On advantages of data driven traffic classification for dynamic routing in optical networks

In this paper, we concentrate on solving the dynamic routing problem with traffic classification using the Software Defined Elastic Optical Networks framework. We first define three categories of traffic based on the CISCO Traffic Reports. Then, we propose two algorithms to solve the Routing, Modulation and Spectrum Assignment problem with the support of traffic class recognition - the Split Spectrum and the Buffer. We evaluate our approach using representative network topologies and a wide range of simulations. The results clearly show that the Butter algorithm provisions efficiently the incoming traffic in the network. In addition, we show the reduction of request blocking for the highest priority traffic that cannot be interrupted.