I'm seeing a huge amount of interest in the early definition of 5G networks, which are expected to start appearing sometime around 2020 - even though there is thus far no formal definition. I attended a Huawei-sponsored event about 5G in Munich a couple of months ago, which is just one of many similar conferences and gatherings involving most of the traditional industry. And there certainly seems to be a long list of technologies - and potential requirements - vying for inclusion and consideration.
However, in my view 5G should take a different path to standardisation to 4G/LTE. That process definitely had some highlights - especially the bringing-together of the formerly separate CDMA and GSM/UMTS worlds. LTE has also taken off rapidly in some countries (especially the US, Japan and South Korea), paralleling and being catalysed by the rise of smartphones.
On the other hand, LTE has some downsides. In terms of business model and user-behaviour, it is still largely "like 3G use, but more so". It's faster, cheaper (per-MB) and has lower latency. But it's also often patchy in coverage, has far too many separate frequency bands, and of course is sub-optimal for telephony, with CSFB's compromises and VoLTE's huge delays and cost/complexity, occurring right at the same time as "peak telephony". The LTE speedboat has had to drag the ugly anchor-weight of IMS along with it, as the 3GPP standards have meshed them so tightly. (Or to use my 2009 metaphor, the dead parrot of IMS has been nailed to the LTE perch).
LTE also sits somewhat uneasily with the growth of WiFi almost everywhere. Ignore the HetNet hype for a moment - most WiFi is, and will continue to be, totally separate from the mobile network. WiFi is mostly either private (part of a home or office LAN enviroment), controlled by fixed/cable carriers, or provided as an amenity (rather than a service) by venue owners, event organisers, software developers and others.
There is a reason for this - WiFi is not constrained to just a single business/user-interaction model, ie a "subscription". It does not need a SIM card. It can be subscribed-to with an ongoing business relationship, or it can be transient, free, sponsored, venue-based, time-based, anonymous, tethered or assorted other approaches. This stands in contrast to 3G/4G, which for all its lobbyist whining about Net Neutrality, still comes as a Henry Ford-like "any business model you like, as long as it's a subscription".
I've never had a conference organiser (or a cafe) give me a code for "free LTE" while on-site.
This is partly because WiFi's technical standards (defined by IEEE & WiFi Alliance) do not include elements that pre-define its usage model. It does not need a SIM (subscriber ID module - the hint is in the name). There are multiple authentication models, and can have "users" rather than subscribers, with access not requiring many of the characteristics expected in the cellular world. As such, it is a "multi-stakeholder" technology - it involves network operators (fixed & mobile), end-users, enterprises, device vendors, developers, municipalities, venue owners, tenants, OS suppliers, aggregators, advertisers and many other interested parties. WiFi also does not mandate a specific service or control infrastructure - while IMS can theoretically be used, it in almost all cases is not.
As a result, unlicenced wireless data has a hugely diverse range of use-cases and manifestations, which has driven immense amounts of innovation, consumer benefit, application consumption and, ultimately, economic and social gain.
We should be thinking the same way for 5G network architecture. We need (we=governments, users, regulators, vendors, investors) to make sure that it too is "multi-stakeholder". Unlike 4G/LTE, we have an opportunity now for many other groups to get involved in defining "requirements" for 5G, and especially, making sure that whatever technical standards emerge do not constrain either business models, nor application/user interactions. Clearly, if it is expected to operate in licenced spectrum, it will need adequate mechanisms for management (especially for interference and probably aspects of performance/quality), but it needs to be technology-neutral end to end. In particular, it needs to be core-neutral and not assume a particular architecture.
Interestingly, we're already seeing discussions to put LTE into unlicenced bands - but at the moment, just with the same structures and architectures as "normal" cellular.
In order to get to those endpoints, the discussion needs to involve many more parties than just the cosy vendor/MNO process seen in 3G and 4G. The upfront discussions defining the technology need to involve a similar - indeed, broader - range of parties to WiFi.
While clearly Vodafone, Verizon, Ericsson & Huawei will need to be heavily involved in the 5G technology definition, so do Ford, Boeing, IBM, Comcast, Google, General Electric, Hilton, Apple, GlaxoSmithkline, Sony, Disney, ABB, WPP, Shell, Westfield, Starbucks, NATO and the Greater London Authority. There also need to be marketeers, behavioural psychologists and other social scientists involved, who can help steer the direction towards what customers actually want - rather than just what (old, mostly male) engineers think we should have.
Governments and regulators need to get involved immediately to ensure that 5G standards definition is not inherently anti-competitive. If we believe in the Internet of Things, eGovernment, Cloud, wearables and digital inclusion, we need to ensure that 5G is not just "4G on steroids". With the coming of virtualisation, we also need to make 5G much more easily "hackable", especially if it's used in unlicenced spectrum. Bits of the radio technology should have developer kits, or even be open-sourced if possible. As long as there is adequate protection against interference, 5G should allow experimentation. We need to be certain that there is not a cartel-like grip on IPR, that funnels 5G into being (to all intents and purposes) merely an overlay/upgrade for 3G and 4G networks. The idea of wholesale - at multiple levels - needs to be ingrained upfront as well.
It should be remembered that the telecom industry is not the only source of capex or managed-services opex. It is in ALU & Ericsson & Huawei's interests to develop versions of 5G that they can sell direct to governments, Facebook, Exxon and electricity companies, as well as traditional network operators. We already see LTE starting to appear for public safety or industrial uses - the future architecture of 5G needs to enable that approach to be expanded massively, especially as the telecom industry consolidates inevitably in coming years.
This probably means that bodies like 3GPP and ETSI are not the right places to start. It is questionable if the new 5G-PPP organisation is, either. It is not obvious to me that historically non-cellular companies (eg Toyota, or medical device vendors, or train operators) will easily be able to fit into the clubby telecom-standards world processes and strictures. This is already seen in the attempts by the mobile industry to embrace/subsume WiFi, where the other stakeholders are effectively excluded from many of the technical discussions. (To Huawei's credit, the 5G Munich event included BMW and a number of other non-traditional participants, although none obviously from the web world).
In a nutshell, as we go towards 2020 and 5G, and as mobile technology becomes more pervasive and important, it is critical that we make sure, upfront, that other voices are heard, and that we don't find the standards process just steam-rollering its way to perpetuating the past.
Regulators and governments should inspect the underlying assumptions - for example whether 5G is always going to be a "service" or whether it can also be "owned", or provided as a utility or amenity.
There should be nothing in the technology to preclude this Elements like SIM cards & IMS cores can still remain - but should be entirely optional. The radio technology needs to be decoupled from transmission, from the core, from the service layer, and from software - ideally with open APIs throughout. Equally, there should be no assumption that 5G is to be used just for Internet access - it should be neutral to "back end" network infrastructure and service domain as well.
5G should be open and exploitable by satellite, drone, balloon and device-to-device innovators, as well as traditional base-station infrastructure providers.
In brief, telecoms is becoming too important to just leave it up to the telcos and their vendors. Government needs to exert a heavier hand to make sure 5G standards are not just stitch-ups, excluding newcomers that can prove to be true sources of innovation and value. They need to ensure that tempting investments and lobbying prowess from the incumbent cellular world do not skew the playing field. And other parties, from consumer electronics to property to vehicle manufacture to app developers to defence, need to get involved NOW, and ensure that 5G pre-research includes them, and reflects their needs upfront.
However, in my view 5G should take a different path to standardisation to 4G/LTE. That process definitely had some highlights - especially the bringing-together of the formerly separate CDMA and GSM/UMTS worlds. LTE has also taken off rapidly in some countries (especially the US, Japan and South Korea), paralleling and being catalysed by the rise of smartphones.
On the other hand, LTE has some downsides. In terms of business model and user-behaviour, it is still largely "like 3G use, but more so". It's faster, cheaper (per-MB) and has lower latency. But it's also often patchy in coverage, has far too many separate frequency bands, and of course is sub-optimal for telephony, with CSFB's compromises and VoLTE's huge delays and cost/complexity, occurring right at the same time as "peak telephony". The LTE speedboat has had to drag the ugly anchor-weight of IMS along with it, as the 3GPP standards have meshed them so tightly. (Or to use my 2009 metaphor, the dead parrot of IMS has been nailed to the LTE perch).
LTE also sits somewhat uneasily with the growth of WiFi almost everywhere. Ignore the HetNet hype for a moment - most WiFi is, and will continue to be, totally separate from the mobile network. WiFi is mostly either private (part of a home or office LAN enviroment), controlled by fixed/cable carriers, or provided as an amenity (rather than a service) by venue owners, event organisers, software developers and others.
There is a reason for this - WiFi is not constrained to just a single business/user-interaction model, ie a "subscription". It does not need a SIM card. It can be subscribed-to with an ongoing business relationship, or it can be transient, free, sponsored, venue-based, time-based, anonymous, tethered or assorted other approaches. This stands in contrast to 3G/4G, which for all its lobbyist whining about Net Neutrality, still comes as a Henry Ford-like "any business model you like, as long as it's a subscription".
I've never had a conference organiser (or a cafe) give me a code for "free LTE" while on-site.
This is partly because WiFi's technical standards (defined by IEEE & WiFi Alliance) do not include elements that pre-define its usage model. It does not need a SIM (subscriber ID module - the hint is in the name). There are multiple authentication models, and can have "users" rather than subscribers, with access not requiring many of the characteristics expected in the cellular world. As such, it is a "multi-stakeholder" technology - it involves network operators (fixed & mobile), end-users, enterprises, device vendors, developers, municipalities, venue owners, tenants, OS suppliers, aggregators, advertisers and many other interested parties. WiFi also does not mandate a specific service or control infrastructure - while IMS can theoretically be used, it in almost all cases is not.
As a result, unlicenced wireless data has a hugely diverse range of use-cases and manifestations, which has driven immense amounts of innovation, consumer benefit, application consumption and, ultimately, economic and social gain.
We should be thinking the same way for 5G network architecture. We need (we=governments, users, regulators, vendors, investors) to make sure that it too is "multi-stakeholder". Unlike 4G/LTE, we have an opportunity now for many other groups to get involved in defining "requirements" for 5G, and especially, making sure that whatever technical standards emerge do not constrain either business models, nor application/user interactions. Clearly, if it is expected to operate in licenced spectrum, it will need adequate mechanisms for management (especially for interference and probably aspects of performance/quality), but it needs to be technology-neutral end to end. In particular, it needs to be core-neutral and not assume a particular architecture.
Interestingly, we're already seeing discussions to put LTE into unlicenced bands - but at the moment, just with the same structures and architectures as "normal" cellular.
In order to get to those endpoints, the discussion needs to involve many more parties than just the cosy vendor/MNO process seen in 3G and 4G. The upfront discussions defining the technology need to involve a similar - indeed, broader - range of parties to WiFi.
While clearly Vodafone, Verizon, Ericsson & Huawei will need to be heavily involved in the 5G technology definition, so do Ford, Boeing, IBM, Comcast, Google, General Electric, Hilton, Apple, GlaxoSmithkline, Sony, Disney, ABB, WPP, Shell, Westfield, Starbucks, NATO and the Greater London Authority. There also need to be marketeers, behavioural psychologists and other social scientists involved, who can help steer the direction towards what customers actually want - rather than just what (old, mostly male) engineers think we should have.
Governments and regulators need to get involved immediately to ensure that 5G standards definition is not inherently anti-competitive. If we believe in the Internet of Things, eGovernment, Cloud, wearables and digital inclusion, we need to ensure that 5G is not just "4G on steroids". With the coming of virtualisation, we also need to make 5G much more easily "hackable", especially if it's used in unlicenced spectrum. Bits of the radio technology should have developer kits, or even be open-sourced if possible. As long as there is adequate protection against interference, 5G should allow experimentation. We need to be certain that there is not a cartel-like grip on IPR, that funnels 5G into being (to all intents and purposes) merely an overlay/upgrade for 3G and 4G networks. The idea of wholesale - at multiple levels - needs to be ingrained upfront as well.
It should be remembered that the telecom industry is not the only source of capex or managed-services opex. It is in ALU & Ericsson & Huawei's interests to develop versions of 5G that they can sell direct to governments, Facebook, Exxon and electricity companies, as well as traditional network operators. We already see LTE starting to appear for public safety or industrial uses - the future architecture of 5G needs to enable that approach to be expanded massively, especially as the telecom industry consolidates inevitably in coming years.
This probably means that bodies like 3GPP and ETSI are not the right places to start. It is questionable if the new 5G-PPP organisation is, either. It is not obvious to me that historically non-cellular companies (eg Toyota, or medical device vendors, or train operators) will easily be able to fit into the clubby telecom-standards world processes and strictures. This is already seen in the attempts by the mobile industry to embrace/subsume WiFi, where the other stakeholders are effectively excluded from many of the technical discussions. (To Huawei's credit, the 5G Munich event included BMW and a number of other non-traditional participants, although none obviously from the web world).
In a nutshell, as we go towards 2020 and 5G, and as mobile technology becomes more pervasive and important, it is critical that we make sure, upfront, that other voices are heard, and that we don't find the standards process just steam-rollering its way to perpetuating the past.
Regulators and governments should inspect the underlying assumptions - for example whether 5G is always going to be a "service" or whether it can also be "owned", or provided as a utility or amenity.
There should be nothing in the technology to preclude this Elements like SIM cards & IMS cores can still remain - but should be entirely optional. The radio technology needs to be decoupled from transmission, from the core, from the service layer, and from software - ideally with open APIs throughout. Equally, there should be no assumption that 5G is to be used just for Internet access - it should be neutral to "back end" network infrastructure and service domain as well.
5G should be open and exploitable by satellite, drone, balloon and device-to-device innovators, as well as traditional base-station infrastructure providers.
In brief, telecoms is becoming too important to just leave it up to the telcos and their vendors. Government needs to exert a heavier hand to make sure 5G standards are not just stitch-ups, excluding newcomers that can prove to be true sources of innovation and value. They need to ensure that tempting investments and lobbying prowess from the incumbent cellular world do not skew the playing field. And other parties, from consumer electronics to property to vehicle manufacture to app developers to defence, need to get involved NOW, and ensure that 5G pre-research includes them, and reflects their needs upfront.
Interesting viewpoint. As someone who has been thinking about 5G now for the last 3 or 4 years, I find your thoughts refreshingly different.
ReplyDeleteThe problem I have is that all cellular mobile technologies, because of their requirement for ubiquity, are all "top-down". It costs billions to deploy a national network, and so the business model requires millions of subscribers.
Wi-Fi is a "bottom-up" technology, that can be deployed at zero cost to the applications or core providers, because the cost of spectrum is nothing and the access points cost a few hundred dollars, and can be subsidized by other sources of income such as hotel rates, coffee or book purchases, or seminar fees.
The idea of making 5G (or 4G come to that) more "Wi-Fi" like keeps foundering on the economic realities of expensive spectrum and network infrastructure, which demands a captive audience (subscribers) to make it viable. 4G/5G Small Cells go part way, but like Moses, don't quite get to the promised land.
Another DB - got to be on the right track with those initials!
ReplyDelete5G is going to be a mix of the two, top-down & bottom-up. Especially as it's going to (probably) include less-expensive spectrum >6GHz as well as potentially include unlicenced.
It may be that we have to frame the licence terms for those bands in such a way that ubiquity is *not* mandated. Maybe it looks more like electricity grids, which have both central & distributed/local generation?
I don't have the answers yet - I'm just saying we need to be framing the questions in such a way that it doesn't preclude some of these possibilities.
What we *must* guard against is the tendency of the cellular industry to try to force top-down strictures onto the parts of the market that are bottom-up (eg WiFi). Notable that the Israeli regulator was last year thinking about banning WiFi offload by cellcos.
Dean
This comment has been removed by the author.
ReplyDeleteSorry "DB", but my "DB" is just a nom-de-plume :)
ReplyDeleteI've been in wireless (vendor product marketing in the UK and the US) for 20 years, but I don't think we've ever met.
Above 6GHz is going to mean relatively poor propagation and building penetration, which means very small cells and indoor access points - much more Wi-Fi-like. So it may well end up become a mixed model with public/private ownership.
The key may be better bits/sec/Hz/sq km, which plays to the small cell world. Unless something comes along with better outright spectral efficiency so that under 6 GHz can be refarmed. I right now know of 2 ways to do that in theory, and I'm working with the inventors to try and prove it out in practice. It's a matter of sneaking around Shannon when he isn't looking.....
"Sneak around Shannon"....
ReplyDeleteMy expectation is we'll move towards beamforming to give the equivalent of "switched wireless", with each lobe just serving one/a few users, at close to "line rate".
Although that may just reflect my memory of LANs going from being shared-media to switched-ethernet!