Finite state predictors for Gaussian sequences

https://doi.org/10.1016/S0019-9958(78)91039-2Get rights and content
Under an Elsevier user license
open archive

A finite state predictor for a Gaussian sequence with known power spectrum may be obtained by quantizing the optimal linear predictor. We analyse the structure, memory, and prediction error of the predictor by combining information theory and methods from linear filtering. The mutual information between the past and the future of the sequence provides a useful estimate of the number of bits of storage in a good predictor.

Cited by (0)