On recovering missing values for discrete time signals with finite sets of spectrum degeneracy

Abstract

The problem of recovery of missed values for discrete time processes is considered in a pathwise setting without probabilistic assumptions. A new class of transfer functions is proposed for recovery of finite sets of missed values. It is shown that error-free and uniform recoverability is feasible for classes of square-summable sequences with Z-transform vanishing with a mild rate at periodically located isolated points. Some robustness with respect to noise contamination is established for the suggested algorithm.