeexplore.ieee.org/document/9385084