Gaussian Multiterminal Source-Coding with Markovity: An Efficiently-Computable Outer Bound

Abstract

We provide a method for outer bounding the rate- distortion region of Gaussian distributed compression problems in which the source variables can be embedded in a Gauss- Markov tree. The outer bound so obtained takes the form of a convex optimization problem. Simulations demonstrate that the outer bound is close to the Berger-Tung inner bound, coinciding with it in many cases.