There is a broad assumption that 5G will end up being an umbrella platform which includes a range of radio access technologies under its heavily virtualized canopy. Many of the R&D projects feeding into 5G – from vendors, governments, operators or academic institutions – are focused on this endlessly flexible, sliceable, programmable network. But in terms of specific radio standards, the process is being driven, just as in 3G and 4G, by the 3GPP, and while there was a very wide range of submissions to its kick-off meeting last September, its processes and concerns do not seem to have changed as radically as the brand new requirements of 5G might demand.
There is no real reason why this should be so. The ultimate decisions on how 5G will look will depend on the ITU, when it decides on its candidate standards for IMT-2020, and then the market itself – which of those candidates gains mass adoption, and how exactly the technologies are deployed in the real world, and integrated with other relevant platforms. Any standards body can develop a ‘5G’ platform, as long as it can demonstrate that it meets the technical and use case requirements defined by the ITU’s Working Group 5D.
However, the weight of the 3GPP’s cellular industry membership, and its emphasis on achieving broad consensus rather than making every deep technical decision, has seen that organization seizing the initiative and finding itself, once again, almost synonymous with 5G, at least at radio level.
This raises the question of where WiFi, and its standards organization, the IEEE’s 802.11 group, fits in. WiFi has become an increasingly integral part of the 4G picture, both for mobile operators (through offload, HetNet and so on), and as an alternative broadband wireless ecosystem. In 5G, it is widely assumed that it will take an even more integrated role in the new networks, and it has multiple standards, defined or in the works, which will address the various use cases defined for 5G by the NGMN Alliance and others (802.11ah for low power wide area, 802.11ac and 11ax for high speeds and density, 802.11ad and 11ay for high frequency spectrum, and so on). But if these continue to be developed outside of the IMT-2020 and/or 3GPP processes, there is the risk – however much the WiFi Alliance and Wireless Broadband Alliance position themselves on a 5G roadmap – that WiFi will be the ‘poor relation’ in the standardization process.
That does not mean it will not be hugely significant in terms of real world commercial deployment, regardless of what the 3GPP defines, but there are significant advantages to being at the top table right from the start – advantages in terms of recognition, large-scale procurement deals, intellectual property rights and others, but also the chance to shape future standards in line with the ethos of the WiFi industry rather than the cellular. The alternative is to develop future platforms in parallel with the cellular industry and then create bridges between them at a later stage, to support multi-RAT HetNets for a wide range of cellular and non-cellular service providers.
That situation raises a real risk that there will be compromises to the vision of 5G because the standards process will be led by the same old cellular industry hands. There is new blood – far more Chinese contributions, for instance – but greater diversity would be welcome. At an IEEE 802.11 plenary meeting in November, held just after the 3GPP kicked off its 5G standards program, the dilemmas were clear to see in several presentations. Cisco’s Andrew Myles, manager of wireless and security standards and a leading light in the WiFi Alliance, put the question of whether 802.11 should be submitted to the ITU as a candidate to be an IMT-2020 standard.
In his presentation, Myles said that the IEEE needs to exert a “stronger influence to ensure WiFi industry and market experiences in a diversity of use cases are reflected in 5G requirements”. Another motivation would be “to ensure that ‘others’ do not succeed in promoting requirements that implicitly prioritize other technologies over WiFi”. (Though, of course, many contributors to the 3GPP standards come from companies with an interest in the expansion of both cellular and WiFi.)
This is an interesting debate in its own right, but even if 802.11 were submitted and approved, it would not guarantee WiFi a role at the heart of driving 5G. The IEEE has been there before – its 802.16m/WiMAX technology was an IMT-Advanced (4G) official platform but was squashed by the force of the cellular industry and its entrenched technologies and operator models. Would 802.16 have fared better had it worked closely with the 3GPP/LTE efforts and achieved the kind of consensus platform which was, in fact, attempted for WiMAX – but after the event and far too late to have a significant impact on the outcome?
The experience of WiMAX certainly shows how difficult it is for a standard to succeed in the mobile world, if it comes from outside the family. And that certainly suggests that, as well as considering whether to submit to IMT-2020, IEEE 802.11 should formulate a closer relationship with the 3GPP, to ensure that the two sets of standards are in step, and end up being naturally complementary rather than clumsily tied together, or even competitive.
This was the topic of another presentation at the November IEEE meeting, by Joseph Levy, principal engineer for standards at InterDigital, a company with a long history of patent power in 3GPP platforms, but also active in developing WiFi and converged systems for future networks. Levy had attended the 3GPP’s 5G kick-off workshop, at which there were submissions from a wide range of stakeholders. As well as member companies, some of these came from 3GPP Market Representation Partners such as the Small Cell Forum. As a standards body itself, the IEEE does not meet the requirements to be a partner of this kind, but in fact, the meeting also saw presentations from external 5G groups, such as the European Union’s Metis and 5G PPP initiatives.
In other words, the 3GPP is opening itself up to wider influence and inputs, even if its current structure makes it hard for another radio standards body to participate. But of course, it does have rising levels of cooperation with 802.11 and the WiFi community as cellular and WLAN technologies have to work together. Levy, outlining the potential for 802.11 to be a 5G RAT within the 3GPP architecture, reminded the audience of the ongoing efforts to standardize WiFi/cellular interworking. These include:
3GPP Core access via:
- WAF (Wireless Access Gateway)
- Packet Data Gateway
- (ePDG) enhanced packet data Gateway
- 3GPP Access Network Discovery and Selection Function (ANDSF)
- 3GPP RAN Controlled LTE-WLAN Interworking (RCLWI)
But Levy argues that, despite these efforts, WiFi remains poorly integrated with 3GPP core networks. He cites:
- Limited Control of Data Flow (for multiple RATs)
- Limited 802.11 configuration control (only some elements)
- No real time handover, limited to relative performance rules
- No true QoS capability, only some relative best effort rules
- No real time access control, limited non-real time access control
- Limited service control
- Limited support of security, user identification, and privacy
Not all these issues can or should be addressed in fundamental standards. Many improvements are made in response to requirements which emerge in the real world, and are driven by industry bodies before being submitted to the 3GPP or IEEE. For instance, the Small Cell Forum has a partnership with the Wireless Broadband Alliance, and they recently released joint specifications for Trusted WLAN interfaces, to connect WiFi access points and controllers to 3GPP-defined Trusted WLAN gateways in an integrated small cell/WiFi network.
However, while such projects are important enablers of the HetNet, they are very much ‘after the event’ standardization – they do not see the WiFi community pushing its core technologies into the base level platforms for new networks.
This is what leads Levy to ask “does 802.11 attempt to enable/define/specify the capabilities to make 802.11 a fully functioning RAT that can be integrated in a 5G network?” Or alternatively, “does 802.11 continue to rely on 3GPP to define 802.11’s place in the 5G network?”
This is not just about WiFi’s political position vis-à-vis cellular, nor about practical interworking issues. The WiFi community, as Levy and Myles argue, would enrich the 3GPP process, because the 802.11 agenda increasingly dovetails with the multi-RAT architecture which most believe will be essential to support all the 5G use cases, from multi-gigabit video to ultra-low power M2M to ultra-dense venue networks.
There are very real contributions the 802.11 community could make to 5G. The lines between licensed and unlicensed spectrum technologies are blurring, even at the 4G stage. LTE is moving into licence-exempt bands with LTE-U and LTE-LAA, and the tortured process of making that work for both sides will set important precedents for true HetNets and for 5G. In reverse, WiFi could expand into an exclusive spectrum model for selected use cases, something hinted at by Andrew Myles, manager of wireless and security standards at Cisco, who gave the November presentation to kick off discussion of participation in the 3GPP deliberations.
And many of the IEEE and WiFi industry activities are covering ground which will also be important to 5G, such as hyper-density (802.11ax) and high frequency spectrum bands (802.11ay, which aims to have specifications for WiFi above 45 GHz during 2017). This will be, in effect, a supercharged version of 802.11ad and so will initially target the same 60 GHz band, but promising speeds up to 20Gbps for short range applications. However, the work to pack more bandwidth into 60 GHz would make a valuable contribution to the wider process of defining 5G standards for high bands (likely to be mainly a focus of the second wave of 3GPP work, from 2020).
Another reason is that, during the creation of the 4G standards, it was not yet obvious that there was a meaningful role for IEEE 802.11, because the process was so focused on mobility. Even WiMAX never achieved the levels of high speed mobile hand-off seen in LTE. But these days, operators are accustomed to combining the local area capabilities of WiFi with cellular’s wide area mobility, and anyway, many of the most promising 5G use cases are not highly mobile, but relate to the Internet of Things, high availability and even to fixed broadband.
All this makes it clear that the IEEE 802.11 group and associated WiFi industry have significant potential to influence the 5G process. They need to decide how best to do this – by establishing far closer ties with the 3GPP; or by developing their own 5G RAT and submitting it for IMT-2020 (or conceivably both).
The deadline to submit key definitions of proposed candidate technologies to the ITU-R WP 5D is June, though the formal submission process does not start until February 2017 – but clearly the clock is ticking.
Myles listed some key benefits of being part of IMT-2020, even while acknowledging the workload involved, and the lack of a guaranteed outcome (recalling 802.16). However, he argued that many developing countries rely on ITU recommendations when procuring or making policy; that WiFi would be able to ride on the 5G hype rather than creating its own noise; that WiFi could be better positioned for the Internet of Things.
Perhaps most importantly, in the age of blurring lines between licensed, shared and unlicensed spectrum – and the overall race for airwaves – 802.11 “might be able to access IMT-2020 identified frequency bands, including exclusive as well as shared spectrum access”. Certain countries may require IMT-2020 compliance for shared licensing in future.
Myles even suggested that the IEEE might take advantage of a 3GPP-driven technology too enhance its credentials with the ITU – notably LTE-LWA (LTE-WLAN Aggregation), part of 3GPP Release 13. This supports aggregation between LTE and WiFi at or just above the PDCP (packet data convergence protocol) layer and so could help 802.11 address any ITU requirements for ‘macro-type’ operations, since these could be satisfied by the cellular partner network.
If IEEE 802.11 takes the IMT-2020 route, or decides to pursue its own path to ‘5G’, it is likely to tap into broader IEEE activities to an increasing extent. Myles pointed out that IEEE is already included in the ITU-R WP 5D’s list of external bodies engaged in 5G research – organizations with which it wants to connect. However, it does not include 802.11 specifically, nor the WiFi Alliance, on that list (though the WiMAX Forum is there, ironically, in a generally cellular-heavy list).
So it needs to take advantage of being part of a wider family. While the ITU may focus on macro networks, real world 5G will embrace all kinds of networks, including WLANs and WPANs – both areas where IEEE technologies rather than cellular ones have prevailed. So the IEEE’s 802.15.4 WPAN groups will be important partners for 802.11, not to mention the 802.3 Ethernet units. Then there is 802.1, which works to define overall architectures for interworking between local, metropolitan and wide area networks from the 802 initiatives.
Above all this, the IEEE has its own 5G initiative, which falls within its Standards Association (IEEE-SA). This, in turn, is part of the Global Standards Collaboration (GSC), which also includes the ITU plus bodies like ETSI, the TIA and ARIB (also 3GPP organizational or market partners). At the top level, this is no longer a world of segregated technologies and standards bodies, even if it still often looks that way when it comes to the market positioning of actual solutions. And 5G in its true sense will be far more than one or more RATs – it will also have to take account of the core network and other areas which the 3GPP does not touch, including IP standards (IETF) and the higher network layers, right up to applications.
So it is important that WiFi/802.11 pushes itself forward to be a key participant in this multi-faceted process. It will be in the interest of key vendors (Intel, Cisco, MediaTek and others which are potentially disruptive to the cellular status quo are prominent in the 11ax and 11ay committees), and Intel may even see the chance to impose its ethos on 5G as it failed to do with WiMAX in 4G. In the real 5G platform which emerges in the 2020s, the actual RAT will play a smaller role in the total architecture than ever before, but of course it is still a fundamental enabler. This may be the last generation in which the 3GPP holds all the best cards in defining that RAT – but hold them it does. In the interests of innovation and diversity, it is important that the IEEE moves to center stage too.