Peter Kietzmann: New Crypto-fundamentals in RIOT

Internet-connected devices need to provision crypto components in order to communicate securely in the network. Random number generation, ID building, authentication, or secret key generation rely on decent platform resources that provide unpredictable but reliable high-entropy numbers. Common solutions require a certain level of hardware and software complexity and thus device cost that are typically unavailable in the IoT.

Physically unclonable functions (PUFs) are a promising class of solutions to this problem. They extract output from individual hardware properties that persist due to manufacturing inaccuracies. Recent work has identified SRAM memory as a feasible, widely available entropy source. The extracted physical characteristics from uninitialized memory pattern can be used to generate random numbers but also to produce device unique secrets, which facilitates various application scenarios.

This talk starts from basic SRAM memory analysis of well-known RIOT devices. We point out applicable PUF primitives, its current state of integration into the OS, as well as its use in an emerging collection of crypto-safe primitives in RIOT.

Slides