Contributions on using embedded memory circuits as physically unclonable functions considering reliability issues
[eng] Moving towards Internet-of-Things (IoT) era, hardware security becomes a crucial research topic, because of the growing demand of electronic products that are remotely connected through networks. Novel hardware security primitives based on manufacturing process variability are proposed to enha...
| Autor: | |
|---|---|
| Tipo de recurso: | tesis doctoral |
| Estado: | Versión publicada |
| Fecha de publicación: | 2021 |
| País: | España |
| Institución: | CBUC, CESCA |
| Repositorio: | TDR. Tesis Doctorales en Red |
| OAI Identifier: | oai:www.tdx.cat:10803/674156 |
| Acceso en línea: | http://hdl.handle.net/10803/674156 |
| Access Level: | acceso abierto |
| Palabra clave: | SRAM cells characterization Reliability margins Mismatch metrics SRAM PUFs Enginyeria electrònica 621.3 |
| Sumario: | [eng] Moving towards Internet-of-Things (IoT) era, hardware security becomes a crucial research topic, because of the growing demand of electronic products that are remotely connected through networks. Novel hardware security primitives based on manufacturing process variability are proposed to enhance the security of the IoT systems. As a trusted root that provides physical randomness, a physically unclonable function is an essential base for hardware security. SRAM devices are becoming one of the most promising alternatives for the implementation of embedded physical unclonable functions as the start-up value of each bit-cell depends largely on the variability related with the manufacturing process. Not all bit-cells experience the same degree of variability, so it is possible that some cells randomly modify their logical starting value, while others will start-up always at the same value. However, physically unclonable function applications, such as identification and key generation, require more constant logical starting value to assure high reliability in PUF response. For this reason, some kind of post-processing is needed to correct the errors in the PUF response. Unfortunately, those cells that have more constant logic output are difficult to be detected in advance. This work characterizes by simulation the start-up value reproducibility proposing several metrics suitable for reliability estimation during design phases. The aim is to be able to predict by simulation the percentage of cells that will be suitable to be used as PUF generators. We evaluate the metrics results and analyze the start-up values reproducibility considering different external perturbation sources like several power supply ramp up times, previous internal values in the bit-cell, and different temperature scenarios. The characterization metrics can be exploited to estimate the number of suitable SRAM cells for use in PUF implementations that can be expected from a specific SRAM design. |
|---|