Calculation of multidisturber crosstalk probabilities — application to subscriber–loop gain

The possibility that excessive crosstalk limits the application of gain to long voice-frequency loops was studied. This problem was examined by means of a probability model incorporating data on such random variables as activity coincidence of disturbing and disturbed subscribers, disturber volume level, coupling path loss, listener acuity, and various noise sources. A new approach was used with regard to the probability distributions of random variables: the probability distributions were modelled in detail for the calculation of the necessary functions, sums, and functions of sums of random variables. Results showed that gain of 6 dB or less is acceptable, and 9 dB is unjustifiably excessive. The approach used provides the most accurate calculation possible with available data, and is anticipated to be convenient not only for similar crosstalk evaluations but also for other nongaussian probability problems.