Signature Analysis of SRAM-PUF for IoT Decentralized Identifier in Large-Scale Networks.

In this paper, we extract hardware signatures from numerous static random access memory (SRAM) and analyze the feasibility as a signature for Internet of Things (IoT) decentralized identifier (DID). Due to uncontrollable variations in the manufacturing process, SRAM contains inherent randomness that can serve as a physical unclonable function (PUF) to identify the device uniquely. In particular, with the massive deployment of IoT devices, research on SRAM-PUF based distributed device identification is being actively investigated. However, conventional SRAM-PUF research has mainly focused on application to a small-scale IoT environment, so a limited number of SRAM have been analyzed. To generalize the signature utilization, we investigate uniqueness of the PUF from a number of SRAMs. We employed commercial SRAMs and off-the-shelf development boards to extract SRAM-PUFs. Evaluation results demonstrate that SRAM-PUFs are suitable for DIDs even in large-scale IoT networks, considering both chip-by-chip and block-by-block uniqueness.