No. 1017

Title:

Attainment of the capacity of nonwhite Gaussian channel with feedback

Author(s):

Masayuki, Kumon (Risk Analysis Research Center, Institute of Statistical Mathematics)

Key words:

Channel capacity, Choleski decomposition, entropy, mutual information, spectral factorization, stationary process, Wiener filter.

Abstract:

    This paper treats the capacity of nonwhite Gaussian channel with feedback under the average power constraint, and clarifies the structure of the pair of the optimal message and the optimal feedback law which attain the capacity. By the two-step power saving method, the optimal feedback law is at first explicitly obtained, and then a remarkable fact is shown that the deterministic message attains the capacity. The capacity with feedback is related to the capacity without feedback under an increased power. When the channel noise are ARMA(1, 1), MA(2), ans some special cases of ARMA(k, k) processes, this relation gives the closed-form expressions for calculating the capacity.