A digital hardware realization of a random number generator

A digital random number generator that generates number sequences with either uniform or Gaussian distributions has been constructed in TTL digital hardware for use as a peripheral device to a DDP-516 computer. The hardware design is based on an algorithm described by Rader, Rabiner, and Schafer [1]. Approximations to the Gaussian distribution are generated by summing
NA + 1
consecutive numbers from the uniform sequence, where
NA
(the number of additions) is specified by the user and is in the range
0 \leq NA \leq 213
. Measurements of its autocorrelation function, power spectrum, and amplitude histograms indicate that the generator produces uncorrelated samples with amplitude distributions that are very close to the desired uniform or Gaussian distributions. For generating Gaussian numbers, the hardware generator operates at least 600 times faster than an equivalent software realization.