Suppose I have the following from alphabet $\mathcal{X} = \{0 ,1\}$ to $\mathcal{Y} = \{0 ,1\}$. The channel simply does

\begin{align} 0 \rightarrow 0&\quad \text{with probability 1} \\ 1 \rightarrow 0&\quad \text{with probability 1/2}\\ 1 \rightarrow 1&\quad \text{with probability 1/2} \end{align}

The classical capacity of this channel is given by the mutual information between the input register, $X$ and output register $Y$ maximized over input probability distributions, $p(x)$. That is,

$$C = \max_{p(x)} H(Y) - H(Y|X)$$

The appropriate distribution turns out to be $(0.6, 0.4)$ and the capacity is $0.3219$ bits.

How does one actually use this result to practically communicate (in the asymptotic limit) over this channel at this rate? Does it mean that I should use input codewords that obey these statistics over the input alphabet for example?


1 Answer 1


Your channel is known as the Z-channel. Take a look at Marco Mondelli, S. Hamed Hassani, and Rüdiger Urbanke, How to Achieve the Capacity of Asymmetric Channels.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.