A Study on Energy Harvesters for Physical Unclonable Functions and Random Number Generation

As the broad implementation and use of wireless sensor nodes in Internet of Things (IOT) devices increase over the years, securing personal data becomes a growing issue. Physical unclonable functions (PUFs) and random number generators (RNGs) provide methods to generate security keys for data encryption. Transducers used in the energy harvesting systems of wireless sensor nodes, can generate the PUFs and RNGs. These transducers include piezoelectric devices (piezo), thermoelectric generators (TEG) and solar cells. This research studies the electrical properties of transducers at normal and low operating levels for electrical responses that can be used in PUF generation and random number generation respectively.

The PUF generation discussed in this study analyzes the resonance frequency of 10 piezos, and the open-circuit voltages of 5 TEGs and 5 solar cells. The transducers are tested multiple times over a 10-day period to evaluate PUF reproducibility and reliability characteristics. The random number generation is accomplished by applying a low-level vibration, thermal or light excitation to each respective transducer. The generated electrical signals are amplified and digitally processed and analyzed using the National Institute of Standards and Technology (NIST) Statistical Test Suite.

The experiment results for the PUF generation are promising and indicate that the piezos are the better choice due to their stable frequency output. Each transducer was able to produce random numbers and pass the NIST tests, but the TEGs passed the NIST tests more often than the other transducers. These results offer a preliminary basis for transducers to be used directly in security applications.