Description (eng)
With the rise of Internet of Things (IoT) devices, security challenges have increased. While various authentication methods exist, most of them rely on storing access information in non-volatile memory. However,
if an IoT device is physically compromised, these credentials may be easily stolen.
Static Random-Access Memory Physical Unclonable Functions (SRAM PUFs) provide an innovative approach to generating cryptographic keys. They rely on the irregular sub-microscopic variations in SRAM
to produce a unique digital fingerprint for each chip, making it impossible to copy and ideal for device authentication. However, due to intra-chip-variability, the fingerprint changes slightly with each use and can
also be affected by external factors and temporal instability. Overcoming these challenges entails creating a
secure and reliable authentication system without storing any access information on the device.
This master’s thesis introduces an SRAM PUF-based authentication system for IoT, tested and designed
specifically for the ESP32 microcontroller. Differing from other authentication systems proposed in the literature, this system uses a bit selection method instead of error correction methods and relies on Application
Programming Interface (API) keys rather than cryptographic methods. The use of the API key-based system
as an additional factor to Mutual TLS (mTLS) authentication provides an additional security factor and can
be seamlessly integrated into existing authentication systems.
As part of the preparations, extensive long-term measurements were carried out. The influences of different ambient temperatures and power supplies on the SRAM were evaluated, and the differences in the
uninitialised SRAM values between microcontrollers of the same series and between different series were
analysed. A detailed literature review also provides a comprehensive overview of the IoT and Physical Unclonable Function (PUF) technologies, specifically focusing on SRAM PUFs and error correction methods.
The results of this master thesis provide a solid foundation for future research and development in SRAM
PUF-based security mechanisms.