Prognostic Epsilon Entropy of a Gaussian Message and a Gaussian Source

It is shown that the nonanticipatory and prognostic epsilon entropies and message generation rates for Gaussian messages and sources are realized in input and output messages that together form Gaussian processes. Representations of these processes are found for Markov messages and sources, and analytical expressions are given for the epsilon entropy and message generation rate.