Broadband World Forum (BBWF) used to be an event which scarcely appeared on Wireless Watch’s radar, but that is no longer true. Wireless is becoming more central to the broadband world, whether because of the revival of interest in fixed wireless access (FWA) or because telcos and cablecos need to add a mobile service to their multiplay bundles.
And wireline broadband technologies are increasingly vital to mobile operators. Backhaul is no longer a matter of choosing between a leased T1 line or purchasing a microwave licence. As cells get smaller, and networks become virtualized, many different fibre and copper options are being considered for backhaul and fronthaul (see separate item on G.fast).
Deutsche Telekom took the wireless/wireline convergence theme a step further at BBWF, which was held in its homeland this year, in Berlin. Franz Seiser, VP of core networks and services at the telco, said the fixed and mobile industries need to work in parallel to ensure full convergence, to meet “new internet” requirements and make sure 5G can fulfil its potential.
Without this convergence of infrastructure – not just of services and billing – 5G’s prospects will be constrained, he argued. “Video going to HD and ultra HD, AR and VR applications – all need more and more bandwidth, but that doesn’t create ARPU in the same way. How do we innovate and stay profitable? Today’s way of running networks will not deliver,” he said.
This, of course, echoes a near-universal theme among MNOs as they think about 5G. The applications and user experiences which 5G could support are tempting, but there will have to be certainty that they will also boost revenues.
Operators remember how few of them succeeded in charging a premium for 3G or for 4G when those networks were deployed at great cost. They did drive down the cost of data delivery and network ownership, as 5G should too; and they did encourage higher usage levels which could support upselling to premium services and content. But the returns are diminishing in the consumer market, so MNOs have to look to less saturated user bases, as well as a radical new way to deploy and run networks. And Seiser believes that, in the consumer space, there will be virtually no ARPU advantage in 5G, only cost efficiencies.
“5G is a unique chance, but we need to … make sure it’s going in the right direction,” Seiser went on. “To enable innovation you need to throw away a lot of legacy and as an industry we are very bad at that … it hampers us going forward.”
While there may be some immediate benefits from early 5G – first mover advantage, the cost efficiencies and so on – this must be a network which continues to deliver incremental benefits, primarily new revenues, throughout its life, and that will require convergence, in DT’s view. Seiser said: “The big question is what happens afterwards. Are we able to start working on the future challenges? Today we are running two totally separate networks, fixed and mobile, to supply the same service. Wouldn’t it be better to bring everything together?”
Seiser did a fair bit of chest-beating about operators’ failure to change sufficiently quickly, but he also had harsh words for the vendors and their slow pace of innovation. “It doesn’t work if you continue vendor-dominated innovation. They control innovation and they do it in incredibly small steps,” he told Mobile Europe.
Again, this is a recurring theme. MNOs, from the traditional giants like AT&T to the new disruptors like Reliance Jio, are trying to behave more like webscale companies and introduce some of that flexibility and responsiveness, as well as sheer scale and automation, into their cultures. This means putting pressure on their traditional vendors to behave differently – and that pressure often comes from the threat to work with new suppliers, including the increasingly influential open source initiatives such as CORD (Central Office Re-architected as a Datacenter), ONAP (Open Network Automation Protocol) and Facebook’s TIP (Telecom Infra Project).
Seiser believes the shift to virtualisation, and to architectures whose roots lie in the data center rather than the network equipment providers, has helped operators to force the pace of change. “Vendors would never have kicked that off,” he said, referring to virtualization. “We need more of that and we need to step out of this old model.”
Of course, DT has been a major supporter of TIP, and is one of the first three operators to set up a TEAC (TIP ecosystem acceleration centers) to incubate start-ups whose approach to the network could revolutionize the cost base and the supply chain. The other early movers are BT and Orange, and Spain’s Telefonica is also backing the initiative, though it does not yet have a TEAC established.
Telefonica’s TIP representative, Farid Singh – who also co-chairs the new TIP edge computing group – was also urging operators to think more like web giants, making their networks smarter rather than just fatter, and moving towards “service-aware connectivity”, not dumb pipes. Clouds like Microsoft’s or Amazon AWS’s are aware of their customers connectivity requirements and call up the resources they need, when they are needed, he said – a concept which will be a foundation for 5G network slicing and the fully dynamic, on-demand network. “The service will be what you sell and the connectivity will be irrelevant,” Singh said.
He pointed to recent alliances between cloud operators and MNOs or telcos, such as AWS’s with AT&T. “We can’t compete with the expertise of AWS, Microsoft and we shouldn’t,” Singh told Mobile Europe. “We should collaborate with them and say, can you squeeze your cloud to my edge?”
He wants to get away from the stand-off between operators, which resent investing in networks and then seeing the web providers taking most of the revenue. It is essential to the survival of the MNOs that they find a model of mutual benefit, otherwise the web players could start to harness shared spectrum and partnerships with alternative operators, to reduce their reliance on the cellular networks. Singh proposes a profit sharing model in which each stakeholder invests a certain share of the cost of the network. The incentive for the cloud providers would be to enhance their capabilities with enriched connectivity, while MNOs would be able to offer a wider range of cloud-based services and move towards high value enterprise customers and network slicing.
A hint of this approach was seen recently when China Unicom announced a restructuring of its ownership – several web providers and industrial companies took stakes, which will improve the telco’s ability to roll out 5G, but will also give its new partners a greater say in how that network should be deployed and optimized to suit their needs.
The operators’ comments show how hard it has been to convince MNOs and vendors of the benefits of a multi-technology, open approach – but there was an air of change at BBWF. Analysts from Wireless Watch’s sister service, Faultline Online Reporter, highlighted US network vendor Adtran as an example of this new atmosphere, when they attended the supplier’s BBWF event.
In this, Adtran discussed a mixed bag of network technologies, but what stood out was that every single presentation throughout the day mentioned a multivendor approach.
Voice-activated devices and services are some of the primary drivers for faster broadband connectivity, according to Adtran’s CTO Ronan Kelly, who kicked off the event at Deutsche Telekom’s Berlin base. While voice services are not bandwidth-hungry like 4K or VR applications, they are extremely sensitive to latency – making connectivity upgrades critical to the development of voice services as a market.
In other words, data rates and bandwidth are not everything. To rescue the MNO business model requires flexibility and a service-driven architecture which adapts to changing customer needs. For instance, not all consumers want to pay for gigabit speed broadband connections when they may only download a handful of large files a month or stream a couple of 4K videos a week. Kelly said operators were missing the opportunity to charge a fee for a temporary upgrade – for instance, when downloading a large file such as a new console game, a user could be alerted to the option to pay a one-off fee, perhaps $5 or so, to upgrade to a 1Gbps connection temporarily, enabling the download of a 60GB file eight minutes or so, instead of the two hours it would take on a 100Mbps connection.
This brings us on nicely to where software-defined access network (SD-Access) technologies are coming in to make this mutually beneficial service a reality.
SD-Access enables the centralization of elements in a network such as high level analytics – enabling operators to easily launch over-the-top services with less expense and complexity. SD-Access makes this transition easier as operators no longer need vendor-specific platforms and can bring in a self-service activation model to differentiate from the pure play providers, in addition to making significant truck roll and call center savings.
To accelerate the path to SD-Access, Adtran is working alongside open source projects for virtualized networks, including ONAP for open source management and orchestration (MANO), ON.Labs’ ONOS SDN controller, VOLTHA (Virtual Optical Line Terminal Hardware Abstraction) and Open Daylight. While all slightly different, the general idea here is to save operators a bunch of integration work. ONAP has a release scheduled for mid-2018 focusing on wireline-only carriers and business services, whereas previous announcements have centered more around mobile operators and 5G.
All the packets being sent via wireless networks must come back to wireline at some point, which is where PON (Passive Optical Network) technologies, namely NG-PON2 (Next Generation PON), enter the fray – to upgrade current technology by being deployed as an overlay on existing fiber. AT&T passed 6m homes with PON technology as of this quarter, while Verizon passed 19.8m two years ago, according to Adtran, and is currently in NG-PON2 trials with Verizon.
In the case of multi-dwelling units (MDUs), fiber-to-the-home deployment is expensive and laborious, so existing copper lines will often be used, harnessing recent innovations in Dynamic Time Assignment (DTA) for G.fast to allow broadband subscribers to access the highest possible upstream and downstream capabilities. Adtran has a coordinated DTA (cDTA) prototype in the works for twisted pair bundles, plus an existing independent (iDTA) system for coax.
Circling back to the open approach discussion, Nokia’s president of fixed networks, Federico Guillén, said in a keynote session at BBWF: “The strategy of deploying fiber to the most economical point in the network is still valid, but the combination of fixed fiber, wireless and other access technologies is now even more crucial to the operator’s business case.”
ONAP claims to be in active discussions with almost all of the top 50 global operators – Deutsche Telekom being a key member. DT’s senior program manager, Robert Soukup, gave his view on next generation networks and open environments. “We don’t like proprietary stuff on our network because it makes it difficult to apply design to cost capabilities. We need to trust vendors and we welcome more kids on the block.”
DT has been a member of the one-year-old ONF (Open Networking Foundation) for around three months now. Soukup believes ONAP is perhaps five years away, contradicting Adtran’s own prediction that ONAP adoption is about two years out. Adtran’s AVP of cloud, Robert Conger, explained that the contrasting timeframe expectations are down to ONAP not being just one big piece of code – with the complexity hidden in multiple layers meaning some components will be ready to go in one year, but others will take much longer.
But the differences of opinion also simply highlight the immaturity of these platforms, and of operators’ plans to deploy them in anger.