Using Optical Imaging to Combat Counterfeit Components
See original post
Counterfeit parts and materials pose a serious threat to the United States defense supply chain. The National Defense Authorization Act (NDAA) of 2012 (Section 818) laid out strict guidelines for DoD prime contractors for detection and avoidance of counterfeit electronic parts1. The Defense Logistics Agency (DLA) identified six federal supply groups in their supply chains that are at high risk for counterfeiting including electrical and electronic components, bearings, hardware and abrasives, pipes and fittings, engine accessories, and vehicle components2. DLA is championing development of anti-counterfeiting technologies that also provide traceability for the parts in its supply chain. A solution called QuanTEK developed by Chromo Logic LLC, involves novel optical imaging methods and has been found to work on a broad class of the DLA identified high-risk federal supply groups.
Concept of Operation
Figure 1. QuanTEK Concept of OperationThe QuanTEK system exploits the fundamental fact that all items exhibit unique, information-rich surface features as a consequence of the normal manufacturing process. These micron-scale surface features on each item, like a fingerprint or iris pattern used to authenticate humans, are highly random and impossible to forge or copy3. QuanTEK obtains these patterns with a non-contact imaging system and, using proprietary algorithms, tracks products back to an enrollment scan at an approved source to verify authenticity of the product.
As shown in Figure 1, the QuanTEK system consists of a camera system with built-in adaptive lighting/magnification, a raised base for placing the parts, a mini-PC for the control and image processing algorithms, and a tablet with a Graphical User Interface (GUI) to operate the system. The concept of operation is a two-step process. Step 1 consists of taking an optical image (noncontact) of the surface of the part in an enrollment scan carried out at the start of the supply chain, for example at an original component manufacturer or at an authorized distributor. This Step 1 may also occur at an independent distributor who complies with the appropriate provenance processes with the parts tested at an authorized test laboratory. From the enrollment scan, QuanTEK’s proprietary algorithms choose several regions of interest (ROI’s) in the image and convert those ROI’s into a first virtual tag or vTAG™ stored in a secure cloud database.
Step 2 occurs at the end of the supply chain where the user, the original equipment manufacturer, the depot, or the Services, takes a second optical image as a verification scan and generates a second vTAG for cloud storage. The two vTAGs are compared and if they match, the part is declared as secure. If there is no match for the second vTAG, the part is suspect.
As shown in Figure 1, the matches and non-matches are quantified by a metric called the normalized Hamming distance (NHD). In the case of a perfect match, all the ones and zeros of the two vTAGs will be identical, giving a normalized Hamming distance of zero. If there is a perfect non-match, on average half the bits will match and half of them will not match, giving a normalized Hamming distance of 0.5. In most real cases, the matches and the non-matches will fall within a slight distribution reflecting hardware noise and variation between units. An appropriate threshold setting is selected that gives low false positive and false negative rates, then a normalized Hamming distance below the threshold is declared as a match, and the normalized Hamming distance above it declared as a non-match.
Figure 2. Uncoated steel washer (left) and results on a kit of 10 identical (same part number) uncoated steel washers.The QuanTEK system parts handling capability is robust, is not affected by the material of the part and it performs in both dry and damp heat conditions, including thermal and humidity cycling. The system does not require any special treatment of the parts under test. More importantly, QuanTEK eliminates the need to mark a part – a process that could potentially violate its intellectual property or its function.
Multiple verifications can be carried out whenever a part changes custody establishing traceability of parts through the supply chain, and the database may also contain time/location stamps and other information.
Successful testing was done on commonly used washers with no identifiable external markings. Figure 2 shows an uncoated steel washer with results from tests consisting of ten such identical washers. All ten washers were enrolled using the QuanTEK system. Each enrolled washer was verified with itself, and the results showed a low NHD as represented by the blue bars. When any one of the enrolled washers was verified against any of the other nine washers, the results showed a high NHD represented by the red bars.
A level 2 test is done using the same QuanTEK unit but at different times with potentially different orientations, while a level 3 test represents two different QuanTEK units, one for enrollment and the second for verification. The results show negligible unit to unit differences and the separation in the average NHD values between the blue and red bars shows the ability to clearly differentiate between similar looking washers.
Comparing this result with an actual use case, if ten washers were enrolled at a manufacturer, and if somewhere in the supply chain, a few of those washers were removed and substituted with similar looking washers, then at a downstream verification point, the QuanTEK system would be able to correctly identify the similar looking unenrolled suspect washers. Similar successful results have been obtained for chrome plated washers, case hardened steel washers with blackoxide finish, and painted washers.
Figure 3 shows results for shielded semi-ground radial steel bearings. As before the testing involved 10 identical bearings, and the level 3 results shown on the right clearly demonstrate Figure 3. Steel bearings (left) and results for a kit of 10 identical (same part number) bearings (right).QuanTEK’s ability to identify individual bearings and differentiate between each of them, ensuring that unenrolled suspect parts are identified and removed from the supply chain.
Other mechanical parts tested were brass aircraft fasteners as shown in Figure 4. A simple jig was constructed so that the head of the fastener could be imaged. As before, 10 identical fasteners (same part number) were enrolled on a QuanTEK system. Each of the ten parts was verified with itself and with the other nine parts. These measurements were carried out with the parts having ten different part orientations from 0 degrees to 360 degrees. QuanTEK could clearly distinguish between each of the ten fasteners, independent of their angular orientation in the jig.
Figure 4. Brass aircraft fastener (left) and QuanTEK system results showing orientation independence.Rigid pipes presented special challenges while imaging their cylindrical surfaces. The QuanTEK system uses proprietary optical methods to ensure there were no spurious artifacts caused by the curvature of the part. Figure 5 shows the QuanTEK’s ability to image pipes with an outside diameter as small as 0.25 inches, offering clear separation between the matched (blue) and nonmatched (red) distributions.
QuanTEK is a novel tag-less anti-counterfeiting solution that works for both electronic and mechanical parts providing on-site and immediate verification of suspect parts. QuanTEK’s secure cloud database is expected to enable prime contractors and their sub-contractors to comply with 2012 Figure 5. Rigid pipe (left) and QuanTEK system results (right).NDAA Section 818 which requires mechanisms to be put in place to ensure traceability of parts as they move through the supply chain. The first generation systems have manual enrollment and verification processes, and the algorithms execute in just a few seconds, providing potential for further automation and integration in manufacturing workflows. QuanTEK has been successfully pilot tested by DoD prime contractors on integrated circuits and is now ready for commercial use.
This article was written by Matthew Mattson, Senior Scientist, ChromoLogic LLC (Monrovia, CA); Naresh Menon, CEO, ChromoLogic LLC; and Paula M. George, Quality/Technical Analyst, J-344 Technical and Quality Division, Defense Logistics Agency Headquarters (Fort Belvoir, VA). For more information, contact Mr. Mattson at firstname.lastname@example.org or visit http://info.hotims.com/61059-203.
1. National Defense Authorization Act for Fiscal Year 2012 (http://www.gpo.gov/fdsys/pkg/PLAW-112publ81/pdf/PLAW-112publ81.pdf).
2. Christine Metz, “Traceability in the Age of Globalization,” panel discussion, ERAI Conference (www.erai.com), Orlando, Florida, April 18, 2013.
3. J. Buchanan, R. P. Cowburn, et al, “Fingerprinting” documents and packaging, Nature, Vol 436, pp. 475, 2005.