Noise Processes in Discrete Communication Systems
This chapter focuses on noise processes in discrete communication systems. The problem with white Gaussian noise process discretization is that a strict definition implies that the noise has theoretically infinite power. Thus, it would be impossible to generate discrete noise, because the sampling theorem requires that the sampled signal must be physically realizable, that is, the sampled noise needs to have a finite power. To overcome this problem, noise entropy is defined as an additional measure of noise properties, and a truncated Gaussian probability density function is used. Adding entropy and truncated density to the definition of the noise autocorrelation and power spectral density functions allows mathematical modelling of the discrete noise source for both baseband and bandpass noise generators and regenerators. Noise theory and noise generators are essential for a theoretical explanation of the operation of digital and discrete communications systems and their design, simulation, emulation, and testing.