We live in an increasingly connected world, generating new applications supported by new devices such as smartphones, tablets, smart TVs and sensors. The Internet of Things represents a major challenge for the future, with over 50 billion connected devices expected by 2020. Machine-to-machine data streams are another hurdle: increasing and continuously growing data traffic places stress on network and DC architecture. Traffic volume is expected to increase in both access and metro networks. By 2019, it is expected that Metro traffic will be 2.5 times the volume of 2014, topping at 168 Exabyte per month. The same applies to Access Networks.
More and more data is going into Data Centres. However, inside Data Centres the situation is even more critical: it is widely accepted that one bit of information generates five bits of traffic inside datacentres.
Furthermore, a vast amount of data is being exchanged within and between datacentres, due to new services and big players such as Google, Facebook and Amazon. Data Centre construction could result in increased long-haul capacity. This would make it possible to scale the capacity between Data Centre farms (also related to SDN and network virtualization), implement redundancy (data replication between remote locations) whilst improving Quality of Service as well as reliability and routing options.
SAFARI project and Optical Crosstalk
Scientists from the University of Southampton's Optoelectronics Research Centre (ORC) are working with Denmark’s Coriant and DTU Technical University and Japan’s NTT and Fujikura on the SAFARI project. SAFARI (Scalable And Flexible optical Architecture for Reconfigurable Infrastructure) focuses on the development of programmable optical hardware that can provide 400Gbit/s per channel. The resulting scalable and flexible high-speed optical transport networks will utilise multicore optical fibres with space division multiplexing.
The team will develop the required optical transport technology and work on ways of providing networking flexibility and controlling networks with SDN (software defined networking). They will also work on combining SDN with ultra-high capacity fibre links to connect to the physical layer. Multicore fibres will each run at a capacity near that of today’s single mode fibre systems. To date, the highest number of cores has been 19, but the team hopes to add many more. However, optical crosstalk can lead to signal degradation. ORC is developing optical amplifiers that operate on each core and across the C band (1530 to 1565nm) which may be extended into the L band (1565nm to 1625nm).
Optical Crosstalk is only an issue when multiple cores and/or modes are used inside one single fibre with space division multiplexing technology. This comes on top of issues usually present wherever standard single-core, single-mode fibre is used. The ‘standard’ impairments depend on the final applications (short reach vs. long reach), how tolerant the system can afford be with respect to these and whether we can (in term of electronics and noise) compensate for them in a cost-effective way in relation to the application (for details see box 1),
Realising higher data rates
Current solutions increasingly point towards a solution for tackling capacity requirements by proposing a disruptive technology. Increased capacity can be addressed in terms of density (capacity per surface area), in term of costs (CAPEX and OPEX), by power efficiency (Watts/bit/second) and finally in terms of spectral efficiency (Bit/s/Hz).
Single mode fibre: Development of new transmission formats and enhanced electronics/signal processing performance can lead to new fibre for high-capacity networks (which has ultra-low loss attributes and large resilience to nonlinearities). On another hand, the number of fibres increases capacity. Reduced diameter optical fibre (e.g. 200µm) allows very high capacity optical cable with a limited cable diameter. This helps reuse existing congested ducts and avoids laying very costly new ducts.
Spatial division multiplexing: The multicore approach: instead of having multiple fibres, several cores are embedded in a single fibre for space saving. Capacity is multiplied by the number of cores (<10). However, several practical issues seem impossible to overcome (manufacturing yields and costs, installation in the field, cable performance), so there are no real cost savings with regard to active infrastructure.
Mode division multiplexing (aka few-mode fibres): here, specially designed fibre can carry several channels (embodied by modes) within the same optical fibre core. Depending on the distance reach, the maturity level is different. For short reach application (DataCom, Data Centre application), the solution is viable and can be envisaged on medium term basis (5 years) as all required components are available.
For long reach applications, there is a lack of commercially available components with established reliability, which are required to complete essential functions such as few-mode amplifiers, or other optical network signal routing (add/drop). Such structures are not anticipated yet at mid-term. For few-mode fibre, capacity increases are in the order of 10-50, depending on the fibre type.
Challenges presented by high data rates
Migration to higher bitrates (i.e. from 10G to 100G) can be efficiently achieved thanks to the development and adoption of signal processing features such as DSP (Digital Signal processing), FEC (Forward-error correction), new modulation formats (High-order Amplitude/Phase Modulation) and Coherent detection schemes.
Scenarios for 400G are more demanding and require bridging the gap on the extra OSNR budget margin and imposing constraints on the networks, particularly in the case of transport networks. The reduction of distance reach can lead to the reconsideration of signal regenerator site locations. While it is possible to gain extra OSNR budget by implementing more advanced FEC circuits, new amplification schemes could be also adopted, (vital for long-haul applications where power supply may be limited such as in the Amazon forest or Africa) as well as possibly using a more resilient fibre type (similar to that used in submarine application). Right now, equipment network providers are addressing these questions. Some concerns about the lack of equipment interoperability may still exist.
To that view, some new fibres for terrestrial applications are being developed in line with the ITU-T G.654 Recommendation. Among those, the Prysmian Group Logline fibre can help bridge the OSNR gap thanks to its extremely high effective area of 120 μm² - 50 % higher than standard single-mode fibre, which dramatically reduces the nonlinear effects offering the possibility of higher power and consequently higher distance.
As for FEC, there are still some drawbacks when improving their performance in term of power consumption, latency (due to processing time of the FEC may introduced up to few 10s µs delay), which can be problematic for real-time applications, and more complex implementations (may even be proprietary).
400Gb/s @ Prysmian Group
Various 400Gb/s-related developments are taking place within Prysmian Group, especially in the field of MMF fibre and Few-mode fibres. We aim to develop solutions for tackling the capacity requirement explosion in the next decade, forge solid partnerships, and eventually launch innovative products. A summary of current developments:
Schematic of the WDM OM4 WideCap fibre. WDM = Wavelength Division multiplexing
Optical fibre: development of the 1st Wideband MMF, called WideCap-OM4, allowing for wavelength division multiplexing over a 100nm window with OM4 grade transmission quality. WideCap-OM4 multimode fibre delivers OM4 performance in the 850-950nm window while maintaining compatibility with current multimode fibres. Traditional OM4 fibres offer high bandwidth in a narrow band cantered at 850nm. To satisfy the exponentially increasing information demand in Data Centres, the capacity of WideCap-OM4 has been extended to longer wavelengths up to 950nm. WideCap-OM4 and multi-wavelength transceivers are a viable solution for future 100 and 400Gbps multi-wavelength systems. WideCap-OM4 incorporates BendBrightxs technology to withstand tight bends and cabling challenges in the data centre. We expect to see cost reduction benefits for deployment compared to existing cabling solutions (1 WideCap-OM4 fibre can replace up to 4 OM4 fibres) and bitrate upgrade thanks to the development of multi-wavelength VCSELs.
Left: Single mode Fibre Right: Multi-Core Fibre
Left: Multimode mode Fibre Right: Few-Mode Fibre
Although Prysmian Group carried out some successful Multi-Core-Fibre manufacturing trials, Multi-Core Fibres are not considered as promising as Few-Mode Fibres
Intensity Profile of the modes
Top left: So called “fundamental mode” which is only employed for
single mode fibre. The other modes are higher order mode and can be
employed in Few-mode fibres
Prysmian Group is very active in this new optical fibre field, thanks to R&D efforts, and has published several papers at top conferences demonstrating new few-mode prototypes and reporting world record transmissions in collaboration with industry leaders.
Prysmian is fully equipped to tackle this new few-mode domain thanks to our proprietary PCVD manufacturing platform, which is the most reliable and suitable manufacturing process for few-mode fibres required for long-reach application as for standard Multimode fibres.
In short
The need for 400Gb/s is driven by global bandwidth growth, which in turn is the result of new applications and technologies and the multiplication of smart devices (Internet of Things). 400G is currently anticipated for backbones and high capacity datacentre interconnects. As we and the industry come closer to a practical solution for realising this, we will keep you posted…
For short reach, the price must be low, and linear impairment such as chromatic dispersion can be an issue if there is too much allocation. The rest is a matter of components specifications.
For long reach (transport networks), the price is higher, so we can plug more technologies to overcome transmission impairment. When moving to higher bitrates, several transmission impairments must be considered due to the more stringent penalties impact caused by the nonlinear response of the transmission medium (optical fibre), the passive/active components throughout the optical path. All of the impairments are expressed in terms of optical crosstalk expressed as the optical signal to Noise Ratio (OSNR), which governs the maximum spectral efficiency, which can be supported within the communication channel. Generally speaking, when moving to higher bitrate, the requirement on the OSNR budget increases (OSNR has to be higher) and unless new enabling technologies are implemented, it results in significantly reduced network reach. See next question for existing solutions.
Right now, the implementation of 400Gb/s is still challenging and demanding. Numerous terrestrial 400G field trials since 2013, by achieved using single (1X400G) or multiple sub-carrier transmission (4x100G). Major Telco Operators have carried out 400G field trials on their existing networks (G.652) or new projects. This trade-off between modulation technique, channel size, and OSNR requirements is at the heart of current 400G research efforts.
Standardisation
Three standardization bodies play a significant role/influence for common industry specifications to be adopted at required levels for the component, module up to the carriers: IEEE, ITU and OIF.
The 400Gb/s has been being studied by IEEE 802.3 Ethernet study group starting from March 2013, to deal with market demand in light of the expected bandwidth need growth, looking for inputs from industry actors (servers, data centre networks, carrier operators…) on a worldwide basis.
For Ethernet-based interconnect, several physical layer specifications were proposed to support various link distance reaches:
- 400GBASE-SR16: ≥100 m over MMF
- 400GBASE-DR4: ≥ 500 m over SMF
- 400GBASE-FR8: ≥ 2 km over SMF
- 400GBASE-LR8: ≥10 km over SMF
All of these physical layer specifications consist of aggregating low-speed links in multiple of 25G / 50G or 100 G to create a virtual higher-speed link. To enable 4x100G lanes, new modulation formats are needed replacing the standard NRZ format like Pulse-Amplitude Modulation (PAM). Standards are expected by 2017.
The ITU-T is responsible for defining the recommendation to support transport of such 400Gb/s bitrates across optical transport networks.
The OIF is in charge of coordinating by addressing interoperability and proposal general architecture among the industry players. To that purpose, the OIF has issued a 400G Framework Document back in August 2015.
PUBLICATIONS FOR FURTHER READING
Wideband OM4
- First report of a Wideband OM4 over 100nm: D. Molin, F. Achten, M. Bigot-Astruc, A. Amezcua-Correa, P. Sillard, “WideBand OM4 Multi-Mode Fiber for Next-Generation 400Gbps Data Communications,” European Conference on Optical Communications, ECOC’14, paper P.1.6 (Cannes, France, 2014)
- First report of a Wideband OM4 over 150nm M. Bigot-Astruc, D. Molin, F. Achten, A. Amezcua-Correa, P. Sillard, “Extra-Wide-Band OM4 MMF for Future 1.6Tbps Data Communications,” Optical Fiber Communication conference, OFC’15, paper M2C.4 (Los Angeles, USA, 2015)
- 100Gbps transmission over 200m of WideCap-OM4: J. M. Castro, R. Pimpinella, B. Kose, Y. Huang, B. Lane, A. Amezcua, M. Bigot, D. Molin, P. Sillard, “200m 2×50Gbps PAM-4 SWDM Transmission Over Wideband Multimode Fiber using VCSELs and Pre-distortion Signaling,” Optical Fiber Communication conference, OFC’16, paper Th2G.2 (Anaheim, USA, 2016)
- 180Gbps transmission over 300m of WideCap-OM4: S.M.R. Motaghiannezam, I. Lyubomirsky, H. Daghighian, C. Kocot, T. Gray, J. Tatum, A. Amezcua-Correa, M. Bigot-Astruc, D. Molin, F. Achten, P. Sillard, “180 Gbps PAM4 VCSEL Transmission over 300m Wideband OM4 Fiber,” Optical Fiber Communication conference, OFC’16, paper Th2G.2 (Anaheim, USA, 2016)
Few-Mode Fibres
- Tutorial paper: P. Sillard, “Few-Mode Fibers for Space Division Multiplexing,” Optical Fiber Communication conference, OFC’16, Th1J.1 (Anaheim, USA, 2016)
- First 6-spatial-mode PON transmission: H. Wen, C. Xia, A.M. Velázquez-Benítez, N. Chand, J.E. Antonio-Lopez, B. Huang, H. Liu, H. Zheng, P. Sillard, X. Liu, F. Effenberger, R. Amezcua-Correa, and G. Li, “ First Demonstration of 6-Mode PON Achieving a Record Gain of 4 dB in Upstream Transmission Loss Budget,” Journal of Lightwave Technology, DOI: 10.1109/JLT.2015.2503121
- Record Spectral efficiency over a single-core fiber: H. Chen, R. Ryf, N. K. Fontaine, A.M. Velázquez-Benítez, J.E Antonio-López, C. Jin, B. Huang,M. Bigot-Astruc, D. Molin, F. Achten, P. Sillard, R. Amezcua-Correa, “High Spectral Efficiency Mode-Multiplexed Transmission over 87-km 10-Mode Fiber,” Optical Fiber Communication conference, OFC’16, paper Th4C.2 (Anaheim, USA, 2016)
- Longest 10-spatial-mode transmission over a 50μm MMF: J.J.A. van Weerdenburg, A.M. Velázquez-Benítez, R.G.H. van Uden, J.E. Antonio-López, P. Sillard, D.Molin, M. Bigot-Astruc, A. Amezcua-Correa, F.M. Huijskens, F. Achten, H. de Waardt, A.M.J. Koonen, R. Amezcua-Correa, C.M. Okonkwo, “10 Spatial Mode Transmission over 40km 50μm Core Diameter Multimode Fiber,” Optical Fiber Communication conference, OFC’16, paper Th4C.3 (Anaheim, USA, 2016)