Journal of Mathematical Sciences: Advances and ApplicationsAuthor's: Lei Lin
Pages: [17] - [30]
Received Date: June 10, 2017
Submitted by:
DOI: http://dx.doi.org/10.18642/jmsaa_7100121841
We present an algorithm to compute the capacity of compound channel, which is specified by maximin of mutual informations. The algorithm utilizes the alternate optimization method similar to Arimoto-Blahut algorithm except for the more complex optimization method at the second phase. By transforming the original problem with non-differentiable objective function into a convex optimization problem with differentiable objective function, we obtain the necessary and sufficient conditions for the capacity-achieving input distributions. We also derive both upper and lower bounds of the maximin value, which provide a criterion to terminate any iterative methods. In addition, the convergence of the algorithm is proved. Finally, the numerical results are presented.
Arimoto-Blahut algorithm, convex optimization, compound channel, maximin of mutual informations.
