SELECTIVE FADING OF SUPER-HIGH FREQUENCY SIGNALS 1197 



400-mc observed band was 8 db less than mid-day normal; and in one 

 case it was more than 10 db less than mid-day normal. 



Another family of curves is shown by Fig. 6. These curves were ob- 

 tained by dividing the 400-mc observed band into a number of 20-mc 

 bands, chosen at random with regard to the shapes of the path-loss 

 versus frequency curves. These data show the difference between the 

 maximum and minimum losses within a single 20-mc broad-band channel 

 which might accompany a fade of a given depth. Such data are of use in 

 estimating possible distortions of a modulated signal occupying a band 

 width of 20 mc. 



Frequency Diversity 



The fact that the instantaneous fading may be different on different 

 frequencies within the same frequency range offers a means for mitigating 

 transmission impairments caused by fading. During periods when there 

 is fading in excess of a specified value on the regular carrier frequency, 

 the carrier can be shifted to an alternate frequency in the hope that 

 the fading on the alternate frequency may be less severe. The merit of 

 using this type of frequency diversity can be gauged from the statistical 

 distribution of fading on the alternate frequency during periods when 

 there is fading in excess of the specified value on the regular frequency. 



The data obtained in these tests indicates that there was no correlation 

 between the fading on frequencies separated by: (1) 40 mc or more 

 during periods when there was fading of 10 db on one of the frequencies 

 and (2) 160 mc or more during periods when there was fading of at 

 least 20 db on one of the frequencies. However, during periods of severe 

 fading, the data indicate considerable correlation between the fading on 

 regular and alternate frequencies separated by 80 mc or less. 



Fig. 7 shows the distributions of fading on alternate frequencies at 

 specified frequency separations from the regular frequency, when there 

 is fading of a specified depth on the regular frequency. Table II indicates 

 which of the curves on Fig. 7 applies to a particular set of conditions. 



Curves G and H on Fig. 8 show the distributions of depths of fade 

 at 4190 mc during the entire months of July and August, respectively. 

 Comparison of the 4,190-mc data with data from tests made over other 

 paths indicates that the fading occurring during the Iowa tests was nearly 

 as severe as during the "worst month" observed on any path to date. 



Curve B on Fig. 7 shows the statistical distribution of fading on any 

 frequency (within the range under consideration) during periods when 

 important fading is prevalent. The fading shown by this curve is con- 



