Superchannels, flex-grid, multilayer switching key developments for next-gen transport networks
By GEOFF BENNETT and SERGE MELLE
To meet the dramatic growth in Internet demand, the optical industry must find new methods to increase the total capacity of existing fiber networks and ensure these new technologies are economically efficient, operationally simple, and scalable. It's only via such combinations that network providers can continue scaling network bandwidth while limiting infrastructure investments.
Coherent transmission technology, which combines advanced amplitude/phase modulation with sophisticated digital signal processing, enables 100-Gbps ultra-long transmission with about 10X the fiber capacity compared to 10-Gbps transmissions using intensity modulation with direct detection. But to meet the dramatic growth in bandwidth demand, service providers need to be able to turn up more than 100 Gbps at a time. The industry response has been the introduction of coherent superchannels operating at 500 Gbps, in which multiple coherent carriers are digitally combined to create an aggregate channel of a higher data rate on a single high-density line card that can be deployed in one operational cycle.
The operational deployment of 500-Gbps superchannels in service-provider networks and demonstrations of future superchannels having 1 Tbps or more of capacity will drive further development of WDM technologies. We'll review three of these development areas: the expansion of flexible coherent modulation capabilities, a new flexible WDM grid (flex-grid) that maximizes WDM capacity and operational ease-of-use, and a tiered multilayered photonic and digital switching architecture.
Flexibility is key
A key innovation enabled by coherent modulation is the ability – simply using software – to tune the type of modulation to the application in question. Early "flex-coherent" modulations supported both QPSK modulation for long-haul terrestrial transport and BPSK for ultra-long-haul or submarine applications. Future flex-coherent modulation capabilities will further broaden the spectrum of software-tunable modulation formats, enabling, for example, 16QAM modulation that will provide 2X the capacity per carrier (at the expense of optical reach and thus targeted for use in metro applications). Further expansion of flex-coherent modulation will enable service providers to deploy a single type of WDM line card for a range of applications and use software to enable operationally simple tuning to the reach and application required.
The first generation of flexible modulation 500-Gbps superchannels used a WDM grid with fixed channel spacing to ensure backward compatibility with existing line systems. But as more sophisticated flex-coherent modulation technologies become available, more flexible means will be required to manage the WDM optical spectrum. So the old fixed-grid world may no longer be the most efficient way to build an optical network.
In Figure 1, each of the spectra shown is a 100-Gbps single-carrier transmission. On the left, we see a conventional PM-QPSK transmission will fit into a channel spacing of about 45 GHz and is nicely compatible with existing 50-GHz fixed-grid systems and associated reconfigurable optical add/drop multiplexers (ROADMs). In the center is a 100-Gbps PM-16QAM wavelength, which only consumes about half the spectrum of a 100-Gbps PM-QPSK wavelength and thus would fit well into a 25-GHz channel spacing. But on the right is a "pulse-shaped" PM-16QAM wavelength. Pulse shaping applies Nyquist filtering at both the transmitter and receiver, with the goal of eliminating the intersymbol interference between channels. Since the shaped pulse spectrum is now much narrower than the 22 GHz of a normal PM-16QAM signal, instead of basing the grid spacing on individual wavelengths, it would be more practical to look at the overall superchannel spectral width and build a flexibly sized channel to match. The shaded areas indicate the wasted spectrum for the shaped pulse spectrum.
To alleviate this waste and enable manufacturers a way to build "superchannel-friendly" devices, the ITU-T has defined a flexible grid in the latest WDM wavelength grid specification, G.694.1. The new flex-grid defines WDM channels having a granularity of 12.5 GHz, combined with the ability to define an aggregate superchannel spectral width of N × 12.5 GHz, to accommodate any combination of optical carriers, modulations, and data rate. In addition, the N × 12.5-GHz spectral width of a flex-grid superchannel can be tuned, enabling rapid changes when the operating specifications of the superchannel are varied.
Figure 2 shows how this flex-grid would work. On the left of the spectrum is a conventional 100-Gbps PM-QPSK signal, with no pulse shaping and with a spacing of about 45 GHz. This fits into a flexible channel built from four 12.5-GHz units. In the center is a pulse-shaped PM-QPSK superchannel of 500-Gbps capacity and with a spectral width of about 375 GHz. This would be supported using a flexible channel built from 30 12.5-GHz units. On the right is a pulse-shaped, PM-16QAM superchannel also of 500-Gbps capacity. As expected, this occupies a flexible channel made up of only 15 12.5-GHz units.
The bottom line is that the huge boost in fiber capacity enabled by coherent technology combined with the application of superchannels for data rates beyond 100 Gbps enables operational scaling by allowing service providers to turn up optical capacity in larger increments with the same effort. The move to more advanced modulation formats such as 16QAM can further double fiber capacity – for shorter reach metro applications – while use of a flexible WDM grid will deliver another 25% of optical capacity by allowing the subcarriers in WDM superchannels to be squeezed more closely together.
A switch in time
The question then becomes how to best manage these superchannels in typical service-provider networks. Traditionally, WDM networks operated with single-carrier wavelength channels and used ROADMs to do photonic-only switching of these wavelengths. Since the majority of end-user client services remain 10 Gbps or less, muxponders are used to aggregate these services in a point-to-point fashion onto 100-Gbps wavelengths.
However, many studies have shown that photonic-only switching can be inefficient, because ROADMs used with 100G muxponders do not provide any ability to groom or switch subwavelength traffic within or between wavelengths. That can lead to low utilization of deployed bandwidth and therefore over-deployment of 100G wavelengths, also termed the "muxponder tax."1 In addition, as network traffic volumes increase and more and more wavelengths are used, it becomes increasingly difficult to find unused end-to-end wavelengths across the network. That creates what is called "wavelength blocking" and requires the use of incremental and otherwise unnecessary regens for wavelength conversion to eliminate the problem.2
To alleviate this problem, networks have been built with converged WDM and switching, where digital subwavelength grooming is integrated with WDM optics to enable highly efficient packing of WDM waves. Sometimes called "digital ROADMs" or "integrated WDM/OTN," this approach minimizes the deployment of costly WDM optics, eliminates all wavelength blocking, and can reduce network capital expenditures (capex) by approximately 30%.3
As wavelength channels evolve from single-carrier 100G to multicarrier 500G superchannels, however, digital grooming of every superchannel may not be necessary or desired. Once digital grooming has fully filled a 500-Gbps long-haul superchannel, it's most cost-effective for that channel to optically express through intermediate add/drop nodes and terminate only at its end destination. In this context, ROADMs enable superchannels to be easily switched and reconfigured to maximize operational flexibility, while the integrated WDM/switching platforms maximize bandwidth and capex efficiency.
Thus, to maximize efficiency, next generation optical transport networks should deploy network nodes supporting multilayer switching, which combines the benefits of both photonic ROADM switching and integrated digital OTN switching, as shown in Figure 3. Network studies have shown that such multilayer switching can reduce the total number of WDM ports by 13–22% compared to all-optical or all-digital switching alone.4 To enable the capacity scaling made possible by 500-Gbps superchannels and flexible WDM grids, next-gen ROADMs will therefore need to support flex-grid technologies to allow reconfigurable tuning of superchannel spectrum allocation and location on the WDM grid.
Going forward, next generation optical transport networks will need to make the most use of the flexibility and scaling enabled using advanced coherent modulation technologies. More advanced flexible coherent modulation will support a wide range of modulations tailored to specific applications, capacities, and reaches. Flexible WDM grids will enable more efficient and flexible use of optical spectrum to maximize capacity. And a tiered multilayer photonic and digital switching architecture will provide the most cost-effective and efficient network architecture for ultra-high-bandwidth scaling.
References
- M. Bertolini, et al., "Benefits of OTN Switching Introduction in 100Gbit/s Optical Transport Networks," OFC/NFOEC Conference, Session NM2F, Los Angeles, March 4–8, 2012.
- S. Melle and V. Vusirikala, "Network Planning and Architecture Analysis of Wavelength Blocking in Optical and Digital ROADM Networks," OFC/NFOEC Conference, Session NTuC, Anaheim, March 25–29, 2007.
- Deore, et al., "Total Cost of Ownership of WDM and Switching Architectures for Next-Generation 100Gbit/s Networks," IEEE Communications Magazine, vol. 50, no.11, pp. 179–187, November 2012.
- S. Roy, et al., "Evaluating Efficiency of Multi-Layer Switching in Future Optical Transport Networks," March 17–21, 2013 proceedings of OFC/NFOEC Conference in Anaheim.
GEOFF BENNETT is director of solutions and technology and SERGE MELLE is vice president, technical marketing at Infinera Corp.
Past Lightwave Issues