Harvesting Entropy from On-board Sensors of Constrained Devices for Hardening Security of IoT Communication Mechanisms

Abstract

One of the basic fundamentals for the current cryptosystems is high entropy random number generator.
The cryptosystems that been based on the weak random number generators will not provide
adequate level of security. The high entropy random number generators for the Internet of Things
are very hard due to the fact of the inherited limitations of the highly constrained devices. For the
purpose of the random number generation in the constrained environments, four entropy harvesting
methods have been proposed and evaluated. As the potential source of the entropy the on-board integrated
sensors (i.e. temperature, humidity and two light sensors) have been analysed. Additionally
the costs (i.e. time and memory consumption) of the presented approach have been evaluated to
assess the requirements for the Internet of Things applications. The results obtained from the first
method using temperature and both light sensors have been characterized with high Shannon entropy
of around 7.5. The fastest entropy harvesting methods were based on the light sensors. Presented
results indicate that it is feasible to design high entropy true random number generator that have been
based on the integrated on-board sensor that will be fast and require very few memory resources from
the constrained device.