Scalable quantum random number generator for cryptography based on the\n random flip-flop approach

For globally connected devices like smart phones, personal computers and\nInternet-of-things devices, the ability to generate random numbers is essential\nfor execution of cryptographic protocols responsible for information security.\nGenerally, a random number generator should be small, robust, utilize as few\nhardware and energy resources as possible, yet provide excellent randomness at\na high enough speed (bitrate) for a given purpose. In this work we present a\nquantum random number generator (QRNG) which makes use of a photoelectric\neffect in single-photon avalanche diodes (SPADs) as a source of randomness and\nis scalable to any desired bitrate. We use the random flip-flop method in which\nrandom bits are obtained by periodic sampling of a randomly toggling flip-flop.\nFor the first time we investigate this method in detail and find that, out of\ntwo main imperfections, bias is due only to hardware imperfections while\nautocorrelation predominantly resides with the method itself. SPADs are\nintegrated on a silicon chip together with passive quenching and digital\npulse-shaping circuitry, using a standard 180 nm CMOS process. A separate FPGA\nchip derives random numbers from the detection signals. The basic QRNG cell,\nmade of only two SPADs and a few logic circuits, can generate up to 20 Mbit/s\nthat pass NIST statistical tests without any further postprocessing. This\ntechnology allows integration of a QRNG on a single silicon chip using readily\navailable industrial processes.\n