Chaotic Oscillations in Josephson Junctions for Random Number Generation

We investigated chaos generation in Josephson junctions by numerical simulations for the purpose of applying it to physical random number generators. In the simulations, an RCSJ model by the current bias was used. Current biases were applied to the Josephson junctions and calculated the time evolution signal of the voltage when they were irradiated with an external microwave. To judge whether the chaos signals were generated, Lyapunov exponents, calculated by Shimada-Nagashima algorithm, were used. In the mapping of chaos-generating regions, a certain kind of regularity was observed. The high-frequency component was cut from the acquired chaotic signal, and the simulation of random number generation was done. The autocorrelation function from the time series data was obtained like a delta-function, which suggested that it has nearly perfect random nature. The binary data sequence, which was assumed to be 0 and 1 by the upper and lower sides of a threshold voltage, was verified by a statistical evaluation (NIST SP 800-22).