Abstract
It has often been found more necessary in the engineering of data transmission systems to consider impulsive noise than conventional Gaussian white noise. A model is proposed for the impulsive noise, which describes empirically an amplitude distribution and a time distribution. Because the latter has been described in experimental work principally in terms of error occurrences, the description is translated into these. The notable characteristics of impulsive noise are that at low occurrence frequencies the amplitudes are much larger than for Gaussian noise, and that impulses or errors tend to be more "bunched" than expected from a Poisson distribution.

This publication has 4 references indexed in Scilit: