414 BELL SYSTEM TECHNICAL JOURNAL 



The average of all those measurements showing no effect of steering 

 is 8.0 decibels compared with the theoretical figure of 10 log 6 = 7.8 

 decibels, for random addition. The random assumption employed in 

 the analysis is thus justified on the average. 



Summarizing Discussion 



The aim of the signal-to-noise study described in this section has 

 been not so much to evaluate the intrinsic merit of the experimental 

 MUSA system as to compare its behavior with a simple theory. The 

 element of research has been to find out how well the transatlantic 

 waves fit into the background of the simple theory. To this end some- 

 what artificial devices, thermal noise and rectified carrier, were sub- 

 stituted for static and speech signals. 



The study included an analysis of diversity action in which the effect 

 upon the signal-to-noise ratio of (1) delay equalization, (2) detector 

 characteristics, and (3) automatic gain control action was displayed 

 prominently. Although those effects, taken together, are important, 

 they are individually small and could be separately evaluated only by 

 locally controlled test methods. 



We propose now to review the results of the tests and studies. The 

 slight decrease in improvement which would have occurred had the 

 speed of the automatic gain control been reduced, and the increase in 

 improvement which would have resulted had noise power been aver- 

 aged do not affect the fundamental considerations and will be neg- 

 lected here. 



One wave bundle received with one branch (Table V) yielded an 

 improvement of 7 decibels which is about one decibel less than 10 log N 

 = 7.8 decibels calculated by simple theory (Fig. 34, Case I). Com- 

 ments relating to this have been given. 



One wave bundle received with two branches steered on each side of 

 the center of the bundle yielded eight decibels improvement (Table V). 

 Of this, one decibel is due to primary diversity action and three 

 decibels are due to the secondary diversity gain. This three-decibel 

 gain was assumed in the simple theory although it was not then 

 designated as secondary diversity gain. It accrues by virtue of voltage 

 addition of the signals and power addition of the noise. Both of these 

 conditions are satisfied in Table V. There remain four decibels which 

 represent the signal-to-noise improvement in each branch, referred to 

 one antenna. This indicates a loss of three decibels as compared with 

 a single branch steered at the center of the bundle, which gives seven 

 decibels; this is reasonable when it is remembered that the branches 

 were steered apart by a phase difference of about 60 degrees {4>a — <i>B 



