A statistical analysis of random noise signals for Johnson noise thermometry

The results of a statistical analysis on the Johnson noise signals, equivalently band pass filtered random signals are described in this paper. We determined an optimal sampling time needed to extract the most information from the cross power spectral density for a given interval of frequency band, showed that the band pass filtered sensor signal and the channel noises are statistically uncorrelated in long term averages, estimated the number of signal blocks for the long-term averages required to meet a desired accuracy, estimated the amount of time required for processing signals to obtain the desired accuracy, and finally showed how accurate is the linearity of the processed signal