A cognitive decision system for heterogeneous reconfigurable optical networks

The European Union FP7 CHRON project addresses the challenge of controlling and managing the next generation of heterogeneous optical networks supporting the Future Internet. For that aim, the CHRON project proposes a Cognitive Heterogeneous Reconfigurable Optical Network, which observes, acts, learns and optimizes its performance. The core element of such network is the Cognitive Decision System, which makes decisions on how to deal with traffic demands and network events by using knowledge acquired from previous experience and considering the current network status. In this paper, we present the architecture of this Cognitive Decision System, and then show a set of methods developed for two of its building blocks: the Virtual Topology Design module and the Quality of Transmission (QoT) estimator module. Regarding the former module, we present two multiobjective genetic algorithms to solve the impairment-aware virtual topology design problem, and evaluate their performance under realistic traffic demands. Results show that one of the methods, which is enhanced with additional cognition, leads to better results. For the latter module, we present a cognitive QoT estimator, which achieves more than 99% successful classifications of optical connections into high or low QoT categories, and is much faster for on-line operation than an existing approach.