Appraisal of a Random Bit Generator Utilizing Smartphone Sensors as Entropy Source

TL;DR

This paper evaluates the potential of smartphone sensors, specifically video and audio, as entropy sources for a random number generator, achieving full success in NIST tests and highlighting the importance of proper data collection methods.

Contribution

It introduces a novel smartphone-based entropy source for random number generation and demonstrates its effectiveness through successful statistical testing.

Findings

Smartphone sensors can provide viable entropy sources for randomness.

Proper data collection methods are crucial for generating strong random numbers.

Cosmic radiation was initially considered but replaced by video and audio sources.

Abstract

We aim to access entropy sources available within smartphones in order to construct and evaluate a random number generator which is competitive in comparison with existing and proven random number generators. A prototype utilizing the herein proposed algorithm shall generate data that can be tested against the Statistical Test Suit provided by NIST. Although our initial intention of using cosmic radiation failed, we were able to extract randomness from incoming video and audio sources. We found that it is possible to access these sources of entropy utilizing sensors from smartphones resulting in 15 out of 15 successful passed tests within the Statistical Test Suit. We also found that wrong methods of sensor data collection using our prototype eventually generates weak random numbers and fails NIST's Statistical Test Suit. Finally, we suggest that in order to reach the initial goal of providing a smartphone-based true nondeterministic random number generator the detection of muons shall be researched.