Autumn 2013 Optical Connections Magazine

Sponsoredby:

ADVA

Software-defined optical networks – Transforming the optical layer into a programmable resource

reconfiguration. Benefits of such an approach include fewer hardware variants, lower equipment costs, and an increased network efficiency. Multiple sub-carriers can be bundled to deliver higher aggregate capacities. Flexible wavelength grid technology allows signals to occupy contiguous strands of optical spectrum with an aggregate bandwidth of n x 12.5GHz, which is particularly important for services requiring more bandwidth than today’s 50 GHz slots. As these concepts are very similar to those used in software-defined radio (SDR), we refer to them as software- definedoptics (SDO). A recent study showed that SDO has the potential to save more than 40% of spectrum resources in a 400 Gb/s backbone network [1]. While a single optical super-channel is most spectrally efficient, large contiguous spectral strands can lead to an increase in wavelength blocking. Having the flexibility to split an aggregate signal across multiple optical strands is thus an important tool to minimize orphan bandwidth and optimize network resource usage. Figure 1 illustrates how two 400 Gb/s signals can be transported making use of different available spectral resources. While it is apparent that optical communications is closely mimicking approaches previously adopted inmobile communications, transmission capacities

as integral parts of their optical infrastructure and therefore require resource control and allocation on a network-wide scale. And Going Beyond The marriage of SDO with software- defined networking (SDN) gives rise to what we call a software-defined optical network (SDON). A SDON turns the optical network into a programmable resource under centralized control. While software controlofelectricalpacketandcircuit networksisrelativelystraightforward, the control of transparent optical networks is more complicated due to their analogue optical nature [2]. Fortunately, mature control plane and path computation engine implementations for wavelength- routed optical networks already exist which can be readily extended to also cover SDO. A control approach is desirable which hides optical layer complexity and allows an abstracted representation and sharing of its network resources. Network abstraction itself can happen at different levels. In the simplest case, network boundaries are defined at the electrical client interfaces. The network operator is in full control of all optical equipment functions. He offers his clients a virtual circuit-switched Network as a Service (NaaS), which they can access over pre-defined attachment circuits. The clients will still be able to

see an abstracted internal network topology which they can use for their path calculation. Optical layer details, though, are hidden from them and may even be changed at the discretion of the network owner. Extending known alien wavelength concepts, a logical extension of the NaaS approach is to eliminate the electrical interfaces at the network boundaries and directly provide Optical Spectrum as a Service (OSaaS). A simple yet practical example is a carrier who builds a new coherent express layer and wants to share his infrastructure cost with one or more partners by “licensing” them parts of his optical spectrum. A more sophisticated scenario, in which optical signals can transparently pass through multiple optical domains, has recently been demonstrated using an SDN-controller for inter-domain coordination [3]. While clients still would not need to see all optical layer details, an information exchange on usable spectral resources, the signal format, and signal performance would still be necessary. Summary SDO transceivers, a flexible coherent express layer, and SDN-assisted network control are the key building blocks to transform the optical layer from a static network infrastructure into a programmable network resource. Together, they help to improve network efficiency, allow higher levels of automation, and facilitate the development of new network services such as OaaS. Jorg-Peter Elbers VP of Advanced Technology at ADVA Optical Networking [1] A. Autenrieth, et al., “Will Flexgrid Networks be Worth the Investment for just 30% Improvement?“, OSu1F workshop presentation, OFCNFOEC 2013 [2] J.-P. Elbers, et al., „Extending Network Virtualization into the Optical Domain“, paper OM3E.3, OFCNFOEC 2013 [3] M. Channegowda, et al., “First Demonstration of an OpenFlow based Software-Defined Optical Network Employing Packet, Fixed and Flexible DWDM Grid Technologies on an International Multi-Domain Testbed”, paper Th.3.D.2, ECOC 2012

By Jorg-Peter Elbers F ollowing low-loss fibres in the 1970s and EDFAs in the 1990s, DSP-enabled coherent transceivers are the latest disruptive innovation in long-haul optical communications. Propelled by steady traffic growth1 and advances in CMOS technologies, massive digital signal processing led to a renaissance of coherent optical transmission. The user benefits are substantial: Coherent 100 Gb/s transceivers deliver a ten-fold increase in DWDM system capacity over conventional 10 Gb/s technology. At the same time, they drastically simplify operations with adaptive electronic equalization that eliminates optical dispersion compensation, makes system performance more reproducible, and significantly eases transmission design. What we are seeing now though, is only the beginning. Learning from Mobile Networks For decades, optical line interfaces were designed to operate at fixed data rate and bandwidth. With the latest DSP technology, transceivers are becoming software-programmable and can adapt the data rate, modulation format, forward error correction and electronic signal equalization to the needs of the application. By exploiting the 100 Gb/s ecosystem and using the same optics and RF electronics, an optical carrier can support 50 Gb/s (BPSK), 100 Gb/s (QPSK), 150 Gb/s (8QAM) and 200 Gb/s (16QAM) speeds at different reaches by simple DSP

and distances in optical core networks are orders of magnitude higher than in mobilenetworks.Theneedto deal with restricted spectral resources, programmable modulation,andaggregation over multiple frequency sub- bands are common themes in both domains. Yet, there are fundamental differences, too. While wireless communication is limited to the links between user equipment and the base stations in a particular cell, optical networks are multi- hop and meshed in nature. They comprise of optical amplifiers and ROADMs

Figure 1 - SDO Cockpit illustrating the delivery of two different 400Gb/s services

1 According to the latest Cisco Visual Networking Index, global IP traffic is growing at 23% CAGR from 2012-2017.

| Optical Connections 2013 | www.opticalconnectionsnews.com | 29

Made with FlippingBook - Online catalogs