This event has morphed and changed considerably over recent years and is now a stream within TECHXLR8. I picked out a handful of relevant talks where RAN planners shared their view on standards and commercial progress. It seems to me that the use cases, requirements and business case for 5G remain quite vague, with plenty of wishful thinking.
A few years ago, this event was known as the LTE World Series and was then rebranded as 5G World. The scope has expanded to incorporate everything from IoT, VR/AR, Cloud to 5G but the size means that there isn’t huge detail about anything in particular. Think of it like a department store, serving a wide variety of interests but not being particularly specialist in any of them. While it does increase footfall, I didn’t see too many senior executives on the presentation schedule. There are a very large number of parallel tracks, as many as 10 at a time, making it difficult to navigate through the program and causing clashes between the few specific topics you want to attend.
For me, there was a new physical format for the sessions on the show floor. Attendees all wear receive-only headsets avoiding loud booming audio in the exhibition areas – it’s all very calm and quiet. But it wasn’t obvious where or how to grab a headset, and this made it difficult to dip in and out or change between streams. Signposting was poor and the overall program somewhat complex and overwhelming.
The event featured a Connected Car section with a Tesla, a large area devoted to startups looking for further investment (Project Kairos), many IoT devices and Apps, as well as some AR/VR. There was a good Public Safety stream, including a presentation by Mike Roth, CEO of FirstNet, which sadly I was unable to attend.
NGMN indicates much of 5G still to be clarified
The Next Generation Mobile Network is a Forum dedicated to championing 5G in much the same way as the Small Cell Forum has promoted Small Cells. They write documents, such as an Architecture Framework, but don’t try to set standards.
Javan Erfanian, Technical Staff with Bell Canada and NGMN, introduced the scope of 5G and all the big promises it continues to make. However, I left with the impression that there remains a large amount of vague and conflicting requirements still to be worked out. He noted some potential confusion about the meaning of Cloud RAN, which moves much of the functionality away from the cellsites – reducing complexity and equipment at the edge at the expense of more demanding fronthaul bandwidth and latency.
This approach conflicts with moving more of the intelligence to the edge for MEC (Mobile Edge Computing) in order to reduce latency for local applications. So he thought the answer is to be dynamic, allowing functions to be either at the edge or the core. I can’t see this provides clear direction, business case or architecture.
Erol Hepsaydir, RAN and Device Strategy for Three UK
Erol is one of the key strategists who will shape the technology choices to be adopted and deployed. His timescale for large commercial deployment of 5G is five or more years away, and so there is plenty of scope for 4G (specifically LTE Advanced) to address short to medium term needs.
Erol highlighted three critical factors for the success of 5G
- Interoperability with LTE
- Harmonised global solution, especially around spectrum
- Low cost
He’s pretty happy with 3GPP progress on the first two but concerned about the third.
He noted that LTE Urban Small Cell deployment has been held back by high deployment costs rather than equipment prices or technology. Everything from the cost of closing the road during installation, planning permission, site availability and rental are all factors. This will be at least as critical for the cost of 5G urban deployments. However these aspects typically don’t apply for in-building systems.
Andy Sutton, EE/BT
Andy explained the progress of 5G architecture in 3GPP standards. They’ve looked at how best to separate the radio head from the rest of the basestation, such as already found in Remote Radio Heads (RRH) and considered eight different options.
The radio head is called the DU (Distributed Unit) and the remainder is the CU (Control Unit). An example of this split in 4G today would be a macrocell BBU (Baseband Unit) and Remote Radio Head (RRH).
Andy highlighted two specific splits which seem most likely:
Option 2, which retains a lot of the functionality in the radio unit and is not nearly as demanding in fronthaul bandwidth as CPRI. The downside is that it won’t handle the full feature of COMP (Co-Ordinated MultiPoint), but instead could use Massive MIMO and Beamforming as an alternative with at least as good results. The bandwidth and latency to backhaul Option 2 is not much more than for a standalone small cell, so he thought this would be a popular choice.
Option 7 is also known as enhanced CPRI (eCPRI) and likely to be ratified by mid 2017. This does require low latency high bandwidth but unlike CPRI can operate over Ethernet. The link range will be shorter than for today’s CPRI. Andy thought this scheme could be used either where CU and DU are combined on a site, or where daughter DU’s are located fairly close by. The maximum distance might be in the order of 5km.
Security is important and Option 2 will require IPSec (or MACSEC) to secure the data path from cellsite. Frequency and phase sync will be essential throughout.
Andy thought the first batch of 5G small cells (DUs) are likely to be costly and connected by fibre but for long term mass deployment then costs would have to come down.
Franz Seizer, Deutsche Telekom
Franz hoped that 5G could allow the industry a fresh start and dump much of the legacy baggage accumulated over recent decades. He thought 3GPP was still very cellular centric and had not adopted some of the technical or architecture approaches found elsewhere. He seemed to be proposing that 5G system software should be written from scratch using the latest virtualisation architecture and techniques.
He also reaffirmed that there remains vagueness in the scope and terminology around 5G. For example, he’s seen many interpretations of what people mean by Network Slicing – his view is that each slice is completely compartmentalised and indepdent. And it’s not just technology that comes with different interpretations. He believes that many vertical industries get different views from each operator. There needs to be more consensus about what the industry is trying ot achieve and the benefits it will deliver.
While many of the full benefits of virtualisation come from a completely new architecture, I find it hard to believe that the industry can afford to completely redevelop software from scratch to adopt a new architecture, especially when the requirements and use cases remain uncertain. Compatibility with 2G/3G/4G will limit what’s possible. New entrants into the market would benefit by being able to take this approach, but there are few with the budget to embark on such investment.
Trying to pick your way through the 5G hype can be quite hard. I have no doubt there will be headline 5G launches in the coming year or two but I struggle to see how the high costs of deploying very large numbers of DUs (many with fibre) is commercially justifiable. Many more will be required than for 4G using today’s small cells and RRHs because 5G uses a very much higher spectrum with limited range.
So for the next five years at least, I'd expect LTE will dominate industry investment while 5G finds its niche.
3GPP does seem to be making good progress with the 5G new radio standards which would work alongside the existing 4G networks but I can’t see the justification for major re-writing of core network software at this stage.