Ultrafast Physical Random Number Generation Using Chaos

Ultrafast true random number generators are ideal for data encryption, Monte Carlo testing, and other data transport applications. Torch"s team has extensive experience designing, building, and testing oscillators based on the innovative, hybrid, exactly-solvable, chaotic oscillator theory as developed by Corron, Blakely and Pethel. These oscillators are capable of generating provably-chaotic waveforms that yield true random numbers with exactly-computable entropy. Our previous work with these oscillators yielded high frequency oscillators having chaotic waveforms that can be efficiently encoded with message bits, passed through a noisy channel and decoded with a chaos matched filter at the receiver. In this proposal we describe a plan for designing, building, and testing true random number generators based on this innovative theory to operate in the chaotic regime and provide ultrafast bit throughput. The plan has four parts: 1) design studies to identify feasible approaches for hardware realization, 2) circuit modeling and simulation for evaluation of feasible designs, 3) fabrication and testing to verify performance in terms of bit randomness and bit rate, and 4) redundancy reduction algorithm techniques for potential post processing. Our target application is stream ciphers. We envision our best designs will provide ultrafast true random bits in a robust, provable way.