Digital transmission in the presence of impulsive noise

The transmission of digital data over telephone channels has been considered previously in the literature, and the effects of Gaussian noise have been analyzed. With experience, however, it has become apparent that while a background of Gaussian noise is present, the limiting noise is not Gaussian but impulsive in nature. It consists of bursts of high amplitude which occur at random considerably more often than is predicted by the rms value of the Gaussian background noise. Measurements of the statistics of this noise have been initiated, and some results have been reported. In this paper, a model for the noise is constructed to be reasonably consistent with these measurements without becoming too complex to be handled analytically. Various modulation systems are analyzed to determine their performance in such a noise environment. Conditional error rates, in terms of the average number of bit errors per noise burst, are determined as functions of a convenient signal-to-noise ratio which is defined. The systems are ranked as to their performance in such a noise environment, and the ranking is found to be the same as that for Gaussian noise. The improvement to be gained by employing complementary delay networks is investigated. Networks with linear and sinusoidal delay characteristics are considered.