Extension of some results for channel capacity using a generalized information measure (Q1101061)

The authors present a new formulation, based on the duality theory for convex programming, for the channel capacity problem. The natural structure of this dual representation is suitable for recent numerical schemes in the mathematical programming literature, and is quite useful for computational purposes and the derivation of bounds. The results are derived in a unified manner by formulating the channel capacity problems as a particular case of a general class of concave programming problems involving a generalized information measure introduced recently by the reviewer and \textit{C. R. Rao} [IEEE Trans. Inf. Theory IT-28, 489-495 (1982; Zbl 0479.94009)]. The results also show that this new information measure of Burbea and Rao can be successfully used to develop a generalized channel capacity theory.