Abstract :
Experimental results, supported by theory, indicate that the error liability after regeneration of the output of a limiter-discriminator frequency-modulation radiotelegraph receiver, when fed with steady signals plus noise, can be described in terms of a simple exponential relation involving a single parameter, which characterizes the receiver performance. This parameter, defined as the signal/noise energy ratio required to give an error liability of 1/2¿, is also the amount by which the receiver falls short of the ideal in the non-diversity detection of Rayleigh-fading signals in noise. It is therefore a convenient index of performance. The losses entailed in diversity by selection with this type of receiver as compared with ideal diversity combination, in the reception of fading signals, are determined for 2-, 3-, and 4-path diversity; they are small, but perhaps not insignificant. Test results suggest that the practice, at present almost universal, of specifying receiver performance in terms of telegraph distortion, with an allowance for associated line tails, causes an appreciable waste of channel capacity.
