Search

In the vulnerability analysis of System on Chips, memory hierarchy is considered among the most valuable element to protect against information theft. Many first-order side-channel attacks have been reported on all its components from the main memory to the CPU registers. In this context, memory hierarchy encryption is widely used to ensure data confidentiality. Yet, this solution suffers from both memory and area overhead along with performance losses (timing delays), which is especially critical for cache memories that already occupy a large part of the spatial footprint of a processor. In this paper, we propose a secure and lightweight scheme to ensure the data confidentiality through the whole memory hierarchy. This is done by masking the data in cache memories with a lightweight mask generator that provides masks at each clock cycle without having to store them. Only 8-bit Initialization Vectors are stored for each mask value to enable further recomputation of the masks. The overall security of the masking scheme is assessed through a mutual information estimation that helped evaluate the minimum number of attack traces needed to succeed a profiling side-channel attack to 592 K traces in the attacking phase, which provides an acceptable security level in an analysis where an example of Signal to Noise Ratio of 0.02 is taken. The lightweight aspect of the generator has been confirmed by a hardware implementation that led to resource utilization of 400 LUTs. [ABSTRACT FROM AUTHOR]

Counterfeiting of an Integrated Circuit (IC) has become a significant concern for electronics manufacturers, system integrators, and end users. It is necessary to find a robust implementation that is efficient, low cost, and noninvasive in detection and avoidance of ICs counterfeiting. In this paper, we introduce the concept of using a guided radiofrequency (RF) wave technique to authenticate ICs. The approach discussed in this work highlights the use of electromagnetic (EM)/radiofrequency (RF) response that has been further evaluated to assign fingerprint or signature of ICs for the purpose of authentication. Our approach is to use EM/RF guided wave to sense the response of the ICs, extract the manufacturing-based process variation of an IC and finally generate identifier or signature of that IC. As a proof-of-concept, we performed experiments over different field-programmable gate array (FPGA) boards of the same family. The post-processing technique was applied on the measurement results to statistically quantify the error probability of the authentication technique. [ABSTRACT FROM AUTHOR]

A high-level (i.e., in terms of security) authentication method is proposed in this communication, where the chipless RFID is extended to the chipless authentication. The proposed method is based on low-cost inkjet-printed square-check-patterned chipless tags, whose design is explicitly optimized keeping the randomness inherent to the inkjet-printing process in view. These optimized chipless tags are very difficult to duplicate, as their unique backscattered EM responses depend on proximate coupling among the possible separated squares that happened naturally due to the randomness of inkjet printing. The performance of the proposed method is analyzed by a low-cost impulse radio (IR) UWB chipless RFID reader as well as by the highly accurate VNA-based chipless RFID reader. The achieved probability of error is comparable to the various fingerprint evaluation campaigns found in the literature. [ABSTRACT FROM AUTHOR]

In this paper, a chipless solution dedicated to the authentication of manufactured products is proposed. The proposed technique is an extension of chipless radio frequency identification (RFID) with the prevention of tags’ cloning due to the use of natural randomness in the fabrication process. For the first time, we proposed a methodology to characterize the chipless RFID tags for the purpose of authentication. The proposed method has a high level of security due to the employment of a database of authenticity. With a nonintrusive and nondestructive operation, the proposed method has promising possibilities for authentication applications, for example, the chipless tags can be inserted (or hidden) in the product packaging. The chipless tags are realized two times intermittently, where each realization is constituted of 45 tags. These two different realizations share the same company, same technology but different film mask to ensure the natural dimensional randomness. The similarity analysis is conducted inside each realization and also between two different realizations. With the help of this technique, it is possible to authenticate two chipless RFID tags exhibiting natural dimensional variations in their designs with a low probability of error. [ABSTRACT FROM AUTHOR]

In this article, we report on video-rate identification of very low-cost tags in the terahertz (THz) domain. Contrary to barcodes, Radio Frequency Identification (RFID) tags, or even chipless RFID tags, operate in the Ultra-Wide Band (UWB). These THz labels are not based on a planar surface pattern but are instead embedded, thus hidden, in the volume of the product to identify. The tag is entirely made of dielectric materials and is based on a 1D photonic bandgap structure, made of a quasi-periodic stack of two different polyethylene-based materials presenting different refractive indices. The thickness of each layer is of the order of the THz wavelength, leading to an overall tag thickness in the millimetre range. More particularly, we show in this article that the binary information coded within these tags can be rapidly and reliably identified using a commercial terahertz Time Domain Spectroscopy (THz-TDS) system as a reader. More precisely, a bit error rate smaller than 1% is experimentally reached for a reading duration as short as a few tens of milliseconds on an 8 bits (~40 bits/cm2) THID tag. The performance limits of such a tag structure are explored in terms of both dielectric material properties (losses) and angular acceptance. Finally, realistic coding capacities of about 60 bits (~300 bits/cm2) can be envisaged with such tags. [ABSTRACT FROM AUTHOR]