RELIABILITY OF RADIO TELEPHONE CIRCUITS 193 



Duration of Interruptions 



In the traffic operation of radio telephone circuits we are concerned 

 in a practical way with the effect of transmission improvement upon 

 the duration of intervals when circuits are unsatisfactory as well as 

 upon the percentage lost or unsatisfactory circuit time. Obviously, 

 the reduction of lost time must be accompanied by a reduction of the 

 length of unsatisfactory intervals, or what may be termed interruptions 

 to service. It is of interest to know how the distribution of inter- 

 ruptions of various lengths may be expected to vary with improve- 

 ment of a circuit. 



In Fig. 7 the upper curve shows how the number of circuit interrup- 

 tions for the year as indicated by the ordinate value varied with hours 

 duration on the transatlantic short-wave radio telephone circuits 

 during 1930.^ This upper curve was obtained from traffic data. In 

 determining the points shown it was necessary to exclude all interrup- 

 tions of uncertain length that occurred at the beginning or end of the 

 periods when circuits were in use so that only the slope of this curve is 

 significant. 



From the field intensity measurements obtained regularly on the 

 transatlantic circuits it is possible to obtain a useful check on the 

 traffic experience. For example, the interruptions may be defined as 

 the intervals of time during which no signal could be heard by beating 

 in the carrier received on a short-wave measuring set to an audible 

 tone with a local oscillation. By this means signals 30 db or more 

 below those required for a barely satisfactory radio telephone circuit 

 can be heard. In Fig. 7 the lower curve shows the distribution of 

 interruptions based upon such a standard. This curve has sub- 

 stantially the same logarithmic slope as the one obtained from traffic 

 experience. That is, the slope remains the same when the conditions 

 defining the point of interruption are shifted by perhaps 30 or 40 db. 



The summation of all interruptions shown by curves such as those 

 in Fig. 7 should agree with the observed lost circuit time. If it is 

 assumed, as the rather meager evidence cited above appears to indicate, 

 that the logarithmic distribution of the interruptions remains constant 

 for different circuit conditions, it is possible to show how the probable 

 number of interruptions of various lengths will vary with transmission 

 improvement. With the slope shown the position of curves corre- 

 sponding to different db improvements is established by the reliability 

 curve of Fig. 6 and the requirement that the summation of interrup- 

 tions equal the lost time. Curves obtained in this manner are shown 



^ Interruptions during other years have been too infrequent to provide depend- 

 able data. 



