A universal divergence-rate estimator for steganalysis in timing channels

This paper proposes new tools for steganalysis of queue-based stegocodes over covert timing channels. We study a universal estimator for the Kullback-Leibler (KL) divergence rate between the covertext process and the stegotext process. We empirically illustrate the performance of our estimator on some simple queue-based stegocodes and study its convergence properties.