The Cell Dependency Analysis on Learning SRAM Power-Up States

Zhonghao Liao, Yong Guan

Abstract: The authentication and verification of a device requires strong and cost-effective digital fingerprint based solutions. The power-up state of Static Random Access Memory (SRAMs) has been widely used for a variety of authentication and identification purposes, such as SRAM-based Physical Unclonable Functions (PUFs) and Random Number Generators (RNGs), which form the basis for many cryptographic algorithms and protocols. Many published work have made various assumptions on the SRAM power-up states. In this work, we have conducted a comprehensive set of empirical SRAM cell spatial dependency studies, and the statistical and physical analysis thereof, to validate these critical assumptions and results in this domain. This paper details the results of the experimental tests on different models of SRAM chips and show that the power-up states of SRAM cells must be carefully selected in order to avoid some unwanted dependency effects. Otherwise, the probability of key disclosure will increase, and the randomness of dynamic seeds used in critical cryptographic functions will be challenged. Furthermore, our purposed method provides a more reliable cell dependency analysis of SRAM chips and can be used as a guidance for key selection and/or randomness extraction for many other similar security applications.

External IDs:dblp:conf/asianhost/LiaoG18