Ramazan Ayasli, Research Assistant, Albert-Ludwigs University

Abstract: Actual authentication methods, especially for electronic products, are based on extrinsic features, which can be differentiated into trackers, e.g. RFIDs, and markers, e.g. fluorescent ceramic particles, hologram stickers, DNA-Markers, etc. The use of extrinsic features for authentication is therefore associated with additional production steps and production costs. Higher security against counterfeits is related with higher expenses for manufacturers.

A marker and tracker free authentication system based on intrinsic product features is presented as a cost-efficient alternative. Intrinsic product features result directly from the manufacturing process and are inseparable connected with the product itself. Due to statistical variation of process parameters during the manufacturing process, intrinsic features underlie a statistical variation which can be used for a databased authentication system, without additional expenditures. Intrinsic features can be classified into two types depending on the features statistical variance.

A first type shows only a small variance. This type can be named as direct feature, since its features are usually direct consequences of product design, layout, material choice etc. On the contrary, the second type shows a wide distribution, e.g. the surface structure of molded products or visible particle inclusion in coatings. An authentication system can be build up by calculating an ID out of an intrinsic feature for each produced unit and providing a database with all IDs. Two ID calculation methods will be presented. One method includes detection of predefined structures in an image with pattern recognition methods and using informations of position, orientation, size and shape of the structures in an ID. Another approach includes the calculation of mean pixel values for predefined areas on the product image. Authenticity can be ensured, if a customer is able to reproduce a known ID with a given method.