Cambridge Wireless, a UK non-profit organisation, runs an annual event to ponder the direction of the wireless industry. Freed from some of the commercial pressures seen elsewhere, it gives a different flavour and insight into what might shape our future.
The wider context
David Cleevely, CW’s Chairman, reminded us of the enormous disruption that IT and communication technology causes. Vertically integrated businesses have been dissected and dis-intermediated. “If you think the CEO of Uber is going to stop with just taxi services, you’ve got another thing coming.”
David Wood, Delta Wisdom, followed up on this theme, illustrating how a burgeoning industry can be steamrolled by a larger tech trend. Examples included the video rental business, where Blockbuster didn’t buy Netflix for $50m at an early stage (now worth >$20 billion), newspaper ads have been superceded by Internet ones, printed booked by Kindle eBooks. Sometimes company staff knew exactly what had to be done to deal with the threat but were unable to implement solutions. Technology steamroller effects can also be positive though, such as where industries have used IT capabilities to grow quickly.
He thought that we should expect to see a layer of Artificial Intelligence added everywhere – from driving cars autonomously to learning how you like your coffee – more insight on that here. Smart Glasses/Augmented Reality were seen as a key driver of data growth. Security was considered to be a growing issue which most still don’t fully comprehend.
Wireline Evolution to 500Mbps copper broadband
Tim Whitely of BT explained their next phase of wireline broadband would use G.Fast, which can achieve up to 500Mbps – anything faster would justify fibre throughout. Many techniques have been adopted from the wireless industry, with “extra spectrum” being used to achieve these high data rates. G.Fast uses up to 170MHz bandwidth compared to only a few MHz in early versions of DSL.
For core network links, BT has trialled and proven speeds of 3Tbps over 359km with Huawei kit using optical fibre - that's 3,000 Gbps.
Jeffrey Ju, SVP for Wireless at Mediatek, who supply processor chipsets for a large proportion of the world’s smartphones (alongside RF chips from Qualcomm) explained how these products have killed off the need for many separate standalone gadgets – GPS navigators, pocket gaming and even cameras. So what’s next? They are focussing on three aspects:
- Power efficiency, because batteries aren’t improving at the same rate as other technical components
- Visual processing (we won't be staring at our smartphone screens in future, ultimately we will have some form of heads up display)
- IoT (Internet of Things)
I’d heard of multi-core chipsets before, but not appreciated that each processing core doesn’t have to be identical. Their latest 10 core chip has three different core designs – 2x fast, 4x medium, 4x slow – each size better suited to different types of Apps. Think of it like gears in a car, which balance power and performance. This saves around 30% of total power consumption.
There was some comment during the Q&A that the industry (and academia) aren’t training enough computer programmers familiar with multi-threading/parallel processing skills to make best use of these. I questioned whether App programmers should be more aware of multi-connectivity (i.e. handling multiple different communications channels with their own characteristics) which I feel is not being widely addressed.
It was also said that the proportion of battery power used in a typical smartphone just for the radio (cellular and/or Wi-Fi) is now relatively low (say <20%), with backlit displays and graphic CPUs taking the most dominant roles.
What isn’t 5G?
Industry commentators point out that up to now a new generation has come along every 10 years or so, where networks invest in huge equipment upgrades and we all buy new phones to use it. Some expect this cycle to be repeated, and dream up a huge wishlist and capability set for 5G to satisfy. This led to Moray Rumney of Keysight to pose the question “What isn’t 5G”?
His view is that we need to obey the laws of physics and lead to a mature and sustainable market proposition. Consumers must understand the benefits of product performance and assess the additional value to them. This hasn’t always been the case with 4G, which in many cases hasn’t justified higher revenues directly because of performance (although higher monthly data consumption might lead to that). He points out that perhaps the major benefit of LTE has been to open up access to a wide range of new spectrum bands, which add far more range and capacity than from the new RF technology alone.
While we push the highest speeds of wireless technology to greater limits, he notes that many rural areas remain without good service. He proposed a “No Home Left Behind” campaign to improve geographic coverage, commenting that he couldn’t even get good 3G service when visiting Bonn in Germany recently. This isn’t a technical constraint, and depends more on commercial approach and regulation.
Superfast wireless using millimetre wave technologies (say at 60GHz) were discussed - these look quite promising especially within the same room, but have some significant issues. Moray lists these as the "Five NEINs" to be overcome:
- Death Grip - no amount of equaliser can compensate hands blocking the antenna
- Body shadowing - signals at these frequencies can’t travel through your body
- Moving vehicles - disrupting line of sight and/or reflected multi-path signals
- Outdoor to indoor coverage (it won’t penetrate into buildings)
- Cost - It would be very expensive to build enough of a high density wide area mmWave network with the ability to maintain multiple links per user
My expectation is that mmWave will first appear as a superfast Wi-Fi feature on smartphones (previously called Wi-Gig) early 2017, working well when inside the same room as the access point and with clear line of sight. Blu Wireless, exhibiting at the event, told me that the technology is currently going through the proving and certification phase - everyone wants to ensure they get this right and avoid a false start before mass market rollout. That may explain why things are a bit quiet on the marketing side at this stage, but could ramp up in about a year from now.
5G not required by 2020
An old fashioned debate argued the specific question “5G is the next generation of mobile carrier incubated radio access network technologies that will be ready for early service adopters by 2020”.
The vote at the outset indicated that most delegates were in favour. At the conclusion, the mood had changed with the majority voting against. Tony Milburn, u-Blox, speaking against the motion, argued that the standards-making machine has moved to a point where it is defining things beyond the need of the consumer. He felt the area for investment is coverage, not yet another standard that sucks the capital out of operators and delivers something consumers don’t need.
Cavium champions RAN Virtualisation
So what would make network operators rip and replace their massive installed base of macrocells? Raj Singh, GM Wireless Broadband Group at Cavium, thinks we need to simplify the RAN before moving on to the next stage. They support various architectural splits that centralise functions normally found at each cellsite.
Pointing to OPEX savings reported by China Mobile, he thought the reduction in macrocell power consumption and reduced need to visit each site provides a strong justification to invest in Cloud RAN. When I asked if this extended even to remote/rural sites rather than just a few dense urban areas, he thought that RAN Virtualisation applied everywhere even including sparsely populated areas.
A useful set of presentations combined with networking opportunities made this worth attending. This was by no means a dedicated “Small Cell” event, or even a 5G one, and provides another and wider perspective on where we’re headed.
Key takeaways for me:
- The wider IT evolution has steamroller effects that will continue to disrupt and change business practices in surprising and unexpected ways. Those who can take advantage of it will win.
- The smartphone is far from just a communications device. As the IoT grows, dependence on cellular network operators may be disrupted, but quality and resilience of connectivity will grow in importance.
- The scope and timeframe for 5G remains unclear and uncertain. It’s unlikely to be just another infrastructure technology refresh cycle as was the case before.
RAN evolution has many potential and conflicting technical directions open to it. I don’t think the industry has a clear idea of which to choose, largely because we lack clear and common set of agreed requirements. Perhaps 5G will end up being much more about evolving and even reshaping the wireless industry to serve the underlying needs of consumers, differentiating through more consistent and more resilient service than by headline peak speeds alone.
A full set of presentations from the event are available online (no registration required) at http://www.cambridgewireless.co.uk/futureofwireless/resources/
As a Cambridge Wireless Small Cell SIG champion, I’ve been organising our next Small Cell event (Cambridge UK, 24 Sept) which asks if Ran Virtualisation means the end for Small Cells. We’ll be asking what forms it might take, learn from some field experience to date and how that affects backhaul.