The Blackwell channel is a deterministic broadcast channel model used in coding theory and information theory. It was first proposed by mathematician David Blackwell. In this model, a transmitter transmits one of three symbols to two receivers. For two of the symbols, both receivers receive exactly what was sent; the third symbol, however, is received differently at each of the receivers. This is one of the simplest examples of a non-trivial capacity result for a non-stochastic channel.
The Blackwell channel is composed of one input (transmitter) and two outputs (receivers). The channel input is ternary (three symbols) and is selected from . This symbol is broadcast to the receivers; that is, the transmitter sends one symbol simultaneously to both receivers. Each of the channel outputs is binary (two symbols), labeled .
Whenever a 0 is sent, both outputs receive a 0. Whenever a 1 is sent, both outputs receive a 1. When a 2 is sent, however, the first output is 0 and the second output is 1. Therefore, the symbol 2 is confused by each of the receivers in a different way.
The operation of the channel is memoryless and completely deterministic.
The capacity of the channel was found by S. I. Gel'fand. It is defined by the region:
1. R1 = 1, 0 ≤ R2 ≤
2. R1 = H(a), R2 = 1 - a, for ≤ a ≤
3. R1 + R2 = log2 3, log2 3 - ≤ R1 ≤
4. R1 = 1 - a, R2 = H(a), for ≤ a ≤
5. 0 ≤ R1 ≤, R2 = 1
A solution was also found by Pinkser et al. (1995).