Intriguing new identity technology

Nature magazine has published an intriguing report called Fingerprinting Documents and Packaging that suggests the possibility of interacting digitally with paper documents without the use of RFID tags or modifications to current production processes.

According to the authors:

“We have found that almost all paper documents, plastic cards and product packaging contain a unique physical identity code formed from microscopic imperfections in the surface. This covert ‘fingerprint’ is intrinsic and virtually impossible to modify controllably. It can be rapidly read using a low-cost portable laser scanner. Most forms of document and branded-product fraud could be rendered obsolete by use of this code.

“Our findings open the way to a new approach to authentication and tracking — even the inventors would not be able to carry out a physical attack on this fingerprint as there is no known manufacturing process for copying surface imperfections at the required level of precision. There is no need to modify the protected item in any way through the addition of tags, chips or inks, so protection is covert, lowcost, simple to integrate into the manufacturing process, and immune to attacks directed against the security feature itself.”

How might the technology be used in passports? A passport could be produced using conventional means, and then the main page of the passport could be scanned and its “fingerprint” held in a database along with relevant identifying information.

When examining the passport, authorities could scan it, causing verification of the fingerprint and retrieval of the associated information. This should match what is printed on the passport, and can be taken as authoritative.

The scheme clearly doesn't transform a document or piece of merchandise into a beacon or contravene the fourth law of identity, since the “digital inspection” of the document is entirely analogous to conventional visual inspection. In my view, this is key.

How does it work?

how does it work

“Figure 1a shows the results of scanning a focused laser beam across a sheet of standard white paper and continuously recording the reflected intensity from four different angles by using four photodetectors. (See supplementary information for methods.) Statistical analysis indicates that there are pseudorandom fluctuations that have a minimum wavelength of 70micrometers. The fluctuations from the mean intensity are digitized into ones and zeros to form the fingerprint code for the object.

“Figure 1b shows the digital cross-correlation between this scan and a similar scan from a different sheet of paper from the pack. The absence of any strong peak indicates that the scans are independent of each other. By contrast, the digital cross-correlation between the original scan and a subsequent scan from the same sheet of paper taken three days later, with normal handling of the paper in between, shows a strong peak close to zero positional shift (Fig. 1c); this indicates that the scans are largely identical. Similar results were obtained from matt-finish plastic cards (such as credit cards), identity cards and coated paperboard packaging (as used to pack pharmaceuticals and cosmetics, for example).”

I can hear you complaining that either intentional desecration or old fashioned wear and tear would render the fingerprint useless. So here is what I found to be perhaps the most astonishing aspect of the invention:

“Recognition was good even after the object had been roughly handled. For paper, this included screwing it into a tight ball, followed by smoothing to leave a badly creased surface; submerging it in cold water for 5min, followed by natural drying; baking it in air at 180 C for 30 min to scorch the surface; scribbling heavily over the scanned area with a ball-point pen and a thick black marker pen; or scrubbing the surface with an abrasive cleaning pad.

“The amplitude of the cross-correlation peak can be used to determine the probability of two objects sharing indistinguishable fingerprints. For the paper studied here, the probability was less than 10 to the 72nd (see supplementary information). Smoother surfaces, such as matt-finish plastic cards and coated paperboard, typically give probabilities of less than 10 to the 20th. The speckle signal therefore serves as a virtually unique fingerprint for the object. Each fingerprint requires about 200–500 bytes of storage space. “

Authors of the paper are James D. R. Buchanan, Russell P. Cowburn, Ana-Vanessa Jausovec, Dorothée Petit, Peter Seem, and Gang Xiong of Blackett Physics Laboratory at Imperial College London; Del Atkinson and Kate Fenton of Durham University Physics Department; and Dan A. Allwood and MatthewT. Bryan of the Department of Engineering Materials at University of Sheffield.

Thanks to Steve Grimaud for bringing this to my attention.

Published by

Kim Cameron

Work on identity.