Abstract
Random time functions or random numbers having any assigned probability distribution function are generally used in the Monte-Carlo method, simulation problems and other related fields. Therefore it is very important to generate high-speed purely random numbers.
The binary random numbers which are most fundamental are generated from physical fluctuation phenomena or M-sequences. However, these methods are inadequate to generate highspeed binary random numbers, because of the frequency-band limitation of physical noise or of the periodicity of M-sequences.
This paper presents a new method of generating high-speed binary random numbers by combining physical noise and M-sequences. 100kc/s binary random numbers are obtained by sampling the 1Mc/s M-sequence with random pulses which are generated by shaping thyratron noise clipped at a certain voltage level. The auto-correlation function and transition probability of the binary random numbers and the probability distribution function of the pulse-interval of the random pulses are measured. It shows that the properties of 100kc/s binary random numbers are almost purely random.
Finally, key points of speeding up the binary random numbers obtained by this method are discussed.
