Signalling a high

Oct 22, 2023

Share this on social media:

The next generation of optical networks will need to provide the capacity required to explosive bandwidth demand, and do so without cost or sustainability compromises. Keely Portway finds out how this delicate balance can be struck.

Image credit: MyCreative/shutterstock.com

If the past three years have done nothing else of use, they have unarguably thrown optical communications into the spotlight in a big way. The need for stable, reliable, high-bandwidth networks was hugely emphasised by the need to work and learn at home, stay entertained indoors and spend time with family and friends remotely. But, despite life having finally returned to a degree of normality post-pandemic, the need for high levels of connectivity has not abated. Far from it.

The appetite for bandwidth was actually in existence long before anyone had even heard of Covid, with a significant number of drivers pushing this demand, which the pandemic served simply to accelerate. Speaking at this year's OFC Conference Market Watch panel, Tim Munks, Senior Principal Analyst: High Speed Optics and Optical Network Technology at industry analyst firm Omdia said: "It's not a newsflash, but the drivers for continued bandwidth expansion in networks worldwide include potential business models such as telemedicine, DocuSign, and gaming, all of which are examples of growth providing growth."

Touching on the pandemic, Munks said: "The pandemic accelerated businesses moving to the cloud; and 5G, which had got off to a slow start, has really taken off now. Moreover, investments continue to grow as governments put a lot of money into broadband expansion where broadband connectivity has become seen as a right."

Such investment in fibre deployment across FTTH, metro, subsea, data centres and front- and back-haul networks has been essential, but investment in R&D for the optical technology at the core of these networks is just as crucial in ensuring that they be future-proofed for next-generation applications.

Coherent optics, for example, have become popular thanks to their ability to modulate the amplitude and phase of light, as well as transmit across two polarisations, meaning that more information can travel through a fibre optic cable.

When I first entered the world of optical communications, the 2018 OFC Conference played host to a number of new coherent optical engines featuring digital signal processing (DSP) technology, which hit headlines because they could transport up to 800Gb/s capacity-per-wavelength. The 2023 iteration of the event was the launchpad of many of the next generation of coherent DSPs, which are now able to reach 1.2 and 1.6Tb/s.

The move to terabit represents a substantial leap forward in data transmission rates, allowing networks to better handle unprecedented amounts of data, and enable faster and more reliable communication across long distances. This could have significant implications for telecoms, but also for industries and applications including cloud computing, data centres, and scientific research.

Coherent optical telecoms networks with such high capacities can allow for seamless streaming of high-definition video content, ultra-fast internet access, and uninterrupted voice communication, allowing service providers to meet the ever-increasing demands of their customers, providing a superior user experience with minimal latency and network congestion.

For data centres and cloud computing providers, the move to terabit coherent optical networks could open up new possibilities for efficient data storage, processing, and distribution if implemented correctly, particularly for access and the edge. In scientific research, fields such as astronomy, genomics, and particle physics generate enormous amounts of data that requires fast and reliable transmission. Coherent optical networks with capacities of 1.2 and 1.6Tb/s could enable researchers to transfer and collaborate on large datasets more efficiently, accelerating scientific breakthroughs.

While the move to 1.2 and 1.6Tb/s coherent presents opportunities such as those detailed above, it can also pose some technical challenges. Ensuring signal integrity and reducing noise interference becomes critical at these ultra-high data rates. The design and deployment of advanced optical transceivers, amplifiers, and dispersion compensating modules play a vital role in maintaining the quality and reliability of data transmission. For network operators there is the worry that upgrading existing network infrastructures to support these higher data rates would require significant investments in both hardware and network management equipment.

A great deal of work has been undertaken in the past few years to provide solutions that can help operators overcome these challenges, and the most recent terabit coherent solutions have been heralded as designed with the operators in mind.

One early announcement in this area came from Nokia, with its sixth-generation coherent photonic service engine, the PSE-6s, which is designed to power the next generation of coherent transport at up to 1.2Tb/s and be deployed in pairs to power a 2.4Tb/s coherent transport solution. It operates at 130+ GBd, and leverages a 5nm complementary metal-oxide-semiconductor (CMOS) with silicon photonics for photonic integrated circuits (PICs).

Nokia's sixth-generation coherent photonic service engine, the PSE-6s, is designed to power the next generation of coherent transport at up to 1.2Tb/s (Credit: Nokia)

With the next generation of coherent technology it is hoped that, as well as additional capacity, operators will experience other benefits. Serge Melle, Director Product Marketing at Nokia told Fibre Systems at the OFC Conference: "For operators, scale, performance and sustainability are very important. You can actually combine PSE6 1.2Tb/s together into a single 2.4Tb/s channel, which means that you can then carry three 800GbE services into that one channel, allowing much more capacity-per-fibre than having individual 800GbEs on individual wavelengths, which helps when it comes to scale."

The performance aspect has also been considered, continued Melle: "The performance part is being able to carry 800GbE, not just in the metro, but in regional long-haul networks, south of 2,000km. Being able to operate an 800GbE over a single wavelength can save operators a lot of costs, because they don't have to put that 800GbE over two lower-speed wavelengths, so it reduces the number of optics. If you can reduce the number of optics and leverage the latest silicon and Moore's law, you can reduce power-per-bit and total network power consumption, which is important from the cost and sustainability angle."

"We're seeing the trend for reduced energy consumption more and more," agreed Robert Maher, CTO at Infinera, who also spoke to Fibre Systems at OFC. "Even when I attended a recent subsea conference, there was lots of talk about sustainable subsea deployments. And we have a customer who deployed our ICE4 optical engine on a subsea cable, and they wanted to replace it with ICE6, and the number one reason was power."

Infinera revealed its own next-generation 1.2Tb/s optical engine, the ICE7, just days before OFC opened its doors. ICE7 is designed to benefit operators by helping to drive down cost, space, and power-per-bit, while supporting symbol rates of up to 148GBd. It leverages a 5nm CMOS and indium sulphide (InP) PICs from its own fabrication facility and enables 800G-based transmission up to 3,000km. It can also be deployed over almost any type of network. ICE7 and an enhanced multi-haul optical line system were designed to increase the power and flexibility of the company's GX Series Compact Modular Platform.

Infinera's ICE7 leverages a 5nm CMOS and indium sulphide (InP) PICs from its own fabrication facility (Credit: Infinera)

"Everything we're doing on the GX platform in general and with the engines, it enables them on the high end, to deliver more capacity and a lower total cost of ownership," continued Maher. "The GX chassis itself enables operators to get access to the newer technology faster, so they can take the advantages more easily without having to deploy and certify an entire new system. There is a need at the edge of the network to grow capacity cost effectively. Coherent is such a powerful tool to make that happen, and we need to enable operators to leverage that tool in more applications."

"And as we get better with DSP processing, and electro optics," iterated Munks, "the performance of those systems end up with increased reach and lower power consumption."

Demonstrating this, Fujitsu's 1Finity Ultra Optical transport platform has the additional benefit of upgraded liquid cooling. The platform is designed to deliver extreme performance and scalability with data rates of 1.2Tb/s on a single wavelength, and the company is already planning a future upgrade to 1.6Tb/s. It features a digital signal processor (DSP) using the latest semiconductor processes and liquid cooling technology which the company says provides twice the cooling capacity of conventional technology.

Speaking to Fibre Systems, Paul Havala, Head of Global Optical Planning at Fujitsu Network Communications, explained: "The system benefits from extreme scale and performance. It features a C + L line system and a 1.2Tb/s transponder was made for terabit networking, which makes sure that you have the maximum capacity for customers that are fibre constrained, or where that traffic really needs to be supported at capacity. And then there are also upgrade paths to extend and increase those rates over time. The improvement to its cooling allows the power to be reduced by 40 to 60%, depending on which technology references you're comparing it to, so that's a huge saving. It also reduces fan noise, so you can run the fans at a lower speed, which allows the system to have longer life. So we think there are benefits beyond sustainability alone."

Ciena went directly to 1.6Tb/s with its own launch, the WaveLogic 6 with 200GBd optics, which was designed to support up to 1.6Tb/s single-carrier wavelengths for metro ROADM deployments; 800Gb/s over long links, and energy-efficient 800G pluggables across 1,000km distances.

Helen Xenos, Senior Director of Optical Solutions Marketing, told Fibre Systems: "Both the WaveLogic 6 Extreme (WL6e) and WaveLogic 6 Nano (WL6n) support up to 1.6Tb/s per wavelength. WL6e is the first to leverage 200GBd optics, while the WL6n can be combined with between 120-140GBd optics for coherent 400G-800G pluggables, 400G long-haul and 800G metro/regional applications, and for interoperable 800ZR data centre interconnect (DCI). We have doubled the capacity of the WL6e, but have the same hardware, so there is a 50% reduction in footprint, a 50% reduction in power-per-bit and 15% increase in spectral efficiency."

Acacia's (part of Cisco) coherent solution, the CIM 8 has already begun to ship to tier 1 carrier customers for network field trials. CIM 8 supports data rates up to 1.2Tb/s, powered by Jannu, the company's 8th generation DSP ASIC, based on a 5nm CMOS using silicon photonic technology. Live network trials have taken place over ultra long-haul, long-haul and regional network distances. They were undertaken in partnership with Adtran, and with carriers including China Mobile and Windstream Wholesale.

Acacia's (part of Cisco) CIM 8 supports data rates up to 1.2Tb/s, powered by Jannu, the company's 8th-generation DSP ASIC (Credit: Acacia (part of Cisco))

Nokia has also field trialled the PSE 6s in a live network, in partnership with European operator, GlobalConnect. A speed of 1.2Tb/s was achieved over metro distances (118km) and 800Gb/s over long-haul distances (2,019km), both using a single wavelength.

And it's not just coherent DSPs moving into terabit territory, OFC also saw a number of announcements pointing to PAM4-based solutions also reaching data rates. On the optics side, as OFC was opening its doors, Marvell announced the launch of its first 1.6Tb/s PAM4 electro-optics platform, the Nova, for cloud, artificial intelligence (AI), machine learning (ML) and data centre networks.

Powered by a 200Gb/s optical DSP, Nova enables 1.6Tb/s pluggable modules for scaling AI clusters. Key features include a 200Gb/s per channel line side transmitter interface to support a range of high-speed lasers; a 200Gb/s per-channel line-side receiver with companion Marvell 112GBd transimpedance amplifiers (TIAs) to offer linearity and low noise; integrated laser drivers and latency-optimised forward error correction (FEC) for 200Gb/s traffic.

In other technology announcements, startup Nubis Communications recently launched its first optical engine, the XT1600, designed to be optimised from the ground up for machine learning and artificial intelligence (ML/AI) networks. The company was founded in 2020 and subsequently raised more than $40m to fund the development of its first product to volume production. Samples are already available to customers. The XT1600 has a total capacity of 1.6Tb/s optical transmit and 1.6Tb/s receive carried over 16 fibre pairs, allowing full fan-out and flexible assignment as 16 independent 100Gb/s full-duplex channels, four separate 400Gb/s full-duplex channels, or two 800Gb/s full-duplex channels. It is optimised for unretimed direct-drive operation, connecting directly with host ASIC SerDes designed for copper links. This approach is designed to lower the power, size, and cost of the solution. The underlying technology behind it is a 16 x 112Gb/s (full duplex) silicon photonics engine with integrated modulator drivers and transimpedance amplifiers.

Founder Peter Winzer (also a 2023 Photonics100 honouree) explained: "Our solution is optimised to work with modern SerDes, not only pitch matching to their edge I/O density but meshing with their capabilities for power-efficient direct-drive. The result is that AI accelerators or similar large ASICs can achieve full bandwidth connectivity within the data centre at a fraction of the power compared to traditional optical solutions. Further, our approach lends itself well to novel emerging box architectures as well as to native chiplet implementations for even tighter integration in the future."

With most of these technologies commercially available this year and early 2024, what next for optical networks? According to Omdia's Munks: "Following the 1.2Tb/s, we're expecting 1.6Tb/s on a single carrier, and those will operate up around 200GBd. These are scheduled in 2024 for sampling and probably 2025 for commercial use. We expect that this will be followed by 2.4Tb/s, and following that, probably 3.2Tb/s. We expect all of this to happen within this decade."