Biometric Electroencephalogram Based Random Number Generator

Abstract

Cryptography is a tool to protect messages confidentiality through codes keys. Random numbers are crucial in the process of making the cryptography keys in which these keys should be unpredictable. EEG, with its chaotic properties, could be selected as one type of noise source for a random number generator where it is unique to an individual and its characteristics are impossible to be faked or compromised hence unpredictable. A raw EEG signal is random but not uniform because it is contaminated with artefacts which tends to produce spikes in the raw signal causing non-uniformity hence potential to be predicted. Therefore, Hilbert Transform is proposed to be added into the randomness extraction algorithm in this paper. There are a total of 112 random number sequences resulting a total of 113,770,496 bits being generated in this paper. These sequences successfully scored 99.40% of success rate through the NIST SP 800-22 statistical test suite which is proven to be random and cryptographically secured.