Skip to Main Content
This paper presents some observations concerning the type of variability one might encounter in determining error rates in digital data transmission systems subjected to impulse noise. The effects of pulse shape and duration are investigated. It is shown that for signal-to-noise ratios (SNR) greater than -5 dB, a variation of 8 to 1 in the average number of errors per pulse might be encountered as pulse shape is varied. For pulsewidths in the range 0.1 to 4 ms, a variation of 6 to 1 in the average number of errors, per pulse at a given SNR could be expected. Experimental findings relating error rates to pulse duration were supported by analytical studies. The probability of making an error, given that a noise burst has occurred, is determined and a worst case analysis is made, from which a critical SNR is found such that if the noise is below this threshold, no errors can occur. It is recognized that the limitations imposed by the equipment available for study results in a loss in generality. However, the results presented are indicative of the type of variations one might encounter in the case of actual impulse noise on an existing practical system.