Download Architecture, Operation and Services for the Next-generation Flexible-grid Elastic Optical Networks Book in PDF, Epub and Kindle
Optical spectrum is becoming scarce with traffic demand growing at 40% per year. Today, the optical spectrum of a fiber is divided using wavelength-division multiplexing (WDM) into optical channels of 50 GHz or 100 GHz. This fixed grid will not be able to efficiently support higher bit-rate transponders (namely, at 400 Gbps and 1 Tbps). Moreover, innovative elastic transponders capable of tuning their data rates by choosing appropriate modulation formats or spectrum width are also being investigated. To exploit the capabilities of elastic transponders and to fit transponders at beyond-100 Gbps rates, the conventional fixed grid has to evolve towards a flexible grid, where an arbitrary number of finer frequency slots (e.g., at 12.5 GHz) can be assigned to serve the client demands. At the same time of the disruptive development of the optical layer, two trends in the general networking/IT fields are prominent: 1. Vertical convergence: convergence of multiple layers of the network stack, e.g., packet and circuit (optical) convergence; 2. Horizontal convergence: convergence of computing, storage, and networking resources, e.g., cloud networking, and information-centric networking (ICN). Hence, we are motivated to study how to adapt the architecture, operation, and services of flexible-grid elastic optical networks to the two prominent trends. For vertical packet/optical convergence, we study dynamic traffic grooming in elastic optical networks. We propose to jointly solve the electrical-layer routing and optical-layer routing and spectrum assignment (RSA). Also, we propose a spectrum reservation scheme that can efficiently utilize the bandwidth variability of lightpaths, reducing operational expenditure (OPEX) as well as increasing spectrum efficiency. We also propose a provisioning policy exploring grooming opportunities both in time and frequency domains in order to achieve high energy efficiency. We review the evolution of traffic-grooming paradigm. Sliceable optical layer based on sliceable transponders and bandwidth-variable reconfigurable optical add/drop multiplexer (BV-ROADM) is identified as a novel technology that could impact the future grooming paradigm by offloading considerable amount of traffic and part of electronic grooming function to the optical layer. We propose two novel network architectures based on sliceable optical layer and their optimal designs. It is found that packet over sliceable (PoS) network architecture consumes the fewest transponders and at the same time achieves either the "lowest-possible" latency or least spectrum usage. We also propose a new dense-wavelength-division multiplexing (DWDM)-centric converged metro/aggregation network, which reduces the usage of electronic packet processing, hence significantly reducing the corresponding OPEX. Both Integer Linear Program (ILP) and heuristics are proposed to address the complex network design problems involving two layers of routing (fiber routing and wavelength routing), wavelength assignment, and survivable design. For horizontal convergence, we investigate how to jointly allocate computing, storage, and networking resources for virtual infrastructures in the problems of network virtualization over both WDM and flexible-grid optical networks. We formulate the problems as mixed integer linear programs (MILP) and propose two heuristics, namely MaxMapping and MinMapping. Numerical examples show that MinMapping performs very close to the optimal results derived by the MILP in both kinds of optical networks, by exploring traffic grooming. Also, it is verified that flexible-grid optical networks can be more spectrum efficient than WDM networks as the substrate for network virtualization. Also, for large service providers (SP) who own their computing, storage, and networking resources, we propose to orchestrate their cache and networking resources in a coordinated way to solve their traffic-engineering (TE) and cache-orchestration (CO) problems in an ICN-enabled networking infrastructure using service popularity. We propose to deal with the joint problems at the service level, in which no specific information about individual contents of services is needed. Numerical results show that our proposed TE and CO schemes can significantly reduce network cost and increase cache hit ratio, with the constraints of satisfying service-level agreements (SLA).