For our next article, let's think more about 5G. It really dominated the headlines and rightly so, with so much activity imminent on the standardization front. For example, during one of its seminars, the GSMA stated that, with clear requirements and new use cases already in place 5G really becomes the “network of the networks” (HetNets), focused on user experience, the new era of automation, and intelligence that will enable an industrial revolution. While that may be true, the good news is that, both in the core and in the RAN, a reality check is in progress.
Already, the race for low latency and high speed has led many to conclude that mmW mobility is indeed a milestone too costly and difficult to pass at the time being. Of course, it's possible that new technologies and MIMO could help further down the road, but that remains to be seen. Until then, as we can already see with Verizon and AT&T, fixed broadband (point-to-multi-point) is the first mmW case to be launched. In addition, better coverage as well as (some) mobility are likely to be obtained using mmW under a lower frequency umbrella.
Meanwhile, Cambridge Broadband Networks highlighted results of trials of 200Mbps within a 2700kmsq residential area using 28GHz (60MHz bandwidth) and 5GHz spectrums, and 4x4MIMO — very promising indeed.
However, in sub-6GHz bandwidths, the 5G NR testbed with MU 256x256MIMO and also including beamforming from Qualcomm demonstrated 1.74Gbps UE data rates for MIMO rank 4, and 100MHz bandwidth, and data rates almost twice as high at 200MHz. Latencies showed an average value of 2.5ms across various configurations of MIMO rank and bandwidth. This is achieved based on rapid switching between beams without acknowledgement during scheduling. Capacity is about 12 users. In a nutshell, it seems very likely that 5G mmW will have hot-spot coverage (about 200m mobility) — in other words, it's going to be highly concentrated and beyond it, lower bandwidths will be available.
Of course, 5G is not just about the RAN - we should consider the core too. In particular, network slicing has emerged as a key concept to serve a range of IoT verticals (and which is likely to create a host of new business opportunities).
Network slicing and virtualization
Network slicing and virtualization are fast proving to be of real benefit to mobile edge computing. This was seen in a demonstration by InterDigital demo of its EdgeLink solution to dramatically reduce latency. However, to be truly cost effective and to ensure optimized operational costs, network slicing will require network federation (and extensive roaming agreements).
At the same time, the NFV solutions that will replace today's hardware will also require intensive automation. As a result, the automation and mobile edge computing required for virtualization together with the network federation and interoperability needed for slicing have become key for 5G and its use cases.
In fact, one of the most convincing illustrations of this was an exciting talk at the conference, jointly delivered by Ericsson and King's College, on the “Internet of Skills” project. The project showed how intimately 5G technology is linked to key uses cases. Attention was drawn to one of the most intriguing projects — remote surgery.
Such a case is clearly mission critical and it needs the right technology to enable it, which is why 5G needs to be clearly linked to the applications it will support. In this case, there are obviously exceptional technical requirements. The low latencies for the transmitted data streams that have to be supported cannot be achieved without full SDN/NFV operability and network slicing. Similarly, uncompressed video at the right resolution requires very high bandwidth which must be continuously available, and the amount of data that needs to be processed cannot be delivered without mobile edge computing. As you can see, there are a number of challenges ahead to make remote surgery really possible and practical — and widely available.
Addressing IoT verticals
Of course, the topic of remote surgery gained a lot of attention and many companies used it as a reference point, but there is also a host of other applications that are gaining attention. As you can imagine, connected and self-driving cars proved to be particularly popular themes. Collectively, these and other emerging use cases create the opportunity for new IoT verticals to evolve more or less independently from mobile operators, driven by other stakeholders.
For example, the industry is now flooded with new entrants from outside the classical mobile ecosystem, such as Otto, a Pittsburg US based company developing self-driving trucks, as well as Uber and Lyft. However, in the long run, demanding requirements that these use cases require, such as map synchronization with less than 1ms latency, simply cannot be achieved unless features enabled by 5G (or something like it) are available.
That's why the ecosystem must collaborate and it's also why operators have a vital role to play. For example, v2x (where x = vehicle, pedestrian or network) is proving to be a key topic for operators (e.g. SK Telecom and Telefonica) as well as chip vendors (e.g. Qualcomm). In this case, they are showing the benefits of how better support for v2x scenarios can be enabled with 3GPP standards at 5.9GHz and 20MHz bandwidths when compared to IEEE 802.11.
In fact, the industry as a whole understands that generating value for specific IoT verticals is crucial for all stakeholders to thrive. As Softbank's bank CEO pointed out, ARPU has decreased by 47% while traffic has increased 8x times. But, CAPEX is rising too and there must be compelling use cases that will drive return on the necessary investments.
IoT verticals are clearly essential to this and areas such as healthcare, transport, smart cities and smart meters all require cloud connectivity with either very low latencies, or very high density and broad coverage, and often higher speeds as well. So far, LTE MTC, which has various flavors (M with 1MHz bandwidth, NB-IoT with 180kHz and EC-GSM with 200KHz) is the best candidate to support these, as AT&T, Verizon and Orange recognized by each having decided to adopt LTE-M by the end of the year. Other connectivity solution, such as LoRaWAN are here to stay, at least for a while, but the complexity and the need of interoperability may diminish their spread and uptake.
Big data and the analytics challenge
Last, but not least, let's briefly discuss big data and analytics again. As I mentioned last time, artificial intelligence, machine and cognitive learning were all present in various flavors, demonstrations and discussions. Just walking the show was an ideal way to see engaging offers and the way in which these technologies are capturing the imagination, both for pleasure as well as for more serious applications.
This area is also benefiting from additional study items regarding network intelligence that are planned for 3GPP Release 15 and there were demonstrations spanning various applications of these technologies. For example, Telefonica's Aura cognitive platform, to which we referred earlier, offers learning from data sources with the emphasis on delivering an enhanced QoE for subscribers. Similarly, Teoco showed off real-time troubleshooting and diagnosis, enabled by cognitive learning.
Ultimately, we can expect much more — from predictive analytics to intelligent, self-aware applications that use context (user profile, location, preferences, network/CDN load and usage, and so on) to deliver better performance for consumer services, such as Netflix, all the way to more important things, such as location-aware smart life jackets from Korea Telecom.
Indeed, as KT Corporation's CTO said “5G is all about connectivity, speed, capacity, and now is about intelligence”. In similar vein, Huawei's CTO pointed out that understanding data through intelligent analytics is the biggest challenge during the digital transformation era and that this can be achieved only through partnerships across sectors.
Still, that's some way off. For now, it seems that operators have the most urgent need to increase the role played by artificial intelligence and machine learning. It is expected that this will help them reconcile, on the one hand the demand for cost-efficient services and applications across various verticals and, on the other, the need to deliver better, user-centric network performance, with operational optimization underpinning all they do.
So, let's think about what this means in practice. Sometimes it's easy to get over-excited and think about the hypothetical or real applications on show. But, we cannot escape that fact that these applications will have to deliver the correct performance, at the correct time, across diverse and changing network conditions — at the same time as other applications and services compete for the same resources across the same network. Operators and other stakeholders need help to achieve this, which is why solutions that derive and process network intelligence, available from vendors such as Infovista are going to be more important than ever on the route to a fully digitized world.
Of course, we have to evolve too. At Infovista, we are already incorporating machine learning into our tools and implementing use case centric root cause analysis for all data. We're actively engaged with 5G trials and already helping to optimize the orchestration of network virtualization. We're also providing solutions for emerging IoT verticals (focusing on LoRaWAN and LTE-M and network slicing) and rapidly accelerating the automation they offer while, at the same time, virtualizing and moving them to the cloud, so that they can also support the edge computing needed. Things are gathering pace and we will help you with this transition too.
Today, we're moving to support the transformation, so that the promised services can really be delivered. It's an evolution that ensures all stakeholders can rely on the service assurance they will need in an increasingly diverse digital world - watch this space to see how things take shape in the next 12 months!
