Random numbers are widely used for information security, cryptography, stochastic modeling, and quantum simulation. We invent a scalable, robust, energy efficient method for physical true random number generation using a semiconductor laser chip of volume less than 1 mm cubed. Ultrafast emission fluctuations are digitized to produce hundreds of random bit streams in parallel. We achieved a total bit rate of 250 Tb/s, which is 100 times higher than the fastest physical random bit generators.

As we rely more on digital networks, the ability to generate a large number of high-quality random numbers at fast speed becomes increasingly important and urgent for cybersecurity. Most random numbers used nowadays are generated by computers, but they are pseudo random and vulnerable to attacks. True random numbers are generated by sampling real physical phenomena, but the generation rate is low and the cost is high. Our method breaks the record of single random bit stream generation rate, and produces hundreds of independent bit streams in parallel. We design a special laser geometry to vastly increase the number of lasing modes and utilize their interference to create ultrafast intensity fluctuations. Stochastic quantum noise makes these fluctuations unpredictable and non-reproducible. Since our method is based on a stand-alone laser, its implementation is extremely simple yet highly efficient, and the performance is robust against fabrication defects and external perturbations.

Tags: Lasers, Random Number Generation, Cybersecurity