This session outlines the important role of physical object security
There is an increasing demand for identification and authentication of “dumb” objects, i.e., objects that do not communicate or electronic devices that are too weak to communicate securely. Part of this demand is driven by the growing Internet of Things (IoT). The IoT will contain many objects whose identity needs to be determined by users, and whose authenticity needs to be checked. Having effective means to perform such verification is crucial for the success of IoT enterprises.
Furthermore, there is the traditional demand for effective anti-counterfeiting technologies. Counterfeiting of money, medication, mechanical parts, and goods in general poses a risk to public welfare, public health, profitability of businesses and brand value. On the one hand there are traditional anti-counterfeiting technologies that depend on the secrecy of a manufacturing step or the scarcity of a resource, e.g. micro-engravings, special inks. On the other hand there are Physical Unclonable Functions (PUFs), which allow for a more open approach where proposed technologies can be thoroughly tested by the (scientific) community. The latter approach is more transparent, while the former is typically easier to implement and has a large installed base.
This session will cover important directions in modern physical object security systems, based on added as well as intrinsic security features of physical objects.
Novelty and motivation
This session’s topic is novel ways to determine the identity of physical objects and to verify the authenticity of such objects. The main focus is on non-proprietary (PUF-like) techniques, but novel proprietary technologies are also in scope.
The main novelty of this session is to extend and further develop frameworks for physical object identification and authentication and apply them to practical systems. The session will focus on novel methods, findings and links between encoding techniques, machine learning, cryptography, image processing and forensics. More particularly, the session encourages papers on PUFs, added or intrinsic forensic markers, design verification methods, fine-grained recognition, feature extraction, etc. and various aspects related to the accuracy, complexity, memory requirements, scalability and security. The session will also welcome new datasets in these domains promoting reproducible research.
The idea to propose this session is motivated by the high pace of current innovations in this field and by the observation that most PUF research concentrates on silicon PUFs, leaving non-silicon technologies undeservedly underrepresented.