Restoration design in IP over reconfigurable WDM networks

摘要

Today, large IP backbone networks are usually deployed directly over sequences of point-to-point WDM systems or chains of newer ROADM-based WDM systems, interconnected by OEO regenerators. The next generation of long distance opticaI network is evolving to an ultra long haul (ULH) architecture based on multi-degree ROADMs. The major drivers of this evolution are reduction of OEO regeneration cost, the ability to remotely reconfigure connections, and enabling of automatic restoration. In this paper we study how to provide reliable IP and high bandwidth ("wavelength") services by designing a restorable, evolved IP-ovcr-Optical network. We propose two novel restoration schemes: The first is 2-Phase Fast Reroute with an optimized Traffic Engineering algorithm for the IP layer; the second is shared mesh restoration using standby (pre-conligured) connections for the optical layer. They both have the ability to achieve sub-second restoration. Also, by maximizing sharing among potential, non-simultaneous failures, each method achieves the desired objective function to minimize needed network capacity at its own layer (IP or OL). To expand the objective function to minimize the sum of network cost over both layers, we also propose an integrated IP-over-oplical layer restoration strategy that enables even more sharing of restoration capacity among non-simultaneous failures at different layers. Simulation results demonstrate significant improvements using our proposed schemes when compared to existing methods.