A parallelizable chaos-based true random number generator based on mobile device cameras for the Android platform

Abstract

True random number generators are used in high security applications such as cryptography where non-determinism is required. However, they are slower than their pseudorandom counterparts because they need to extract entropy from physical phenomenon. To overcome this drawback, generators have been designed to extract unpredictability from devices such as computer processing units or microphones. This paper introduces a new generator for the Android mobile platform based on images captured by a built-in camera. Although similar generators exist, they suffer from poor performance and a lack of proper security evaluation. The proposed generator implements a chaos-based postprocessing algorithm that eliminates statistical defects and increases its throughput. These goals are achieved by using the inherent properties of a chaotic system to amplify entropy extracted from the captured images. The proposed generator is evaluated in two phases: first, statistical test suites are executed to identify statistical defects. Next, the generator’s forward and backward security is analysed. Results indicate that the proposed true random number generator is able to generate statistically secure true random number sequences faster than existing mobile-based generators. In addition, the generator is designed to support parallel processing, thus allowing its performance to scale according to the mobile device’s multicore architecture.