A computer simulation study of real-zero interpolation

The technique of real-zero interpolation (RZI), first described by Voelcker, permits the interpolation of a minimum-bandwidth waveform to a set of zero crossings on the time axis. RZI can thus lead to bandwidth compression of intrinsically 2-level signals, or to the conversion of analog signals into a type that is impervious to instantaneous non-linear distortion. The purpose of the simulation study described was to examine how RZI performs in a realistic communication environment, i.e., when the transmission channel is noisy, has a finite dynamic range, and when zeros can occur only at discrete points on the time axis. Attention is focused on 2-level facsimile as the most likely candidate for RZI processing, but other types of signal such as synchronous data and synchronous pulse-width modulated waves have also been studied. It is shown that RZI can yield substantial bandwidth saving for signals that carry information by their zero crossings only. The price is paid in increased vulnerability to channel noise; this may limit the applicability of RZI to channels with over 35 to 40 dB signal-to-noise ratio. Excessive amplitude swings, intrinsic to the system, can be controlled by the insertion of complex zeros. RZI may lead to a bandwidth saving of the order of 6:1 for 2-level facsimile signals. Its data handling performance is characterized by the measured rate of 1.0 to 1.6 bit/Hz. An application to direct analog encrypting of speech signals by multiplicative distortion is also described.