Pasar al contenido principal

Resultados relacionados

Resultados totales (Incluyendo duplicados): 3
Encontrada(s) 1 página(s)

Zaguán. Repositorio Digital de la Universidad de Zaragoza
Publicación Artículo científico (JournalArticle)

Novel Machine Learning-Resistant RO-Based PUF Optimized for IoT Device Authentication

  • Aparicio-Téllez, Raúl
  • Garcia-Bosque, Miguel
  • Díez-Señorans, Guillermo
  • Celma, Santiago
In this work, a novel Physically Unclonable Function (PUF) based on Generalized Galois Ring Oscillators (GenGAROs) optimized for IoT device identification purposes has been proposed and analyzed. A GenGARO is composed of a certain number of logic gates with up to two inputs connected in cascade so that one input corresponds to the output of the previous gate and the other to the output of the last one. The bias of the signal of these oscillators has been used to construct a GenGARO-PUF. Firstly, the combination of logic gates which optimize the response of the GenGARO-PUF in terms of identifiability has been studied. Once the optimal configuration of GenGARO has been found, a GenGARO-PUF of 11 Look-Up Tables (LUTs) has been implemented on FPGA and its properties have been analyzed and compared to a conventional RO-PUF implemented in the exact same locations of the FPGA and using the same hard constraints. The proposal of this work shows an average Inter- Hamming Distance (HD) of 49.9 % and an Equal Error Rate EER=1.52⋅10-12 using 100-bit responses. The GenGARO-PUF has proven to outperform the conventional Ring Oscillator (RO) PUF in terms of spatial autocorrelation, uniqueness, uniformity, bit-aliasing, identifiability and resistance to modeling attacks. Furthermore, a 3-LUT GenGARO-PUF has been proposed maintaining the prediction accuracy of the 11-LUT GenGARO-PUF, and showing that a good identifiability can still be achieved using fewer resources of the FPGA.
Zaguán. Repositorio Digital de la Universidad de Zaragoza
Publicación Artículo científico (JournalArticle)

Improving Physically Unclonable Functions’ Performance Using Second-Order Compensated Measurement

  • Fernández-Aragón, Jorge
  • Diez-Señorans, Guillermo
  • Garcia-Bosque, Miguel
  • Aparicio-Téllez, Raúl
  • López-Pinar, Gabriel
  • Celma, Santiago
In this paper, we study the performance of second-order compensated measurement to generate a multi-bit response in physically unclonable functions (PUFs). The proposed technique is based on a novel second-order compensated measurement generating multiple bits instead of a single bit provided by the conventional compensated measurement. A PUF based on this technique has been proposed and implemented in 40 Artix-7 FPGAs, and its uniqueness and reproducibility have been compared to those of another PUF using the compensated measurement technique. In addition, we demonstrate that the best trade-off between identifiability and computation time performance is obtained when using only two bits. At the same time, the good performance of the technique has been demonstrated, improving the identifiability of a ring oscillator PUF (RO-PUF) between 70 and 90% compared to a RO-PUF that uses conventional compensated measurement. In particular, equal error rates (EER) of the order of EER∼10−16 can be achieved by combining the sign bit with another bit extracted using the proposed technique; and up to EER∼10−19 by using one more extra bit. In addition, the high reliability of the responses generated by this technique against possible temperature and voltage variations has been proved. These results show how this new technique improves the performance of the PUF in terms of identifiability, so it can be effectively used for device identification purposes.
Zaguán. Repositorio Digital de la Universidad de Zaragoza
Publicación Artículo científico (JournalArticle)

Exploiting nonlinear oscillator dynamics for on-chip thermal sensing

  • Aparicio-Téllez, Raúl
  • Garcia-Bosque, Miguel
  • Diez-Señorans, Guillermo
  • Celma, Santiago
Real-time monitoring of temperature in FPGAs is crucial to ensure system reliability, as local temperature variations can affect their performance. For these purposes, virtual temperature sensors such as Ring Oscillator (RO)-based sensors are widely used. However, efficiently implementing them with a minimal number of Look-Up Tables (LUTs) remains a significant challenge. In this paper, we propose a novel temperature sensor architecture based on nonlinear oscillators, implemented on a 28 nm Artix-7 FPGA. The proposed approach minimizes the usage of FPGA resources while maintaining accurate thermal sensing. As a result, several nonlinear oscillators are analyzed, achieving better sensitivity (0.020%/°C) over a 25 % wider temperature range compared to other RO-based temperature sensors reported in the literature for the 28 nm technology node. These results highlight the potential of these oscillators as efficient, low-resource alternatives for real-time thermal monitoring in FPGA.