PRIVATE LIXE DATA TRANSMISSION 1485 



ous channels. Thus, the multichannel system B is really worse off, by 

 that small amount, than system A. This may amount to some 3 or 4 db 

 in ten or twelve channels. 



There are occasions where crosstalk into other facilities sets the level 

 permitted for the signal. It may be that concentration of the signal onto 

 a single carrier aggravates this interference, as compared with that from 

 a multi-carrier signal. In such cases system A may be penalized by a few 

 dl), as compared with system B. 



Impulse noise shows correlation among the phases of its spectral 

 components. Thus a noise pulse of voltage amplitude A^ in each channel 

 of system B, cumulates to voltage amplitude lOiV in system A or 20 db 

 greater. On the other hand, a signal of amplitude S in each channel of 

 system B, cumulates to an RMS voltage amplitude -v/lO S, or 10 db 

 greater, over the total system (plus a peaking correction which may be 

 positive or negative as just mentioned). Thus, the single channel A sys- 

 tem is at a disadvantage of 10 db, less the peaking adjustment, with re- 

 spect to the B system. 



A further adjustment may be needed, because a single noise peak that 

 affects all 10 channels of B, or 10 bits, may affect only one bit of A. 

 This adjustment depends upon the word grouping of bits which is used. 

 It may be neglected if all the 10 bits of B are in the same word, and if, 

 in one word, an error of 10 bits is effectively no worse than an error of 

 1 bit. 



Single-frequency noise lies at the opposite extreme of the gamut from 

 impulse noise. The vulnerability of a signal pulse to single-frequency 

 noise varies according to the relationship between the frequencies of the 

 noise and of the signal carrier in the utilized signal band. 



The pattern of sensitivity to noise over an individual channel can be 

 expected to be about the same for a narrow band as for a wide-band 

 channel. Thus the pattern of sensitivities in the single channel of system 

 A is repeated in each channel of system B on a 10 times finer frequency 

 scale. The required S/N ratio in any one channel of B remains the same 

 as that for A. 



In sj'stem B each of the ten channels must put out onh' one tenth of 

 the power of the single channel of system A (less the correction for peak- 

 ing which as before may be positive or negative). Thus any one of the 

 ten channels of B is 10 db (plus a peaking correction) more vulnerable 

 to single frequency noise than the single channel of A. 



It must be noted that there are occasional special circumstances where 

 the single frequency noise may be persistent and steady. The multi- 

 channel system B may in such cases have an advantage in permitting 



