Noise generators for the simulation of digital communication systems

The generation of random numbers, having uniform and Gaussian distributions, with application to the time-domain simulation of communication systems is considered. The specific goal is to identify appropriate noise generation algorithms for use in system simulations for the determination of the symbol error probability of a digital communication system. Several methods of generating both uniform and Gaussian random numbers are examined. The algorithms considered were taken from those discussed in recent literature that appeared most appropriate for the authors' problem. The algorithms were programmed in FORTRAN for use on a PC with math-coprocessor. Tests to evaluate the statistical properties of the generators were performed and the results are presented.< >