An encoder whose input is a binary equiprobable memoryless source produces one output of rateR₁and another of rateR₂. LetD₁, D₂, and D₀, respectively, denote the average error frequencies with which the source data can be reproduced on the basis of the encoder output of rateRₗonly, the encoder output of rateR₂only, and both encoder outputs. The two-descriptions problem is to determine the regionRof all quintuples(R₁, R₂, D₁, D₂, D₀)that are achievable in thc usual Shannon sense. LetR(D)=1+D log₂ D+(1-D) log₂(1-D)denote the error frequency rate-distortion function of the source. The "no excess rate case" prevails whenR₁ + R₂ = R(D₀), and the "excess rate case" whenR₁ + R₂ > R(D₀). Denote the section ofRat(R₁, R₂, D₀)byD(R₁ R₂, D₀) =\{(D₁,D₂): (R₁, R₂, D₁,D₂,D₀) ∈ R}. In the no excess rate case we show that a portion of the boundary ofD(R₁, R₂, D₀)coincides with the curve(1/2 + D₁-2D₀)(₁₂ + D₂-2D₀)= 1/2(1-2D₀)². This curve is an extension of Witsenhausen's hyperbola bound to the caseD₀ > 0. It follows that the projection ofRonto the(D₁, D₂)-plane at fixedD₀consists of allD₁ ≥ D₀andD₂ ≥ D₀that lie on or above this hyperbola. In the excess rate case we show by counterexample that the achievable region of El Gamal and Cover is not tight.
No takes yet. Share an insight, caveat, or question.
Zhang et al. (1987) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: