LSA ^TM works on any material that isn't perfectly transparent and which isn't a perfect mirror. It has been tested against paper, cardboard, coated paper-board, a range of plastics (PVC, polycarbonate, nylon), metals and industrial ceramics. It does not work on clear glass.

LSA can withstand high levels of damage to the item. In the case of paper, tests have been performed where the document has been baked in an oven (to simulate aging), put under water, scribbled on with a pen, scrubbed with an abrasive pad, crumpled into a tight ball. In all cases, we were able to recognise the same fingerprint which had initially been measured when in pristine condition even after the damage.

The time taken to read a fingerprint is limited only by how quickly the sense head can be made to move with respect to the item. In the case of small, portable scanners with a small motor, it takes approximately 1 second to acquire the fingerprint. If the item is already moving, however, such as on a production line or printing press, speeds as high as 20 metres per second can be achieved. Speeds as high as this are hardly ever encountered even in the fastest production environment - LSA can therefore acquire fingerprints in real time without slowing production.

The user can select the file size, within the range 125 bytes - 750 bytes. The larger the file size, the greater the damage that can be sustained to the item without loss of recognition. We recommend 500 bytes for coated paperboard and other shiny surfaces; 250 bytes is sufficient for most matt-finish plastics and paper documents.

Why not! At 500 bytes per item, an entire year's inventory would occupy 500 Gbytes of data - just over one hard drive's worth.

No. Most in-the-field operations use a laptop connected to the Internet via a GPRS card. The operator scans an item, its fingerprint is transmitted over the internet to the central database, where a search is performed against the inventory of genuine items. The result is then transmitted back to the operator.

The LSA data format is particularly easy to search against (much easier than human fingerprints, for example). A standard desktop computer can test 10 million targets per second; an enterprise server can test 100 million targets per second. Most inventories can therefore be searched within a few seconds.

Yes. Unless the inventory size is very small, we always use two databases. One is the main database that contains the full fingerprint and all of the metadata (e.g. time of manufacture, destination market etc). The other contains what we call the thumbnail of the fingerprint. The thumbnail is a miniature version of the fingerprint. While it can't offer high confidences of matching, it can be searched very rapidly. We use the thumbnail database to draw up a 'most likely' list; only the items on the most likely list are then matched against the full database, reducing significantly the amount of data that has to be retrieved. We have a further trick for searching very large databases rapidly. In large inventories, there are usually some features of the product (e.g. a sell-by date, a batch number or a particular brand) that can be used to divide the database into smaller chapters. We use these to reduce the number of thumbnails to be searched to around 1 billion or so. We can then usually find a match within 30 seconds.

The intrinsic placement tolerance of LSA is around ±1mm. Our production line scanners can control multiple scan heads, each displaced from the other by approximately 1mm. Multiple fingerprints are then stored in the database; to validate the item, a match is only required against any of them. For example, suppose the item could be offset by up to 4mm on the production line. We would therefore use 4 sense heads, each recording a 1mm track. This increases the size of the database by a factor of 4, but this is rarely of concern.

This is not how LSA is designed to work. If there were a common feature, a counterfeiter could soon learn how to replicate it. Only by storing item level information can you become immune to the attacks of counterfeiters. Database space is so cheap (around 1 USD per gigabyte), there is no reason not to record every item.

LSA is designed to scan a rectangle 5mm high and around 40mm long. However, the length of the rectangle can be easily customised. For paper-based materials, a scanning length as small as 10mm is possible. For plastic and coated paper-board materials, 30mm is the minimum recommended scan length. There is no theoretical limit on the maximum scan length, although the database size will increase and there is rarely any advantage to scanning more than 60mm.

Yes. The LSA scanners can tell which parts of the surface are printed and which are plain. The software can be set to either focus only on those parts that are unprinted or to include everything. In either case, a small amount of calibration is needed so that the software can learn which parts of its signal are fingerprint and which parts are printing. This is usually achieved by passing a few dozen nominally identical items through the system. It quickly learns which parts of the signal are random and which are fixed.

LSA uses a diode laser with a wavelength of 635nm and an output power of 1mW. It conforms to ANSI Z 136.1 class II and IEC/EN 60825-1 class 1M. It is intrinsically safe, provided optical instruments are not used to intentionally view the beam. There are no known health or safety implications when using this kind of laser. The laser itself is less powerful than those used in conference laser pointers.

Yes. The software provides an analysis of whether any local regions within the scanned area fail to match the database fingerprint. This could be evidence of local tampering, e.g. replacing a photograph on a passport or scraping the surface of the signature portion of a document.

For paper documents and matt plastics, around 1 part in 10 to the power of 100. For coated paper-board items, around 1 part in 10 to the power of 30. These probabilities are evaluated using Binomial Statistics. The fingerprint is broken down into 1000 bits. It is firstly demonstrated that these bits are independent of each other, using cross correlation functions, and that the probability of an accidental match of a given bit is 0.5. Binomial statistics are then used to evaluate the probability that n or more of the 1000 bits would match accidentally. As a final check, large numbers of items are scanned and all of the fingerprints paired with each other and a histogram of the match quality is evaluated. The standard deviation of the distribution is related to the number of degrees of freedom in the randomness - this can be checked against the assumptions used. Further details can be found in the Supplementary Information published by the highly respected journal Nature, at
http://www.nature.com/nature/journal/v436/n7050/edsumm/e050728-05.html

Since LSA is based around an optical technique, the laser passes through any transparent overlayers and wrapping, and probes the first non-transparent surface it reaches. Cartons that will be wrapped in cellophane can be scanned before wrapping, and a good match to the fingerprint will still be obtained at a later time by reading through the cellophane.

Very. The operator simply places the item on our scanner, aligns it by pushing it against a guide rail and clicks an icon on the software. The software interface can be programmed to give a simple yes/no response or to give more detailed diagnostic about the item. All software and hardware actions are recorded in a log file, for auditing, security and quality control.

In comparison to most other security and track & trace technologies, LSA works out remarkably cheaply. The cost structure has 3 parts: hardware, licensing and database costs. The big financial advantage of LSA is that there is no 'per-item' cost: you don't have to pay for a chip or an ink to go on every item you manufacture. The fingerprint is already there - all you need is the hardware to measure it and the database to store it. Precise costs are evaluated on a customer by customer basis - please contact us if you would like a quotation.

This is a brand new approach to security and track and trace. We are currently rolling out scanners onto selected company production lines. Please contact us if you would like to arrange a pilot on your site.

Please refer to Professor Cowburn's report in the highly respected peer-reviewed journal Nature (Nature 436, 475 (2005)). A copy of the paper can be downloaded free of charge from:
http://www.docurights.com/drmaker.cgi?vid=5628&objid=310288

Highly technical supplementary information was published with this article. It can be downloaded from:
http://www.nature.com/nature/journal/v436/n7050/edsumm/e050728-05.html

Yes, Ingenia Technology Ltd holds a large number of patents (both granted and pending), covering the basic principles of LSA, different implementations, software algorithms and business method applications.