My presentation at Ofcom: What the year 2030 implies for wireless trends & spectrum policy

On the 8th of February, I gave a presentation at Ofcom (the UK telecom regulator). The event was a day-long discussion of the "Future Wireless World", looking at longer-term trends towards IoT and connectivity (5G, WiFi, mesh, satellite and more), with an implied impact on how spectrum policy needs to be reshaped to meet the changes. It was introduced and moderated by Philip Marnick (Group Director, Spectrum) and also attended by the Ofcom CEO Sharon White. On the same day, Ofcom released its latest thoughts on 5G spectrum (link)

There were about 150 attendees from a range of operators, broadcasters, government bodies, vendors, consultants, Internet and industrial players and internal Ofcom staff. There may be an audio/video recording of the sessions put up online at some point, but I'm not certain of this.

My presentation was a very broad one - I was tasked with imagining how the future economy, consumer and business environment might look like in the year 2030, what disruptions and innovations may occur between now and then, and how that flows back into the use of wireless networks and therefore spectrum. 

In other words, I was wearing my "telcofuturist" hat, where I take generic futurist themes and apply them to the specifics of telecoms and the broader wireless industry. After my presentation, I joined Philip Marnick for a Q&A session with the audience, which was a mix of regulatory, futurist and general analyst-type discussion.

The rest of the event was made up of a series of presentations and panel debates between a broad set of industry luminaries and innovators, including Dino Flore of 3GPP & Qualcomm, Simon Saunders of Google (& formerly the Femto Forum), plus others from O3B, Ericsson, Veniam, BT/EE, Vodafone, Silver Spring and others. There was a really interesting session on mesh networks later in the day, which I also think has a lot of potential. It was a really refreshing change from some of the usual sponsor-driven snorefests, although there was clearly a strong "lobbying" flavour to some of the questions, with people taking advantage of access to the regulator in an open forum.

One thing that struck me about both this event, and another event I attended recently at Tech-UK's Spectrum Policy Forum (link) is a growing frustration in the regularory community. Some people now view spectrum purely as a "mobile" thing, without simultaneously mentioning broadcast, government, WiFi, LPWAN, industrial, satellite, fixed-access and all the other users of the airwaves.

The mobile industry tends to be very good at pitching for more and more slices of spectrum, ideally provided on an exclusive basis with long licence terms (in exchange for quite a lot of cash in terms of fees, to be fair). It has a far bigger and more cohesive lobbying and publicity engine than the broad set of other spectrum stakeholders.

My own view - and, it seems, many regulators' - is that given the finite amount of spectrum, there is ever less rationale for exclusivity. Various forms of sharing and private networks are rising up the agenda. My recent piece on Industrial IoT and sharing [link] has garnered a lot of good feedback, while the National Infrastructure Commission's Dec'16 report [link] recommended that "Government and Ofcom should review how unlicensed, lightly licensed spectrum, spectrum sharing and similar approaches can be utilised for higher frequencies to maximise access to the radio spectrum".

In other words, spectrum-sharing - of various types - is moving up the regulatory agenda very fast in the UK. I think onsite industrial IoT coverage, via private cellular or licenced-band WiFi deployments, is the easiest to conceptualise and "sell", but there are plenty of other angles too.

But as well as the challenges of IIoT, I covered a lot of other topics in my presentation (slides are embedded below the list - apologies that the bullets aren't in the same sequence):

• The impact of AI will be felt on both network "supply" side (eg more efficiently-optimised networks, churn management etc) and "demand" (smarter use of wireless connectivity, least-cost routing and so forth). I wrote a post on this a while back (link)
• Whether the emphasis on mobile uses of spectrum, and the 3GPP/GSMA "national MNO" view of the world could lead to a "monoculture" of cellular connectivity. As in agriculture, the superficial efficiency/yield needs to be considered in the context of risks. Might there be long-term benefits in "network diversity", and should regulators look to protect it, the same way environmental rules protect biodiversity?
• On a similar environmental theme, I considered habitats that are primarily "mono-platform" and fragile to external events (eg coral reefs) vs. "multi-platform" ecosystems which are more resilient (eg rainforests). Obviously this doesn't translate precisely to wireless networks, but the metaphor seems apt. I'm not a biologist, but a quick word with someone who does study ecosystems afterwards suggested my analogy is worth further exploration.
• "Arbitrage Everywhere": future networks - and by extension both spectrum and telecom competition rules - should anticipate devices and applications using multiple connections / service providers, and picking and choosing/bonding connectivity from several options. This is already seen in the fixed world for enterprise with SD-WAN, and should be expected in wireless too. This means that "partial competition" (eg from WiFi, LPWAN, satellite, private cellular) should be considered as well as like-for-like rival infrastructure from other national MNOs.
• Redefining the nature of a "service" - what do we actually mean, when we frame our regulation of "service providers"? Many more organisations are offering connectivity services, while many other models of delivering a "capability" are emerging. WiFi can be a service, owned by a venue, given away for free, provided as an amenity, self-provisioned by a user and so forth. ITU's definition of a service ("a set of functions offered to a user by an organisation") seems to be too narrow given the rise of developers, embedded connectivity in IoT, private networks and more.
• I discussed the relative timing of various industry trends - and the fact that various look like swinging "pendulums". For instance we see a back-and-forth between centralised vs. distributed control, standards vs. proprietary technologies, local vs. national vs. global and so on. I noted that the timing of the various pendulums' swings are not all in sync - and therefore the actual outcome for the wireless sector is really complex to predict. Various external trends (eg open source, Moore's Law, AI, geopolitics, specific companies) can act as weights on the pendulums.
• I noted that many different and new organisations may own/operate/embed wireless connectivity in future. Aircraft engine manufacturers use satellite telemetry and download sensor data via WiFi to optimise their analytics for selling "power by the hour". IoT platforms & MVNOs for specific sectors are springing up (eg Cubic Telecom for automotive). Theoretically, Elon Musk could use SpaceX to launch his own satellites - and provide vertically-integrated connectivity to Tesla cars. Google has numerous wireless initiatives, from Fi to WiFi to white spaces to its Loon balloon project. The Governor of California has suggested launching the state's own earth-sensing satellites, if the current administration cuts federal funding for environmental monitoring. Then there are public-safety LTE networks, WiFi everywhere, new mesh concepts, private LoRa deployments and so on.
• In the Q&A, I also discussed 5G bands, NFV, network-slicing and more. I noted that 5G is being driven initially by fixed-access and 28GHz in the US & S Korea, not the three "mainstream" uses of critical IoT, ultra-mobile broadband and massive IoT. This is outside the "official" bands being pushed by Ofcom as "pioneer" options, and slowly being explored internationally for the ITU WRC event in 3 years' time. This was explored in another post of mine (link). I also expressed doubts that NFV-led network-slicing will deliver all the agility required for creating vertical-specific networks - even if it allows "super-MVNOs", will the host network provide enough fine-grained control and liability-bearing SLAs?

Ofcom Future Wireless World + Spectrum - Dean Bubley from Dean Bubley

Overall, my session seemed to be very well-received. Hopefully I've prodded some parts of the industry. I'd like to see a wider recognition of the changes to some of our fundamental assumptions that will occur over the next decade and beyond. 

A key point is that 5G, delivered by traditional MNOs as a subscription service, is exciting and important - but it must not be allowed to totally dominate discussions around spectrum. Governments and regulators must push for "network diversity" of technologies, stakeholders and business/operational models - including private networks for businesses. Short-term focus on "efficiency" of a monoculture approach may mask wider ecosystem-level risks. 

A key theme is the need for flexibility and agility in wireless networks and related regulation - many of the more radical changes will occur at timespans of 1-5 years, which is much shorter than the investment and planning horizon for a lot of the industry. Whether we need more malleable licences, better secondary marketplaces for spectrum, new forms of sharing (eg using blockchain as a basis for a distributed database of allocations), or a rethink on how competition is measured, there are plenty of options.

Spectrum policy is several steps away from the actual world of consumer and business needs for wireless networks. But it's for that reason it's worth thinking deeply, about the long chain of implications of seemingly small decisions or baked-in business models that are created now.

If you'd like to have a similar presentation and discussion at your own event, or at a private workshop, please contact me via information AT disruptive-analysis dot com

No, VoLTE is not the future of mobile voice

I've lost count of the articles and presentations I've seen recently, loudly proclaiming that VoLTE represents "The Future of Mobile Voice", or worse, "Allows telcos to compete against [so-called] OTTs". The run-up to MWC is likely to generate even more breathless and gushing tributes of this type.

But this is all nonsense. VoLTE is most definitely NOT the future of mobile voice. It's just the past of mobile telephony, with a lick of very expensive IP paint.

(It goes without saying that VoWiFi, RCS & ViLTE aren't the future of mobile comms either. See here for a full explanation, and watch out for another post soon).

VoLTE is still just "boring old phone calls", billed per minute, with interruptive caller-to-recipient ringing, based on E.164 numbers, with no customisation for purpose or context or sense-making. HD audio and embedding calls in apps via APIs aren't new either - they work for circuit calls too. Not only that, but VoLTE doesn't work very well - Amdocs recently providing evidence that calls drop far more frequently now. 

It's also hugely expensive and complex to implement, and in most cases different VoLTE networks don't interoperate, as they're all slightly different. The GSMA has even given a special acronym to its bug-fixing squad - now called VIRTUE (VoLTE Interoperability Resolution Expediting Task Force) - link here.

There are only two real purposes for VoLTE: supporting calls on 4G-only networks, and hopefully reclaiming some 2G/3G spectrum. Worthy, but not "the future". VoLTE will not increase telephony usage. Yes, faster call setup, better indoor coverage (4G or WiFi) and clearer HD might slow the decline a little, but neither are exactly groundbreaking. Fixed PSTN call setup has been instantaneous since the 1980s, and HD-quality voice has been available on both 3G networks and 3rd-party VoIP apps for years. VoLTE is arriving very late to the party, with a bottle of cheap supermarket lager, just as the cool people are heading somewhere else for champagne and whisky.


Peak Mobile Telephony

This is because telephony is yesterday's news, even mobile telephony. Looking around the world, many countries are past the point of "peak telephony", when adding together volumes of fixed+mobile outbound calls. Quite a few are now past "peak mobile telephony" as well, although others are still substituting fixed for mobile, as prices fall because of competition. 

Even those with growing overall call usage are seeing this driven by increasing numbers of users not usage per person, for example in India, where few people had phones at all in the past. Sometimes there's a blip of pent-up demand when flat-rate plans come into the market, or there's a recovery from an economic crash, but that's a one-off gain, not sustained growth.

But the simple fact of the matter is that once people have access to phones & (cheap) calls, there's a limit to how much they want to use. Use-cases are eroded by email, messaging, video communications, social networks, app-based interactions, notifications and the 100 other ways of connecting people and computers - even before WebRTC's help in accelerating the trend. 

There are, simply, better ways of doing almost anything than via a normal phone call.

This is already exhibited by usage trends, before we've even started the main part of this transition period. For example:

• The CTIA's annual survey reported a drop of 6.2% in US mobile voice-call minutes from 2013 to 2014, despite a broad shift to flat-rate plans and a 6% increase in the number of subscriptions. A similar fall in 2015 seems plausible.
• China Mobile did the same number of voice-calling minutes (1.60trn) in Q3'2015 as it did in Q3'2012, despite growing its customer base from 699m to 823m over the same period. That's a 15% decline in average use in three years. (Link)
• In the UK, EE has recently reported average minutes-of-use falling 4.4% year-on-year on average - and 8.6% for postpaid users.
• France bucks the trend - it has seen mobile telephony usage rise consistently, although that's perhaps because more competition from Free (and lower prices) has driven elasticity in usage. The chart on p32 here indicates that growth is flatter, though

In business telephony & UCaaS, there are mixed reports of growth by company, vertical and user role. But if you take out pre-arranged dial-ins to conference bridges (not really a "call" as such), the picture isn't pretty either.

In short, mobile telephony is (or has) peaked. I use fewer minutes per month on my phone than last year, and I bet you do as well. Upgrading to VoLTE won't change that, especially if it's less reliable.

When was the last time you phoned for a taxi, or called someone to invite them to a birthday party? Unless you work in telesales, how often do you use your phone - desk or mobile - to call a client or industry contact? How often do you scowl or tut, when your phone rings unexpectedly and distracts your concentration?

Phone calls won't disappear, no. But then there's plenty of fax machines still around, too.

More importantly still, if demand is falling or flat, while supply of telephony or near-substitutes is rising, the only economic outcome is falling prices. The revenue attributable to voice telephony is falling, even if volumes are holding up.

The painful fact is that telcos don't do "voice" in a general sense. They only do telephony, with an occasional side-order of voicemail, conferencing and push-to-talk. They are almost entirely absent from the 100s (soon to be 1000s) of other voice applications and use-cases. 


VoLTE is misnamed. It's just ToLTE. 

What else is there in voice besides phone calls? 

How about voice chat, asynchronous voice, speech-to-text, recording, real-time translation, priority voice, voice recognition, voice biometrics, voice messaging, stereo audio, aid for the blind, peer-to-peer, realtime captioning, voice PaaS, emergency group voice, gunshot detection, nurse-call functions, radio, virtual assistants, emotional analysis, karaoke, encrypted voice, hypervoice, IoT integration - and so on. 

Those are not phone calls, not billed (or measured) per-minute, not necessarily using phone numbers, often not subject to lawful interception, not regulated - but they are huge sources of current and potential value. But they need thought and innovation, not blind standards.

Then add in the whole rapidly-inflating universe of contextual communications and analytic/metadata-enriched voice, and you start to get the full picture. I'll cover intersections with machine-learning and IoT another time. Telephony is Alexander Graham Bell's view of voice communications, not the 21st century's. (Obviously, regulators and governments are even slower than telcos to work this out, though). 

(A couple of telcos have tried updating VoLTE to pre-announce why people are calling. Good idea, badly implemented. A topic for another post. Hint: blame RCS).

The voice-based service that's currently got the most buzz? No, not Skype or Viber or Whatsapp or any of the UC vendors. It's Amazon Echo and its Alexa virtual assistant. Echo has been around for a while, but in the last month or two everyone seems to have suddenly started creating integrations and cool services. "Alexa, get me an Uber", "Alexa, play Lady Gaga's Telephone on Spotify". Or, ironically, "Alexa, send my calls to voicemail", by innovative VoIP provider Ooma (link).

Add that to Apple Siri, Microsoft Cortana and Google Now, plus assorted other virtual assistants that are becoming the true face of future mobile (or fixed) voice communications. It will be (in part) about input/output and access to information. Not upgrading the string between two tin-cans to IP.

There's an upcoming conference on Mobile Voice in a couple of months. There are no mobile operators or even VoLTE vendors on the draft speaking roster. Ironically, the only CSP represented is cable operator Comcast. Forget the upcoming dinosaur exhibition in Barcelona - this is where the new mobile voice DNA is being developed. 

The obsession with VoLTE is just rearranging the deckchairs on the Titanic. Or creating an IP-powered horse buggy in the age of the car. Create your own analogy - there's plenty of examples of "missing the wood for the trees".


VoLTE is retro

In fact, perhaps the best analogy for VoLTE is the new-shape VW Beetle introduced in 1997, which took a lot of styling clues from the classic car made from 1938 onwards, updated for the modern era. It perpetuated a decades-old model, but with an underlying change in platform. It evolved from rear-engine/RWD, to front-engine/FWD - the equivalent of moving from TDM to IP in terms of phone calls. It still had 4 wheels, it was still internal-combustion, it was still sold via normal dealerships and distribution, and was financed in the usal ways. However, it was more complex and expensive than its predecessor.

While the New Beetle clearly fitted a market demand, and has been moderately successful, in hindsight it was not a game-changer in any way. It didn't represent the "future of cars". It hasn't been a blockbuster, let alone the "people's wagon". It was an up-to-date retro take on an old classic. The original Beetle sold about 21m units in 54 years. The New Beetle sold just 1m in its first 10 years to to 2008, and (with another refresh in 2012) it will probably do about the same in the next 10 years. People think it's quite cute, but it holds nowhere near the universal appeal of the original.

(I'd perhaps contrast the Beetle with the new Mini, which has spawned lots of different variants and now seems almost as iconic as the original 1950s Issigonis classic. Or perhaps another car analogy VoLTE should have emulated is the Porsche 911, which has continued to evolve incrementally for decades - it keeps the family resemblance, but is always fresh. And I'm not even going to invoke Tesla, Toyota Priuses, Google self-drive cars, car-club business models and other innovations).


Summary

VoLTE is the right answer, 7 years too late. It should have been part of LTE at launch on Day 1, designed-in and standardised during Year Minus-3. The industry forgot about it, and has been in panicky catch-up mode ever since, trying to shoehorn immature packet voice onto a complex new air-interface and core network. The focus has been on network QoS, standards and spectrum re-farming, rather than communications purpose, user-experience, analytics, devices or applications. The word "acoustic" didn't even appear in the original IR92 specifications.

It's too little, too late. It's a IP-homage to ye olde phone networkes of yore, not a shiny vision of the future of communications. Yes, we'll likely see continued growth in VoLTE deployments, with vendors making some money, while telco engineering departments look "busy" and justify their existence and budgets. 

Instead of putting R&D efforts in, hiring people who understand speech and behaviour, educating regulators, experimenting with new capabilities and models, or learning from other innovators, we get a group of people titled "head of voice" who mostly seem confused by the lack of Bakelite and rotary dials. And vendors happy to entrench the confirmation bias.

Try an experiment over the next month, especially at MWC: Mentally replace the word "voice" with "telephony" wherever you see or hear it. (Or replace VoLTE with ToLTE). It could be in press releases, CEO speeches, booth demos, business-card job titles, analyst research reports, Tweets or whatever. 

If voice/telephony makes no difference to the sentence's meaning, then that person or company simply doesn't get it. They're not the future of mobile voice - they're the past. They're telco equivalents of VW marketeers, with equally dodgy stats and polluting emissions.