Reducing Aging Impacts in Digital Sensors via Run-Time Calibration

Abstract

County (UMBC) ScholarWorks@UMBC digital repository on the Maryland Shared Open Access (MD-SOAR) platform.

Testing and reliability enhancement of security primitives: Methodology and experimental validation

“…Note that, the number of sensors, and the location they are resided in a chip (to detect FIAs) depend on the number of sensitive blocks we want to monitor and their placement in the chip (refer to fig. 19 of [4]). Figure 13 compares the variation of sensors' AFN index in various voltage quantities vis-a-vis for two sensors in each FPGA, in particular Sensor-1 and Sensor-8.…”

“…There are two aspects linked to that: (i) Testing a PUF means dealing with samples from a joint distribution of manufacturing variations and noise 3 . (ii) The entropy of a PUF is restricted by the limited variations in the manufacturing process and is effectively lowered by noise which often imposes remediation means such as error correction 4 .…”

“…Please note that multi-challenge PUFs like Arbiter PUFs[26] or SUM PUFs[47] and their relatives are rarely used for key storage and are potentially weak against machine learning[36] 3. Potentially, more effects can also be part of this distribution, e.g., temperature shifts, fluctuations in the supply voltage, or device parameters variation owing to its aging 4. Some PUFs, such as the Loop PUF[12], can achieve a required entropy / reliability by adapting the number of oscillations.…”

DELFINES: Detecting Laser Fault Injection Attacks via Digital Sensors

“…In these sensors, instead of quantifying the propagation time, it is checked if the transition feeding the corresponding delay chain manages to propagate to the end of the delay chain at the considered frequency. As will be discussed later, we use such a sensor for detecting LFIAs in this paper [59].…”

Aging-Induced Failure Prognosis via Digital Sensors

“…The classic aging prognosis schemes mainly utilize lookup tables to record the history of operating conditions and use this data to predict aging degradation rate via machine learning models [8,19]. However, as the operating conditions can change frequently during the runtime, using such method is challenging.…”