Provisioning, Recovery, and In-Operation Planning in Elastic Optical Networks

Luis Velasco, Marc Ruiz

• Publisher: Wiley
• Year: 2017
• ISBN:
  9781119340423
• Edition: 1
• Topic: Sciences physiques
• Subtopic: Vagues et mécanique des vagues

Explains the importance of Elastic Optical Networks (EONs) and how they can be implemented by the world's carriers

This book discusses Elastic Optical Networks (EONs) from an operational perspective. It presents algorithms that are suitable for real-time operation and includes experimental results to further demonstrate the feasibility of the approaches discussed. It covers practical issues such as provisioning, protection, and defragmentation. It also presents provisioning and recovery in single layer elastic optical networks (EON). The authors review algorithms for provisioning point-to-point, anycast, and multicast connections, as well as transfer-based connections for datacenter interconnection. They also include algorithms for recovery connections from failures in the optical layer and in-operation planning algorithms for EONs.

Provisioning, Recovery and In-operation Planning in Elastic Optical Network also examines multi-layer scenarios. It covers virtual network topology reconfiguration and multi-layer recovery, and includes provisioning customer virtual networks and the use of data analytics in order to bring cognition to the network. In addition, the book:

• Presents managing connections dynamically—and the flexibility to adapt the connection bitrate to the traffic needs fit well for new types of services, such as datacenter interconnection and Network Function Virtualization (NFV)
• Examines the topic in a holistic and comprehensive way, addressing control and management plane issues for provisioning, recovery, and in-operation planning
• Covers provisioning, recovery, and in-operation planning for EONs at the proposed exhaustive level

The rapid expanse of new services has made the use of EONs (a relatively new concept) a necessity. That's why this book is perfect for students and researchers in the field of technologies for optical networks (specifically EONs), including network architectures and planning, dynamic connection provisioning, on-line network re-optimization, and control and management planes. It is also an important text for engineers and practitioners working for telecom network operators, service providers, and vendors that require knowledge on a rapidly evolving topic.

1
Motivation

Luis Velasco and Marc Ruiz

Universitat Politècnica de Catalunya, Barcelona, Spain

1.1 Motivation

The huge amount of research done in the last decade in the field of optical transmission has made available a set of technologies jointly known as flexgrid, where the optical spectrum is divided into 12.5 GHz frequency slices with 6.25 GHz central frequency granularity, in contrast to the coarser 50 GHz in fixed grid Wavelength Division Multiplexing (WDM) [G694.1]. Such frequency slices can be combined in groups of contiguous slices to form frequency slots of the desired spectral width, thus increasing fiber links’ capacity. To illustrate the magnitude of the capacity increment, a 40 Gb/s connection modulated using Dual‐Polarization Quadrature Phase Shift Keying (DP‐QPSK) can be transported on a 25 GHz slot in flexgrid, instead of 50 GHz needed with WDM [Ru14.1].

In addition to increasing network capacity, subsystems currently being developed will foster devising novel network architectures. These are as follows:

• Liquid Crystal on Silicon (LCoS)‐based Wavelength Selective Switches (WSS) to build flexgrid‐ready Optical Cross‐Connects (OXCs) [Ji09].
• The development of advanced modulation formats to increase efficiency, which are capable of extending the reach of optical signals [Ge12].
• Sliceable Bandwidth‐Variable Transponders (SBVTs) able to deal with several flows in parallel, thus adding, even more, flexibility and reducing costs [Sa15].

The resulting flexgrid networks will allow mixing optical connections of different bitrates (e.g., 10, 40, 100 Gb/s), by allocating frequency slices and using different modulation formats such as 16‐State Quadrature Amplitude Modulation (QAM16) or DP‐QPSK more flexibly. Furthermore, larger bitrates (e.g., 400 Gb/s or even 1 Tb/s) can be conceived by extending the slot width beyond 50 GHz. In addition, the capability to elastically allocate frequency slices on demand and/or modify the modulation format of optical connections according to variations in the traffic of the demands allows resources to be used efficiently in response to traffic variations; this is named as Elastic Optical Networks (EONs) [Na15], [Lo16].

To understand how the finer spectrum granularity together with the flexible slot allocation of EONs can impact the network architecture, let us compare national Multi‐Protocol Label Switching (MPLS) network designs when the underlying optical network is based either on a WDM network or on an EON [Ve13.1]. MPLS networks typically receive client flows from access networks and perform flow aggregation and routing. The problem of designing MPLS networks consists in finding the configurations of the whole set of routers and links to transport a given traffic matrix whilst minimizing capital expenditures (CAPEX). To minimize the number of ports a router hierarchy consisting of metro routers, performing client flow aggregation, and transit routers, providing routing flexibility, is typically created.

As a consequence of link lengths, national MPLS networks have been designed on top of WDM networks, and, thus, the design problem has been typically addressed through a multilayer MPLS‐over‐WDM approach where transit routers are placed alongside OXCs. Besides, multilayer MPLS‐over‐WDM networks take advantage of grooming to achieve high spectrum efficiency, filling the gap between users’ flows and optical connections’ capacity.

The advent of flexgrid technology providing a finer granularity, however, makes it possible to flatten the multilayer approach and advance toward single layer networks consisting of a number of MPLS areas connected through a core EON (Figure 1.1).

Figure 1.1 Three MPLS areas connected through a core optical network.

To compare the designs of national MPLS networks when such flatten network architecture is adopted, we define the overall network spectral efficiency of the interarea optical connections, as:

(1.1)

where b_aa′ represents the bitrate between two different areas a and a′ in the set of areas A, Δf is the considered spectrum granularity (i.e., 50 GHz for WDM and 12.5 GHz for EON), and B_mod is the spectral efficiency (b/s/Hz) of the chosen modulation format. Note that the ceiling operation computes the number of slices/wavelengths needed to convey the requested data flow under the chosen technology.

Let us analyze the results obtained from solving a close‐to‐real problem instance consisting of 1113 locations, based on the British Telecom (BT) network. Those locations (323), with a connectivity degree of 4 or above, were selected as potential core locations. A 3.22 Pb/s traffic matrix was obtained by considering the number of residential and business premises in the proximity of each location. Locations could only be parented to a potential area if they were within a 100 km radius.

Figure 1.2 plots the amount of aggregated traffic injected to the optical core network for each solution as a function of the number of areas. The relationship between the amount of aggregated traffic and the number of areas is clearly shown; more areas entail higher aggregated traffic to be exchanged because less traffic is retained within an area since the areas are smaller. Nonetheless, the amount of aggregated traffic when all the 323 areas are opened is only 6% with respect to just opening 20 areas.

Figure 1.2 Aggregated traffic and network spectral efficiency.

Plots in Figure 1.2 show the maximum spectral efficiency of the solutions for each technology and the number of areas selected. As illustrated, network spectral efficiency decreases sharply when the number of areas is increased since more flows with a lower amount of traffic are needed to be transported over the core network. Note that the traffic matrix to be transported by the core network has |A|*(|A| − 1) unidirectional flows. Let us consider a target threshold for network spectral efficiency of 80% (horizontal dotted line in Figure 1.2). Then, the largest number of areas are 116 and 216 when WDM and EON, respectively, are selected. Obviously, the coarser the spectrum granularity chosen for the optical network, the larger the areas need to be for the spectral efficiency threshold selected, and, thus, the lower the number of areas to be opened.

Figure 1.3 presents the details of the solutions as a function of the number of areas, while Table 1.1 focuses on the characteristics of those solutions in the defined spectral efficiency threshold for each technology. Note that plots in Figure 1.3 are valid irrespective of the technology selected since no spectral efficiency threshold was required.

Figure 1.3 Details of the solutions against the number of MPLS areas: (a) size of the areas, (b) switching capacity of core MPLS routers, (c) size of the internal data flows, and (d) size of the aggregated data flows.

Table 1.1 Solutions details (network spectral efficiency = 0.8).

		WDM	EON
Number of IP/MPLS areas		1...