A Method for Generating True Random Numbers with Multiple Distribution Characteristics

In this paper, a method of generating true random numbers obeying multiple distribution characteristics is proposed. First, two resistance-capacitance (RC) self-excited oscillation circuits are used to generate jittered, periodically unstable square wave signals, and then high-precision high-frequency quartz crystal oscillator sample measure these jittered signals obtain their periods. Due the randomness external environment, there Gaussian white noise in circuit, so periods variables subject distribution. Finally, we use field-programmable gate array (FPGA) perform secondary processing on which can number sequences that conform distribution, Poisson 0-1 uniform etc. The generated by circuit be applied field intelligent control related automatic machine learning. This uses physical entropy source high-quality sequences, easy build, low cost, small size. experimental results show low-cost, highly reliable integrate, providing an effective solution for generation electronic systems.