Quantization Effects

Abstract

Up to now, we have really been studying discrete-time filters and signal processing, not digital filters and signal processing, because the signals were discrete in time but not in amplitude. Such discrete-time systems can be realized by, for example, switched-ca-pacitor circuits where the amplitudes are represented by voltage or charge and are thus continuously variable. However, the vast majority of discrete-time systems are realized by digital computers or hardware, and the amplitudes of both signals and coefficients are also discrete variables in these systems. Strictly speaking, such systems are nonlinear, but fortunately, as we shall see, these effects can usually be modeled as linear perturbations or additive quantization noise.