THE RESULTS OF TESTING TEL 4 IN A REAL-TIME, PCM/FM CONFIGURATION.

A discussion of the way in which noise introduces errors in a PCMFM link is followed by a description of test results. A summary of the test procedure is also included. The effect of time is reported for approximate periods of one day, three days, thirty days and six months. Performance comparisons are made between P-band and S-band operation, for several bit rates, and for two different message structures. All tests were performed, simultaneously, with a demultiplexer that uses a sampling detector and a second demultiplexer that uses an integration detector. Performance data were computer processed to accomplish a regression analysis for each set of test conditions. By this method a best-fit estimate of performance was obtained along with a measure of data precision. This systematized procedure permits a large number of measurements to be analyzed quickly, without human error or prejudice. Discussion includes some considerations of data-point scatter and suggests the use of a Poisson distribution for relating the number of errors counted to the probability of erroneous measurements. Author