3rd Neutral Host Workshop + OpenRAN for shared networks. Early bird still available

NOTE: Owing to uncertainty around the impact of Coronavirus on travel, event attendance, company policies & venues, this workshop has been postponed from 31st March until 7th July. We have contacted existing registered attendees to discuss the options

On March 31st  July 7th I'll be running my 3rd public workshop on Neutral Host Networks in central London, together with colleague Peter Curnow-Ford.

As well as covering the basics of new wholesale/sharing models for MNOs, both with and without dedicated spectrum, we will also be looking more closely at the fit between NHNs and new virtualised vRAN / OpenRAN technologies. 

We'll cover all the various use-cases: metro-area network densification, indoor systems for various venues, road/rail coverage, rural wholesale models, FWA and more. 

The links (and differences) between neutral-host and private LTE/5G will be discussed, as well as alternative models such as multi-MNO sharing or national roaming. (see this post for some previous thoughts on this)

Different countries' competitive, regulatory and spectrum positions will be covered, to assess how that will impact the evolution of NHNs. 
 
Early bird pricing is available before June 7th.

Full details and registration are available here

Predictions for the next decade: looking out to 2030 for telecoms, wireless & adjacent technologies

It's tempting to emulate every other analyst & commentator and write a list of 2020 predictions of success and failure. In fact, I got part-way into a set of bulletpoints about what’s overhyped and underhyped. 

But to be honest, if you read my articles and tweets, you probably know what I think about 2020 already. Private cellular networks will be important (4G, initially). 5G fixed wireless is interesting and will grow the FWA market - but won't replace fibre. 5G is Just Another G and is overhyped, especially until the new core matures. RCS is still a worthless zombie, eating brains. But I don't need to repeat all this in detail, just because I'm a bit more sharp-worded than most observers. It wouldn't tell you much new.

But seeing as I spend a fair amount of time advising clients about the longer-term future, 5-10 years out or even further, I thought I'd set my sights higher. I use the term "telco-futurism" to look at the impacts of technology and broader society on telecoms, and vice versa.

So, at the start of the 2020s, what about the next decade? Assuming I haven't retired to my palatial Mars-orbiting private Moon in 10 years' time, what do I think I'll be writing, podcasting (or neural-transmitting) about in 2030?

So, let's have a few shots at this more-distant target...

• 6G: In 2030, the first 6G networks are already gaining traction in the marketplace. The first users are still fixed connections to homes, and personal devices that look a bit similar to phones and wearables, but with a variety of new display and UI technologies, including contact lenses and advanced audio/haptic interfaces. 6G represents the maturing of various 5G concepts (such as the new core), plus greater intelligence to allow efficient operation. 
• Details, details: Much of the 2020s will have been spent dealing with numerous "back-office" problems that have stopped many early 5G visions becoming real. Network-slicing will have thrown up huge operationalisation and security issues. Dealing with QoS/slice roaming or handoff, at borders between networks (outdoor / indoor / private / neutral / international) will be hugely complex. Edge computing scenarios will turn out to need local peering or interconnection points. All of these will have huge extra complexities with billing, pricing and monitoring. mmWave planning and design tools will need to have matured, as well as the processes for installation and operation.Training and skills for all of this will have been time-consuming and expensive - we'll need hundreds of thousands of experts - often multi-domain experts. By the time all these issues get properly fixed, 6G radios and vendors will exploit them, rather than the "legacy 5G" infrastructure. See this post for my discussion about the telecom industry's problems with accurate timelines.
• Device-Network cooperation: By 2030, mobile ecosystems and control software will break today's silos between radio network, devices and applications much more effectively. Sensors in users' devices, cell-towers and elsewhere will be linked to AI which works out how, why and where people or IoT objects need connectivity and how best to deliver it. Recognise a moving truck with machine-vision, and bounce signals off it opportunistically. Work out that someone is approaching the front of a building, and pre-emptively look for Wi-Fi, or negotiate with the in-building neutral host on a marketplace before they enter the door. Spot behavioural patterns such as driving the same route to work, and optimise connectivity accordingly. Recognise a low battery, and tweak the "best-connected" algorithm for power efficiency, and downrate apps' energy demand.Integrate with crowd-flow patterns or weather forecasts. There will be thousands of ways to improve operations if networks stop just thinking of a "terminal" as just an endpoint, and look for external sources of operational data - that's a 20th Century approach. Expect Google's work on its Fi MVNO & Android/Pixel phones, and similar efforts by Samsung and maybe Apple, Qualcomm and ARM, to have driven much of this cross-domain evolution.
• Energy-aware networks: Far more energy-awareness will be designed into all aspects of the network, cloud and device/app ecosystem. I'm not predicting some sort of monolithic and integrated cascading-payments system linked into CO2-taxes, but I expect "energy budget" to be linked much more closely to costs (including externalities) in different areas. How best to optimise wired/wireless data for power demand, where best to charge devices, "scavenging" for power and so on. Maybe even "nudge" people to lower-energy applications or consumption behaviours by including "power-shaming" indicators. If 3GPP and governments get their act together, as well as vendors & CSPs, overall 6G energy use will be a higher priority design-goal than throughput speed and latency.
• Wi-Fi: We'll probably be on Wi-Fi 9 by 2030. It will continue to dominate connectivity inside buildings, especially homes and business premises with FTTX broadband (i.e. most of them in developed markets). It will continue to be used for primary connectivity on high-throughput / low-margin / low-mobility devices like TVs and display screens, PC-type devices, AR/VR headsets and so on. It will be bonded together with 5G/6G and other technologies with ever-better multi-path mechanisms, including ad-hoc device meshes. Ease of use will have improved, with the success of approaches like OpenRoaming. Fairly little public Wi-Fi will be delivered by "service providers" as we think of them today.  We'll probably still have to suffer the "6G will kill Wi-Fi" pundit-pieces and hype, though.
• Spectrum: The spectrum world changes slowly at a global level, thanks to the glacial 4-year cycle of ITU WRCs. By 2030 we will have had 2023 and 2027 conferences, which will probably harmonise more spectrum for 5G/6G, satellites & high-altitude platforms (HAPS) and Wi-Fi type unlicensed use. The more interesting developments will occur at national / regional levels, below the ITU's role, in how these bands actually get released / authorised - and especially whether that's for localised or shared usage suitable for private networks and other innovators. By 2030 we should have been through 2+ cycles of US CBRS and UK/Germany/Japan/France style local licensing experiments, allocation methods, databases and sensing systems. I think we'll be closer to some of the "spectrum-as-a-service" models and marketplaces I've been discussing over the last 24 months, with more fluid resale and temporary usage permits. International allocations will still differ though. We will also see whether other options, such as "national licenses with lots of extra conditions" (eg MVNO access, rural coverage, sharing, power use etc) has helped maintain today's style of MNOs, despite the grumbling. We will also see much more opportunism and flexibility in band support in silicon/devices, and more sophisticated approaches to in-band sharing between different technologies. I'm less certain whether we will have progressed much with commercialisation of mmWave bands 20-100GHz, especially for mobile and indoor use. It's possible and we'll certainly see lots of R&D, but the practicalities may prove insuperable for wide usage.
• Private/neutral cellular: Today, there's around 1000 MNOs globally (public and private). By 2030, I'd expect there to be between 100,000 and a million networks, probably with various new types of service provider, aggregation hubs and consortia. These will span industrial, city, office, rural, utility, "public venue" and many other domains. It will be increasingly hard to distinguish private from public, eg with MNOs' campus networks with private cores and hybrid public/private spectrum. We might even get another zero, if the goals of making private 4G/5G as easy and cheap to build as Wi-Fi prove feasible, although I have doubts. Most of these networks will be user-specific, but a decent fraction will be multi-tenant, either offering wholesale access or roaming to "legacy MNOs" as neutral hosts, or with some sort of landlord model such as a property company running a network with each occupied floor or building on campus as a "semi-private" network. Some such networks will look like micro-telcos (eg an airport providing access to caterers & airlines) and will need billing, management & security tools - and perhaps new forms of regulation. This massive new domain will help catalyse various shifts in the vendor community as well - especially cloud-native core and BSS/OSS, and probably various forms of open RAN, and also "neutral edge".
• Security & privacy: I'm not a security expert, so I hesitate to imagine the risks and responses 10 years out. Both good and bad guys will be armed to the teeth with AI. We'll see networks attacked physically as well as logically. We'll see sophisticated thefts of credentials and what we quaintly term "secrets" today. There will be cameras and mics everywhere. Quantum threats may compromise encryption - and other quantum tools may enhance it, as well as provide new forms of identity and authentication. We will need to be wary of threats within core networks, especially where orchestration and oversight is automated. I think we will be wise to avoid "monocultures" of technologies at various levels of the network - we need to trade off efficiency and scale vs. resilience.
• Satellite / HAPS: We'll definitely have more satellite constellations by 2030, including some huge ones from SpaceX or others. I have my doubts that they will be "game-changers" in terms of our overall broadband use, except in rural/remote areas. They won't have the capacity of terrestrial networks, and signals will struggle with indoor penetration and uplink from anything battery-powered. Vehicles, planes, boats and remote IoT will be much better-connected, though. Space junk & cascading-collision scenarios like the movie Gravity will be a worry, though. I'm not sure about drones and balloons as HAPS for mass-market use, although I suspect they'll have some cool applications we don't know today.
• Cloud & edge: Let's get one thing clear - the bulk of the world's computing cycles & data storage will continue to occur in massive datacentres (perhaps heading towards a terawatt of aggregate power by 2030) and on devices themselves, or nearby gateways. But there will be a thriving mid-market of different sorts of "edge" as I've covered in many posts and presentations recently. This will partly be about low-latency, but not as much as most people think. It will be more about saving mass data-transport costs, protecting "data sovereignty" and perhaps optimising energy consumption. A certain amount will be inside telcos' networks, but without localised peering / aggregation this will be fairly niche, or else it will be wholesaled out to the big cloud players. There will be a lot of value in the overall orchestration of compute tasks for applications between multiple locations in the ecosystem, from chip-level to hyperscale and back again. The fundamental physical quantum of much edge compute will be mundane: a 40ft shipping container, plonked down near sources of power and fibre.
• Multi-network: We should expect all connectivity to be "software-defined" and "multi-network". Devices will have lots of radios, connecting simultaneously, with different paths and providers (and multiple eSIM / other identities). Buildings will have mutliple fibres, wireless connections and management tools. Device-to-device connections and relaying will be prevalent. IoT will use a selection of LPWAN technologies as well as Wi-Fi, cellular and short-range connections. Satellite and maybe LiFi (light-based) connections will play new roles. Arbitrage, bonding, load-balancing will occur at multiple levels from silicon to OS to gateway to mid-network. Very few things will be locked to a single network or provider - unless it has unique value such as managed security or power consumption.
• Voice & messaging: Telephony will be 150yo in 2026. By 2030 we'll still be making some retro-style "phone calls" although it will seem even more clunky, interruptive, unnatural and primitive than today. (It won't stop the cellular industry spending billions upgrading to Vo6G though). SMS won't have disappeared, either. But most consumers will communicate through a broad variety of voice and video interaction models, in-app, group-based, mediated by an array of assistants, and veracity-checked to avoid "fake voice" and man-in-the-middle attacks of ever increasing subtlety. Another 10 years of evolution beyond emojis, stories, filters and live broadcasts will allow communication which is expressive, emotion-first, and perhaps even richer and more nuanced than in-person body language. I'm not sure about AR/VR comms, although it will still be more important than RCS which will no doubt be celebrating its 23rd year of irrelevance, hype and refusal to die.
• Enterprise comms:  UCaaS, cPaaS and related collaboration tools will progress steadily, if unspectacularly - although with ever more cloud focus. There will be more video, more AI-enriched experiences for knowledge management, translation, whispered coaching and search. There will be attempts to reduce travel to meetings and events as carbon taxes bite, although few will come close to the in-person experience or effectiveness. We'll still have some legacy phone calls and numbers (as with consumer communications) although these will be progressively pushed to the margins of B2B and E2E interactions. Ever more communications will take place "contextually" - within apps, natively supported in IoT devices, or with AI-based assistants. Contact centres and customer interactions will be battlegrounds for bots and assistants on both sides. ("Alexa, renegotiate my subscription for a better price - you have permission to emulate my voice"). Security and verification will be highly prized - just because something is heard doesn't mean it will match what was originally spoken
• Network ownership models: Some networks of today will still look mostly like "telcos" in 2030,  but as I wrote in this post the first industry to be transformed by 5G will be the telecom industry itself. We'll see many new stakeholders, some of which look like SPs, some which are private network operators, and many new forms of aggregator, virtual operator, wholesale or neutral mobile/fibre provider. I'm not expecting a major shift back to nationalised or government-run networks, but I think regulations will favour more sharing of assets where it makes sense. Individual industries will take control of their own connectivity and communications, perhaps using standardised 5G, or mild variations of it. There will be major telcos of today still around - but most will not be providing "slices" to companies and offering deep cross-vertical managed services. There will be M&A which means that we'll have a much more heterogeneous telco/CSP market by 2030 than today's 800 identikit national MNOs. Fixed and fibre providers will be diverse as well - especially with the addition of cloud, utility and muncipal providers. I think the towerco / property-telco model will be important as asset owners / builders as well.

I realise that I could go on at length about many other topics here - autonomous and connected vehicles, the future of cities and socio-political spheres, shifts in entertainment models, the second wave of blockchain/ledgers, the role of human enhancement & biotech, new sources of energy and environmental technology, new forms of regulation and so forth. But this list is already long enough, I think. Various of these topics will also appear in podcasts - which I'm intending to ramp up in 2020. At the moment I'm on SoundCloud (link) but watch out here or on Twitter for announcements of other platforms.

If this has piqued your interest, please comment on my blog or LinkedIn article. This is a vision for 2030, which I hope is self-consistent and reasonable - but it is not the only plausible future scenario.

If you're interested in running a private workshop to discuss, debate and strategise around any of these topics, please get in touch via private message, or information AT disruptive-analysis DOT com. I work with numerous operators, vendors, regulators, industry bodies and investors to imagine the future of networks and other advanced technologies - and steer the path of evolution.

Happy New Year! (and New Decade)

Which will be more important for 4G/5G coverage expansion: Neutral Host, or RAN-Sharing?

There are increasing demands for better mobile coverage in areas that are technically complex, or which struggle economically with traditional MNO deployment models. 5G's use of new and higher frequencies will exacerbate the problems.

Even with a shift to pure private networks for some enterprises, there will still be a need for the public mobile networks to have better coverage for their subscribers in places such as:

• In-building locations, including both private offices and public venues
• Metro in-fill sites, needed to densify cellular networks in busy cities - but where cell-siting and connectivity challenges can be immense
• Rural areas, where mobile users are sparse and sometime lower-ARPU
• Along road and rail routes, especially where new connected vehicle uses are expected
• Anywhere with few people, but more IoT devices
• Business sites where multi-operator connectivity is needed (eg construction sites)

There are various approaches emerging to solve these issues:

• More flexible / cheaper RAN deployment options for individual MNOs to extend their own networks
• RAN sharing (including national roaming)
• Neutral host networks (NHNs)
• Various hybrid schemes with government involvement

The middle pair -  NHNs and RAN-sharing - are perhaps the two most interesting, as they fit with a lot of other developments around local and dynamic spectrum licensingto , OpenRAN and NFV, and a move to multi-MNO collaboration.

Yet which will win out, and in what contexts?
 
RAN sharing involves 2+ existing mobile operators combining network assets to save costs, perhaps through a joint venture. There are various types with differing levels of sophistication, from sharing physical towers & power, through to shared backhaul, core networks, baseband units & even spectrum. (MORAN, MOCN, etc) 

Neutral hosts are 3rd parties which build a RAN (and may have spectrum of their own) and which then sign up national MNOs or new niche/private cellular providers as tenants. Again, there are various technical and commercial models emerging. 

In theory, NHNs are more flexible, and push the capex to the new host operator. 

But what are the practicalities? Many questions arise: 

• Coverage locations & backhaul availability. What works best in rural, metro, indoor or industrial locations?
• Does an NHN need a core network? Standalone? Also VoLTE?
• Does this all apply to 4G, 5G, or both? 
• Where do OpenRAN or modern DAS & small-cells fit best? If these overlap with NFV and netwrk-slicing, can each "tenant" MNO bring its own software, if they want?
• How does security work for all parties? This is a huge and diverse minefield, relating to everything from RF interference and license conditions, to the physical integrity of network elements, down to lawful intercept and data-collection requirements.
• What are the contractual & regulatory hurdles? 
• What about other stakeholders like venue owners, property companies, towerco's and local authorities? 
• Who puts all of this together? What's the value chain, and which systems integrators and other partners will be involved?
• Will a neutral-host also offer neutral-edge computing capabilities?

There are no easy answers to all of this - the answer will generally be "it depends", both on use-case and national market.
 

I'll cover all these topics & more in next week's 2nd Neutral Host workshop in London on November 21st. Full details and registration page here: [link]

5G & Neutral-Host Thought Experiment #1

Reposted & extended from original LinkedIn post (link) - main comment thread on that page

Here's a thought experiment, to test your ideas about 5G, indoor wireless, neutral-host networks, URLLC and network-slicing.

It's a plausible scenario which seems simple, but actually has lots of complexities. It's the sort of thing that marketing departments might suggest as a use-case for 5G, but in reality, "it's not that simple".

Imagine it's the year 2025. 

There's a large office building on a business park... with a faulty elevator. The elevator company sends out one of its local maintenance engineers, who works as a contractor. 

He arrives with an AR headset, running an application to deliver repair instructions and record the fix, linked to the manufacturer's cloud-based diagnostics, image-analysis and compliance/recording platform. Given the safety issues such as fall-risks, it needs a low-latency connection to avoid the risk of nausea and distraction. 

But.... what's the network coverage like in the lift-shaft? Is there outside-in signal with <1GHz 5G? Or is there a DAS or multi-operator small-cell system? Is there a private cellular network with local spectrum? Does it support integration with all outdoor / public networks equally well? Can it support URLLC with a guaranteed SLA? What network is the engineer's headset SIM registered on, anyway? Is there a voice/video connection for looping in a remote expert? And how would that work?

Whose responsibility is all of this? Is it down to the building owner? A smart-building specialist? A neutral-host provider? Should the elevator manufacturer integrate local connectivity with Wi-Fi or 5G NR-U? How do they deal with sub-contractors? Is it possible for "slices" or performance guarantees to work on the indoor (possibly private) network? Is there a separate core network for the indoor system? Who designs, tests or pays for it? Who's liable if the network fails? Is there any need for edge-compute and storage as part of the application design - and if so, where is it and how is it accessed?
  
There are no easy answers here. The real world for many "5G" applications is going to have to deal with these heterogeneous situations, with workarounds and fallbacks. 

In this case, it seems pretty clear that the AR headset will have to have an offline mode, with blueprints & manuals stored on itself, or the engineer's phone or PC. Or the engineer will use the headset to record video, and then go back outside the building to upload it & call in for advice. Inefficient, but safer. When good-quality coverage is available inside the elevator shaft, the work can be concluded faster & more reliably - but it won't always be possible.


This is the first of a series of "5G Thought Experiments" that will help people think more about realistic scenarios and use-cases. I'll be focusing on ones that touch on opportunities for 5G, Wi-Fi6, neutral-host, cloud-native and private cellular. I'll be doing some as podcasts, so sign up here. I'll also be touching on these in my upcoming Neutral Host workshop on November 21st 2019. Details here.