The greening of our fiber networks

Extinction awaits those who cannot rapidly get on top of their energy consumption ...
Ronan Kelly
People using digital devices

Extinction awaits those who cannot rapidly get on top of their energy consumption

During the COVID-19 pandemic, societies across the globe bore witness to tremendous reductions in pollution and greenhouse gas emissions when the daily trudge in and out of our cities and industrial parks was replaced by working from home. For the first time, mountain ranges emerged through the veils of smog that had cloaked them from surrounding cities for decades.

This precious glimpse of what our world could be like if we could reduce and control our emissions was afforded by modern broadband networks. In fact, these networks served as the fabric that held societies together over the last 24 months. As a result, governments around the globe have recognised the importance of broadband in their broader fight to reverse the damage of the last 100 years and provide a path toward a viable future for all.

While broadband networks themselves will be a foundational enabler for many government environmental strategies, the broadband industry has a major role to play in reducing its environmental impact. This comes against a backdrop where bandwidth demands continue to increase from market competition, as do the consumption patterns of consumers as their communications and entertainment mediums steadily transition to over-the-top (OTT) delivery with ever-increasing fidelity. The half a Terabyte per month average download threshold has now been breached, and over 20% of subscribers exceed a Terabyte of downloads every month.

With increasing flexibility, reach, control, and choice, along with lower content distribution costs, we must not anticipate any deviation toward complete OTT delivery of every communications medium in the future. Recent announcements by European satellite and cable content providers like Sky and Virgin Media validate the inevitable transition to full OTT. The only question is, “How quickly will the transition happen?” This will place more load on broadband operators’ networks. It will also result in network operators taking responsibility for the cost of the power needed to deliver said mediums to end-users.

In the simplest terms, the energy required for the traditional transmission of a TV or radio signal over the airwaves – where anyone in range using an appropriately tuned receiver can access the content – is much less than the energy required to deliver the same content via unicast over an operator’s broadband network to address a similar number of recipients. The continued shift from broadcast via airwaves or cable to unicast OTT delivery means that operators will be picking up the bill for the energy consumed during content transmission. Some edge-focused approaches, like multicast ABR via a transcaster attempt to reduce the impact of content-related rising traffic levels. Unfortunately, the consensus thus far is that the gains provided do not yet justify the complexity and increase in equipment costs.

These future OTT-born increases in energy consumption, coupled with the skyrocketing cost of energy, only compound the imperative for broadband operators to prioritise their environmental strategies and initiate a relentless focus on reducing the energy consumption of their networks.

As multi-gigabit services and global supply chain shortages conspire to kill off Active Ethernet, which technologies should operators plan for?

In previous CTO Insights, we spoke about the migration of broadband operators from the power-hungry Active Ethernet Point-to-Point (P2P) fibre access to the more efficient, multi-gigabit capable XGS-PON Optical Line Terminals (OLTs), often deployed over the original P2P fibre infrastructure. This migration has continued to gain momentum as the increasing emergence of multi-gigabit service offerings forces an upgrade beyond current gigabit-capable Active Ethernet switches. This is compounded further by the global supply chain challenges, which are seeing the diversion from common components over to the larger and expanding PON side of the industry.

While this migration will assist some early adopter fibre operators by reducing their energy consumption, they and all the other PON-focused operators must think carefully about their future plans. What is their optimal timing for the introduction of technologies beyond XGS-PON? Standardisation of G.hsp is drawing near within the ITU-T, the industry’s standards body responsible for GPON and XGS-PON standards. G.hsp sees 50 Gbps PON as the next standards-based step forward, while outside of the ITU-T, a movement supported by 12 operators is underway within the industry to encourage operators to begin planning for the imminent introduction of what they describe as 25GS-PON. This is being driven by requirements from large enterprises for symmetric 10GE services, which XGS-PON with Forward Error Correction (FEC) enabled is unable to provide a growing need for small cell backhaul.

Within this blog, we will avoid being drawn into the debate around how many of these 10GE demanding large enterprise customers are proximate to the residential PON networks actively being built. We will also overlook the fact that PON Optical Distribution Networks (ODNs) will outnumber urban and rural Small Cell deployments 20 to 1 by 2026, according to GlobalData. Instead, we applaud every effort to service the needs of enterprise and cellular backhaul over the same PON fibre networks that service consumer access. We believe this approach can deliver substantial energy savings. However, it is critically important that the methods used to realise these bandwidth gains do not come at the expense of operators’ overall energy reduction plans. Many factors must be considered when planning beyond XGS-PON to ensure the desired outcome is achieved for both the target service offering and the company’s environmental targets.

Consolidated enterprise and x-haul services via next-generation PON will be impossible without chipset and OLT architecture evolution

Implications around current chipset capabilities require consideration when enabling G.hsp and 25GS-PON. With the current chipsets optimised around combined GPON and XGS-PON deployment, the introduction of either G.hsp or 25GS-PON will result in the number of ports each chip can service being reduced. Demonstrations thus far have seen the port counts at best halved or quartered. Most operators do not have the capability to rationalise their existing customer base across half the number of access ports on an OLT platform. To maintain the same number of customers, they must double the OLT equipment to uniformly introduce this capability. This drastically increases both CAPEX and energy consumption at a time when operators are desperately trying to reduce both of these expenditures.

Another key consideration when looking at today’s fibre access OLT chassis is that many are operating at the pinnacle of their efficiency. With Combo PON (XGS-PON and GPON simultaneously from the one port transceiver), most modern chassis systems match the line card bandwidth capability with that of the chassis backplane into which it connects (16 x 12.5 Gbps into a 200 Gbps backplane slot), delivering the non-blocking port to backplane capacity that the industry has been accustomed to since the inception of broadband. This is a pivotal consideration when planning for premium enterprise access and cellular x-haul. These services must be delivered with service level agreements around capacity, latency, availability, security, and mean time to repair. When seeking to run these types of services across an over-subscribed backplane, one must pay extra attention to how Quality of Service (QoS) on the backplane is managed. One must ensure that dropping critical enterprise traffic or delaying time-sensitive mobile x-haul traffic are not options. When 25GS-PON is introduced with XGS-PON and GPON onto each OLT port, the resulting capacity is 37.5 Gbps per port or 600 Gbps per typical 16-port line card. With most popular backplanes maxing out at 200 Gbps per slot, the backplane interfaces are now oversubscribed three to one. Suppose you wish to stay within the standards and focus on the ITU-T standard G.hsp utilising the 50 Gbps capability of G.hsp alongside XGS-PON in a next-gen combo fashion. In this case, the contention level of each line card climbs to five to one. This oversubscription will, for many operators, prove unacceptable for launching mission-critical services, which would be a travesty given the important role that convergence has to play if the 10s of millions of residential PON fibre ODNs are to be used for enterprise and cellular x-haul services. This is where, quite literally, out-of-the-box thinking is required.

Web-scale innovations are key to unblocking the path to introducing next-generation PON technologies

One option for operators to safely introduce future PON technologies while meeting their immediate cost and energy reduction targets is to use a configuration that can offer these types of services in an un-contended fashion is by utilising cloud-based, software-defined disaggregation. Leveraging the constantly advancing approaches in data centre Ethernet switching rather than depending on advancements in slow-evolving traditional backplane capabilities, disaggregated OLTs are already leveraging the newer, more efficient QSFP-DD400 interfaces to preserve protected un-contended capacities for Combo PON-based services. With the emergence of QSFP-DD800 Ethernet devices, the same will be possible for Combo PON of G.hsp (50G PON) and XGS-PON.

Adtran table

An indirect benefit of disaggregated OLT solutions is their need to compete on their own without the protective veil of a chassis around them. This means the competitive pressure on disaggregated OLT vendors is always present. Hence, the cadence with which newer disaggregated OLTs are being released to market is much faster than what we have historically seen with chassis-based OLT line cards.

As next-generation 7nm and 3nm chipsets make their way into the OLT ecosystem, disaggregated OLT solutions will lead the charge in their introduction. This will open operators to further power savings with GPON/XGS-PON and Combo PON while also facilitating the introduction of next-generation PON technologies without compromising port densities and power requirements. Only when operators can replace their existing Combo PON port densities with a solution that has comparable power consumption to that of today’s solutions can the industry credibly consider their introduction into the network.

Achieving half a Terabit per second over PON networks today without altering the cost structure of the PON CPE

For those operators who have a pressing requirement to introduce dedicated 10GE or even 25GE services onto their PON ODNs, the option to augment their PON ODN with Point-to-Point (P2P) wavelength overlay is available. There is enough space on the wavelength grid to overlay up to 18 x 25GE, or a mix of 10GE and 25GE wavelengths alongside GPON and XGS-PON, over PON ODNs spanning up to 20 km in distance. In this case, one can extremely cost-effectively address any of the use cases that surpass the capabilities of XGS-PON without burdening OLT solutions with the complex clocking advancements needed for small cell x-haul applications and minimising the increased power needed.

There is no question that the PON ODNs must play a converged role for all service types in the future. Next-generation PON technologies will support many of these advanced use cases. However, there are important technological and environmental considerations that must be considered prior to their deployment. Until the OLT chipsets have advanced sufficiently to enable these next-generation technologies to be deployed with the same port densities and without compromising the operators’ environmental strategies, P2P wavelength overlay remains a low-cost, low environmental impact option that can offer all the stringent clocking capabilities of an actual dedicated P2P fibre service.

As next-generation 7nm OLT chipsets and later 3nm chips make their way to market in the mid and late decade, we will be able to facilitate the introduction of next-generation standards like G.hsp without compromising...

As with all things in life, timing is of the essence, and enhancing our PON networks is no different. Introduce a new technology and you pay for the privilege, not just in equipment costs, but also in the inefficiencies of the components being stretched to grant that early access. Time things correctly and you get to tap into the sweet spot of cost curves and component efficiencies to deliver profitable services.

Related articles