Embedded Obfuscated Barcodes for Identication of Genuine Additive Manufactured Parts

Additive manufacturing (AM) has been adopted for manufacturing complex shaped highly customized components for aerospace, automotive, and medical elds, where intellectual property protection and counterfeit detection are major concerns. New technologies such as Blockchain have been promising in supply chain authentication. However, AM due to layer-by-layer manufacturing process provides opportunities of embedding information inside the part during manufacturing, which has been explored recently to embed identication codes inside the parts. The present work studies the possibility of printing a barcode inside the additively manufactured part and develops a scheme to obfuscate the code design to read differently from different directions to enhance the security and protect the intellectual property. The embedded three-dimensional codes are scanned using a micro-CT scan. This scheme of embedded obfuscated codes proves to be a highly customizable and ecient process while securing product design les.


Introduction
Additive manufacturing (AM) has evolved as one of the promising manufacturing technologies in aerospace in medical sectors [1][2][3]. However, security of intellectual property (IP) and identi cation of counterfeit and unauthorized products have always been an issue in aerospace and medical sectors [4][5][6]. Cloud-based platforms have enabled sharing and collaboration for development of product designs; however, security of information has been a concern for such systems due to increased attack surface [7].
The attack taxonomy has been studied in the available literature and shows several possible attack vectors in the AM eld [8]. Since AM is currently used for development and production of highly valued parts such as aircraft and spacecraft components and patient speci c medical device parts, security concerns are ampli ed for this eld [9][10][11][12]. The additively manufactured products are subjected to cyberattacks, counterfeiting or other illegal activities, causing huge nancial loss and safety concerns [10,[13][14][15][16][17]. Currently, there is very limited discussion on the effective solutions to secure AM and its products. In the following Sect. 1.1, an overview of some of the anti-counterfeiting methods is presented along with the possible implementation of these methods in AM and its products. It is evident that the conventional security labels are highly subjected to removal-reapplication attacks, which will result in severe losses. Our scheme of embedding "security labels" inside the product can reduce the risk of security breach. Furthermore, we also propose a novel security strategy that is rooted in the design les to prevent counterfeiting production by e ciently distinguishing the fake products from genuine ones [18].

Overview of current product authentication tools
Authentication technologies and security tools are essential to all genuine products and their legitimate IP owners [19]. Anti-tamper devices, package and box seal, package sealing tapes are available for tamper resistance, but once exposed, the product identity can be still stolen and reproduced for counterfeits.
Property marking is also made on the surface where products can be marked with security markers and stamping products [20,21]. These markers can be in uorescence but transparent and only visible under an UV light. However, this technique may be limited to protect high-value components with extremely small sizes such as a microchip due to writing di culty. In addition, traces of these UV markings may be seen with naked eyes if not performed properly and subsequently removed by attackers. Other methods like holographic label printing anti-counterfeiting approach enabled by laser holographic technology are also used to prove product authenticity [20,22]. ID card, passport, document laminates can be made using holographic technology with multi-color uorescing inks, optically variable ink (OVIs) and liquid crystal solutions to create unique identi able features [20,22].
Credit cards and banknotes are designed to have holographic security labels on them which change colors as they are tilted [20,23]. However, these labels are still subjected to duplication, reproduction or removal by attackers. Product serial numbers on a sticker label can be placed externally on the product and help to distinguish a genuine product from a counterfeit one but they can be scratched off to prevent product authentication or other logistic management [24]. For example, barcode or QR code labels are attached on the outer surface of consumer products so that they can be identi ed with a code scanner.
This scanning system is favored by many businesses for high e ciency, productivity, and relatively low cost [24,25]. The small barcode can encode su cient amounts of information in a visual pattern.
However, the tracking label can be damaged and lose its functionality. In addition, the radio-frequency identi cation (RFID) tags, with or without an electronic chip, have been widely used in many applications including warehouse logistics, product tracking in a supply chain, passports, and banknotes [23,26]. In many applications like identity cards, passports or credit cards, RFID tags with integrated circuits (ICs) are used to prevent data theft and control access [26,27]. An RFID tag or chip is more resistant to wearing damage but is still subjected to be removed or disabled by attackers similar to other product authentication labels.

Proposed AM product security approach
All the cybersecurity strategies can help to keep the product safe from counterfeiting, but once they are breached, IP can be stolen, and counterfeits can be produced. Therefore, we propose a security approach that protects the product at the design level, where parameters including printer capabilities, layer resolution, scanning resolution are taken into consideration. The aforementioned authentication tools and techniques can be subjected to damage at different levels. When applying these tools in AM and its products, surface labeling such as security marking or label attachments are highly vulnerable to the removal-reapplication attacks. Security labeling such as laser engraving tools can reduce the risks of removal-reapplication attacks to some extent, but also make the security features visible and subject to counterfeiting using 3D scanning reverse engineering. Therefore, in this work we study the possibility of embedding security and authentication codes inside the component that is intended for 3D printing at the product design level. Widely used barcodes are taken as an example in this study for the possibility of embedding them in product design. Such embedded codes would not be visible from outside and cannot be tempered.
Only by imaging techniques such as computed tomography (CT) scanning, the embedded features can be visualized and con rmed for authenticity. In this work, we have provided an extensive description for the CAD development with embedded features, the scanning and reading procedure, and further improved complexity on the embedded features. The design strategies, as discussed in previous studies [15,[28][29][30][31] were utilized to demonstrate the enhancement of product security by modifying the embedded features such that a unique set of processing parameters are required to produce parts with the correct embedded codes.

Design And Process Development
A barcode image representing the four letters "CMML" (that stands for "Composite Materials and Mechanics Lab") is selected and imported into a 3D modelling software SolidWorks 2019 to construct a 3D CAD model of a cuboid with embedded barcode, which is then used for AM.

One-layer barcode
One-dimensional zebra-striped barcodes are typically used in the retail industry for commercial products [32]. These 1D barcodes encode data only horizontally using black bars on a white background, which are su cient enough for storing information such as numbers or short tags. For a string eld with more than 25 characters, however, they are insu cient, let alone a text area eld that has a 30,000 characters' limit or URLs [32]. Despite their limited information storage capability, one dimensional barcodes are still widely used for high feasibility of creating and reading [33].
Code 128 is one of the most widely used 1D barcodes in shipping and packaging industries as a product identi cation code. Therefore, a code 128 is generated for the string character "CMML" using an online conversion tool as shown in Fig. 1(a), which contains machine readable information (in black strips) and human-readable text (in letters underneath) [33]. A 3D CAD model as shown in Fig. 1

(b) is created in
SolidWorks 2019 by rst importing the barcode image in Fig. 1(a) and then using the "Sketch" and "Extruded Cut" functions to de ne contour, locations, and dimensions. The complete 1D barcode is embedded within a solid prism at the same depth and 3D printed using VeroClear transparent photopolymer for visualization of internal features as shown in Fig. 2. When imported in a Slicer program, the STL model is scaled down and printed in different sizes using the Stratasys Objet30 Pro resin 3D printer to evaluate possible miniaturization of using such codes. As depicted in Fig. 2, a part embedded with a 1D barcode can be 3D printed as small as 10 × 6 × 3 mm 3 in length × width × thickness and still have a good visibility of the code to the naked eye. Smaller codes can be generated for reading through imaging methods.

Mechanical Testing on one-layer barcode
Parts with embedded codes are evaluated for their mechanical properties and structural integrity. It is expected that under the optimal design, with the security feature size minimized and properly positioned inside the part, the parts' original functionality will not be compromised.
The standard tensile bar specimen geometry with embedded barcode feature is developed as shown in Fig. 3(a), where the same barcode pattern in Fig. 1 is used here. The volume of "Extruded Cut" barcode is 7.85 mm 3 as compared to 15739.09 mm 3 for the entire solid tensile bar, which gives a volumetric ratio of 0.0499% between barcode and tensile bar as in the CAD program. The tensile bar les are used for 3D printing using the Stratasys Objet30 Pro resin 3D printer and VeroClear photopolymer model material with high printing resolution down to 16-microns layer thickness. Meanwhile a set of standard tensile bars without any embedded codes are also printed using the same method to determine the baseline properties. The 3D printed tensile bar with embedded barcode is shown in Fig. 3(b), where the embedded barcode can be seen clearly from the front view. After visually validating the existing embedded barcodes, which are printed with support material (Stratasys SUP706) rather than the model material, standard tensile test is performed on 5 tensile specimens. Three tensile bars without embedded codes are also tested for reference (fewer than 5 specimens were tested due to high consistency; deviation is less than 1%).
The processed tensile test results are plotted in Fig. 4 with representative data from both categories, showing a close match between the stress-strain curves. The tensile properties are calculated and summarized in Table 1, where embedded code specimens show on average only ~ 0.35% lower tensile strength and ~ 0.39% lower modulus as compared to the reference specimens without embedded code. Therefore, it can be safely concluded that an embedded tracking code such as the one used in Fig. 3 would not signi cantly affect the part's tensile properties, and hence can be a promising scheme to secure the product while preserving its strength and functionality.

Multi-layer barcode
The same barcode representing "CMML" as in Fig. 1(a) is used to create the multi-layer barcode CAD model in SolidWorks 2015 as shown in Fig. 5. Similarly, the barcode image was rst imported in SolidWorks and a total of 22 rectangular prisms were generated using the "Sketch" and "Extruded Cut" functions to de ne contour, locations, and dimensions. With careful placement, the top and right-side views can display the exact same barcode using the multiple layers embedding scheme as depicted in Fig. 5. Distances between each stripe and their widths are used to reconstruct the same barcode pattern on the side. By changing the thickness of each stripe, the side pattern can be dramatically different, leading to different barcode reading, or simply a false barcode. With the distributed layers of codes, it not only obfuscates the appearance of the 3D barcode, but also reduces the stress concentration as compared to the one-layer barcode model.
Once exported as STL le (coarse resolution), the model is sent to the Stratasys J750 3D printer and 3D printed using the multi-material 3D printing method (PolyJet), where the barcode volume is printed in red using VeroMagenta photopolymer, and outside enclosure cuboid is printed using VeroClear photopolymer for transparency as shown in Fig. 5.

Embedded obfuscating barcode models
The embedded tracking code can serve as a unique product signature by distinguishing itself from counterfeiting products that do not have any embedded tracking codes. However, when the entire design les are stolen, it is likely that expert attackers can use these design les to illegally reproduce parts with comparable quality and the same tracking code information. Therefore, we have modi ed the design of these embedded features such that they can be only printed correctly under a speci c set of processing parameters such as printing orientation. In this case, despite being stolen, attackers will need to invest signi cant resources to determine the best combination of processing parameters to produce parts with high quality and the sane tracking codes, which can effectively distinguish counterfeits from the genuine products and protect the original rightful IP owner.

Obfuscating one-layer barcode
The embedded barcode feature can be modi ed with other 3D modelling features such as the surface feature to secure the original product. Here a CAD modeling scheme is developed such that the embedded barcode can only be 3D printed under a speci c set of processing parameters (individual feature size, resolution of STL le and direction of slicing), while other attempts will not generate parts with the embedded barcode for an authentication process. In this way, attackers or other unauthorized users cannot reproduce the part with the embedded tracking code easily without knowing the processing parameters, which results in failure at passing the authentication steps.
The original CAD model for the barcode model is shown in Fig. 1(b), where the barcode stripes are extruded inside a solid prism. This model is modi ed such that the barcode stripes are applied with the "Extruded Surface" function inside the solid prism as shown in Fig. 6, so that it becomes an obfuscating embedded barcode model. Each barcode stripe is represented with an open surface in the same rectangular shape as the stripe. When these barcode stripes are performed with the "Extruded Surface" instead of the "Extruded Cut" function, the model shows orientation-dependent slicing results.
In the original CAD model, dimensions for the solid prism are 12 × 8 × 4 mm 3 , where the minimum width of the bar stripe is 0.13 mm. However, depending on the printer resolution and capability, a scale factor should be applied such that the minimum width of barcode stripe is a multiple of the printer layer thickness. The modi ed CAD model is translated to a standard tessellated le format STL and saved in coarse STL resolution, which is then imported to a slicer CatalystEX for FDM 3D printing. The minimum layer thickness 0.178 mm is chosen for 3D printing using this printer, therefore, the minimum width of stripe is scaled up to 0.178 mm with a scale factor of 1.37 applied to original CAD model dimensions, to ensure that it can be printed. The scaled model is sliced under default orientations and three other manually adjusted orientations, which are 90° rotations about x-, y-, and z-axis respectively. For this modi ed barcode model, auto-optimized orientation option is not available due to missing facets or reversed normal. Out of the four slicing orientations, it is observed that only the manually adjusted 90°r otation about x-axis orientation generates a toolpath for barcode, while other orientations slicing does not generate any barcode toolpath as shown in Fig. 7. It can be further indicated by the geometry of CAD model and slicing results that only when the extruded surfaces are normal to the slicing direction, the barcode extruded surface features can be recognized, and toolpath can be generated. However, when the smallest stripe width is 0.178 mm (scale factor of 1.37) the barcode pattern is not generated correctly as shown in Fig. 7(d) as opposed to Fig. 6(d). Therefore, the scale factor is further increased to 2.74 and 4.1 such that the smallest barcode stripe width is 0.356 mm and 0.534 mm, which are twice and thrice of the printer layer thickness, respectively.
It can be observed from the slicing results that toolpath of the barcode pattern is generated for both scaled model (2.74 and 4.1 scale factors) when placed under manually adjusted 90° rotation about x-axis orientation, however, the larger model with 4.1 scale factor has shown more precise barcode pattern toolpath than other models as shown in Fig. 8. Likewise, under the default orientation, all models (scale factor of 1.37, 2.74, and 4.1) generate toolpath only for the prism without any barcode stripes. These results have shown a close relationship between machine capability and design freedom. While AM allows a high degree of freeform fabrication, careful investigation needs to be implemented to achieve the best manufacturing outcomes. When the original design had a minimum stripe width below the printer resolution (0.178 mm), the security feature was not able to be implemented correctly. Under both XY and XZ orientations, the barcode stripes are printed with support materials in a wrong pattern. No difference can be observed when changing the slicing orientation with respect to the model. The slicing results are validated by the FDM printed parts as shown in Fig. 8, where support material is deposited for larger width barcode stripes and less for smaller widths.
From this experiment, it can be observed that when extra surfaces/planar surfaces are introduced, the STL tessellation le shows two surfaces, unlike the four extruded surfaces as originally in CAD model. When additional extruded surfaces (unclosed) are introduced in the solid prism, the solid prism shows "missing facets or reversed normal" error message in the Slicer when attempting to auto-optimize its orientation. Moreover, under the default orientation, the solid prism model is printed as one solid part without any support material deposited inside. However, when manual orientation to XY (90° rotation to x-axis from default), the embedded column of surface features is printed with the support material in the pattern of correct barcode design, and under all other 90° rotation, the model is printed as one solid part. But this phenomenon may or may not occur under other 3D printing techniques and slicing algorithms. In summary, by introducing four surfaces (in the shape of rectangular tube) with determined dimensions, a column of support materials can be printed without any dependence on the slicing orientation. Authentic parts will be printed by authorized users where support materials are deposited showing the barcode pattern for tracking and authentication process.
A micro-CT scanning and reconstruction by the Micro-CT system is performed on the FDM 3D printed modi ed barcode part as in Fig. 8(a) to validate previous discussion on the reading techniques. The printed prism, Fig. 8(b), is subjected to micro-CT scan and then cut to physically verify the presence of the code Fig. 8(c). When exposed to x-ray radiation as in Fig. 8(c), color indication shows the barcode stripes as printed with a different material than the prism model material. This can also quickly help in validating different printing results for the modi ed barcode model with security features under different orientations. Further, a 3D model can be reconstructed from the computed tomography scanned crosssection images. This step is necessary because the scanning orientation might not be su cient to visualize the barcode pattern based on the assumption that this kind of tracking code can be embedded with 360° degree of freedom. The reconstructed model can be viewed at any angle and any cross-section to visualize the barcode pattern as in Fig. 8(c), which shows relatively good barcode pattern and contrast.
With improved printer capability and resolution, dimensional accuracy can be increased, and the tracking code can be read faster.

Obfuscating multi-layer barcode
By combining the ndings described earlier in the sections on multi-layer barcode and obfuscating onelayer bar code, the multi-layer barcode model as shown in Fig. 5 is enhanced with obfuscation security feature so that only under certain conditions will the correct barcode be printed, and only a certain viewing orientation can read the code correctly. The "Extruded Surface" function is performed on the barcode instead of "Extruded Cut" function. The "Extruded Surface" barcode is 30 mm in length with square sides that are either 0.6 × 0.6 mm 2 , 1.2 × 1.2 mm 2 , or 1.8 × 1.8 mm 2 , all embedded inside a cuboid enclosure that is 56 × 40 × 56 mm 3 in length × width × height as shown in Fig. 9. The STL model is exported in coarse STL resolution, which is then imported into the slicer CatalystEX and sliced under different orientations as shown in Fig. 10. Only under the 90° rotation about x-axis will the barcode be printed correctly while other orientations cannot print the barcode at all. The STL model is then scaled down (scale factor 0.4) to a smaller size compatible for micro-CT imaging and sliced after rotations (x-90° and z-90°) as in Fig. 10 Fig. 11 to validate that the x-90° rotation slicing orientation can properly generate the embedded barcode inside the 3D printed cuboid.

Reading techniques
The tracking codes embedded inside the solid model bene t from the freeform solid modelling stage of AM.
In previous examples, we have developed CAD models with embedded tracking code and additively manufactured 3D objects from these CAD models in a transparent photosensitive polymer material to facilitate visualization and code pattern validation. However, such a scheme can be developed into any 3D solid model and produced with any materials, without the necessity of being transparent. Under the scenario of an opaque or translucent appearance model, direct code reading might not be feasible so an external facility for internal structure visualization needs to be employed. Examples of such nondestructive reading techniques include ultrasonic imaging and x-ray computed tomography (CT). Here we demonstrate the feasibility of using a micro-CT machine to acquire internal information of a 3D printed product with embedded tracking code. During the acquisition, a series of cross-sectional slices will be generated for the object and used for 3D model reconstruction. Once the internal pattern is reconstructed using the CT-scan machine, an e cient image processing scheme can be adopted to validate product authenticity.
From the CAD models and printed objects, it can be observed that currently, AM technologies are capable of manufacturing tracking codes with similar geometries at high resolution and miniaturized size. Therefore, in the processing of this reconstructed image, it is promising to apply pre-determined geometry such as rectangular or squares so as to discriminate signal noises and only acquire the useful information for image analysis and code authentication. With this, our scheme of a robust and highly e cient authentication process can be realized and compatible with a broad selection of products.

Conclusions
The process of designing, printing, and reading an embedded tracking code is successfully demonstrated in this article. The 1D barcode example is used for illustration, but more complex design or a multi-layered code can also be implemented following the similar protocol. Compared to the conventional security labels that are highly subjected to removal-reapplication attacks, our scheme of embedding "security labels" inside the product can reduce the risk of security breach. Combined with the additional obfuscation design, the embedded feature complexity and security is increased to further distinguish the genuine product from counterfeits. Unauthorized users without access to the key process parameters will eventually produce counterfeits without the embedded security feature that will fail the authentication process.

Declarations FUNDING/ACKNOWLEDGMENTS
This work is supported by the National Science Foundation grant DGE-1931724 from the Secure and Trustworthy Cyberspace program. Authors thank NYU Tandon Makerspace for the facilities and support provided for the experimental work. The views presented in this article are those of the authors, not of the funding agency.

Con icts of interest/Competing interests:
The authors have no con icts of interest to declare that are relevant to the content of this article.
Availability of data and material: Data can be provided if requested.
Code availability: Not applicable.
Ethics approval: This article does not contain any studies with human participants or animals performed by any of the authors.

Consent to participate:
There are no human participants that are included in the study.

Consent for publication:
The authors have not included any data taken from other individuals.

Figure 1
A 1D code 128 is generated (a) for the string character "CMML" using an online barcode generator https://barcode.tec-it.com/en, and (b) convert into a CAD model using "Extruded Cut" function (code dimensions 474×300×100mm3, enclosure cuboid dimensions 630×380×200mm3).          otation about x-axis for the printed part to properly generate the embedded barcode.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.