Elastic All-Optical Networks: A New Paradigm Enabled by the Physical Layer. How to Optimize Network Performances?

A new era for optical networks has been started by the development of transceivers based on digital signal processing and operating multilevel modulation formats with coherent and equalized receivers, enabling uncompensated optical links and dynamic transparent wavelength routing in nodes. Connectivity matrices at the physical layer have changed from sparse and static to full and elastic. Therefore, to fully exploit network potentialities and to implement the elasticity paradigm requested by the forecasts of worldwide Internet traffic evolution, the orchestration of logical and physical layer has become a firm requirement. We propose the statistical network assessment process (SNAP) as a method to explore potentialities of physical layer independently of specific traffic allocation. We apply SNAP to a German and to a Pan-EU network topology, first deriving the average bit-rate per lightpath given an any-to-any connectivity matrix-a static metric-in order to compare merits of three typical fiber types and to evaluate benefits of time-division hybrid modulation formats. Then, assuming physical layer characteristics are set, we perform a SNAP investigation versus progressive network loading evaluating the blocking probability versus total allocated traffic-a dynamic metric-comparing merits of different size in traffic grooming operating together with multi- or fixed-rate transceivers. We also show how the progressive loading analysis can be used to address selected physical layer upgrades. As an example, we investigate possible upgrades to hybrid Raman/Erbium fiber amplification on selected links, then displaying quantitative benefits in terms of enabled excess traffic at a given blocking probability.