Software-Defined Fragmentation-Aware Elastic Optical Networks Enabled by OpenFlow

We present a software-defined fragmentation-aware elastic optical network, in which an extended OpenFlow-based control plane intelligently routes connection requests to avoid spectrum fragmentation. The overall feasibility and efficiency of the proposed scenario is validated by using both numerical simulation and experimental demonstration. Introduction Optical networks are evolving from a fixed ITU-T DWDM wavelength grid to a flexible grid in which the optical spectrum is divided into fixedsize spectrum slots (6.25 or 12.5 GHz each). The required spectral resources for a connection request in such a flexible grid or Elastic Optical Network (EON) are adaptively allocated by assigning the necessary number of contiguous spectrum slots according to the traffic bit rate and modulation format. However, the setup and release of connections in a dynamic network scenario can fragment the optical spectrum into noncontinuous small pieces. Such a spectrum fragmentation problem may lead to inefficient resource utilization and a high blocking probability if not dealt with properly. To address this issue, several algorithms for spectrum defragmentation have been proposed recently. Despite significant progress, it should be noted that all previous studies have focused on algorithm design. The issue regarding how to deploy a defragmentation mechanism in a real operational scenario by using a network control plane has not been addressed yet, especially through an experimental approach. To this end, in this paper, we present a software-defined approach enabled by OpenFlow for deploying a fragmentation-aware EON. All the proposed algorithms and OpenFlow protocol extensions are validated by using network simulation and experimental demonstration, verifying their overall feasibility and efficiency for spectrum defragmentation. Network architecture The network architecture for a software-defined fragmentation-aware EON is shown in Fig.1. The data plane is deployed with bandwidthvariable wavelength cross-connects (BV-WXC), which are implemented by using BVwavelength selective switches (BV-WSS). Each BV-WXC is controlled by an OpenFlow agent, which is referred to as an OpenFlow-enabled BV-WXC (OF-BV-WXC). A dedicated OpenFlow controller (e.g. NOX) is introduced to control all the OF-BV-WXCs through the extended OpenFlow protocol (as detailed next). Thanks to the centralized architecture of OpenFlow, the NOX controller is able to know, and can dynamically manage spectrum slots usage information (i.e., free or occupied) for all the links in the network. This information can be utilized by the NOX to perform a fragmentationaware Routing and Spectrum Allocation (RSA) algorithm to intelligently route the incoming requests to avoid spectrum fragmentation. Fig. 1: Network architecture and the assessment of spectral fragmentation To quantitatively assess the spectrum fragmentation in a EON, several metrics have been proposed . In this paper, we use the parameter “Cut”, which is a nonnegative integer (e.g. 0, 1, 2...) accounting for the number of consecutive spectrum that a new connection will break. For example, as shown in Fig.1, an incoming connection from WXC1 to WXC4 for We.3.E.2.pdf