New generations come along about twice as quickly for WiFi than cellular and so we should not be too surprised to find the next version, WiFi 7, being presented even before the WiFi 6, or rather its 6E extension, has attained general commercial availability.
But this time round there is an extra urgency in the WiFi camp because of the sense that with 5G cellular has stolen a march and threatens to intrude into its territory, as well as take a lead in some fundamental performance parameters. We recall how in April, London-based mobile analytics firm Opensignal reported that 5G download speeds exceeded those of WiFi in seven out of eight leading countries, typically by a factor of three or more.
WiFi had led 4G in six of those eight countries and yet even in the USA, the one where it was still just ahead of 5G, that was explained by major operators such as AT&T and Verizon relying on slower low frequency bands to meet their 5G coverage targets.
While it is a mistake to represent 5G and WiFi as adversaries, there is no doubt such data has galvanized activity and accelerated work on the next WiFi 7, even though 6E itself was almost a new generation in that it gained access to a lot of new spectrum wrested from regulators in the 6 GHz area. We can also point to developments common to both WiFi and 5G as wireless technologies, notably Time Sensitive Networking (TSN), which will progressively be incorporated in both to help satisfy ultra-low latency use cases.
Talking of such use cases, it is worth pointing out that WiFi looks like it could largely miss out in two larger key sectors, automotive and the Industrial IoT (IIoT), where 5G combined with edge compute in various guises threatens to take hold. That combines the local performance and wide area network access required most efficiently.
However, that still leaves plenty to play for given the challenges facing 5G over indoor coverage, with WiFi indeed well placed to participate alongside cellular in heterogenous network combinations. That could include fixed wireless access (FWA), which is set for rapid growth in some developing markets especially.
WiFi 7 represents an effort to shut down any edge that 5G may have gained, especially for indoor communication, and also to be ready to leap on board with relevant emerging technologies as they arise, including the TSN. For that latter reason WiFi 7 as currently drafted leaves room for further additions under several clear headings.
Broadly, the standard has two halves, those components that build on 6E to maximize backwards compatibility and those that introduce more radically new capabilities where deemed essential for progress.
WiFi 6E about quadruples the amount of spectrum available by accessing the 6 GHz band at frequencies varying between countries, enabling a peak data rate of around 10Gbps. Apart from headline performance it also alleviates spectrum congestion in more densely populated areas where even the previous top 5 GHz band was now being consumed more heavily.
There is the caveat that range is reduced at the higher frequency, just as it is for 5G in the yet higher millimeter wave bands. But even that has some advantages for WiFi because it increases security, with the signals being more confined to the user’s premise and that lower range also mitigates the impact of congestion among neighboring hotspots in close physical proximity.
The recent research paper for WiFi 7, known under the old naming scheme as IEEE 802.11be, has targeted a further threefold increase in peak bit rate to 30Gbps. But at least as important is low latency, given all the noise about that from the 5G camp. That will be achieved through a combination of various elements underpinned by TSN, which will be integrated into the other components.
On that front, WiFi will be in contention with 5G and the objective is to shave latency down to similarly low values within set bounds for given instances to deliver overall delays that are predictable. That has been elusive not just for wireless networks but also wired ones carrying IP traffic on a best effort basis subject to delays that often vary significantly with traffic levels.
For this reason, TSN evolved initially to make Ethernet timing more predictable as an underlying datalink layer for IP transmission. While the technical details are more complex, the underlying principle is simple enough – to divide traffic into urgent and non-urgent with the former given priority. Then urgent data up to a certain level of traffic can have guaranteed latency within tightly specified bounds.
TSN was defined initially as IEEE 802.1Q to enable deterministic messaging over standard Ethernet networks. This meant it was a Layer 2 technology working at the level of data links transmitting frames between nodes of a network. This is a level below transmission of IP packets between routers of a network since those higher level point to point connections may span multiple lower level data links.
Ethernet is a shared medium, like higher level IP networks, with forwarding decisions made using Ethernet headers, not IP addresses. The advantage of this is that Ethernet can be deployed in almost any environment, so it made sense to build TSN around it. The aim was to give Ethernet for the first time the ability to specify in advance within tight bounds how long a frame would take to traverse a data link, using the standard Precision Time Protocol (PTP) to obtain a common sense of time for measurement of delay.
Two PTP profiles have been developed for TSN, IEEE 802.1AS and IEEE 802.1ASRev. The latter is a more advanced version incorporating fault tolerant mechanisms and shorter fail over by enabling synchronization between multiple so-called grandmaster clocks that receive accurate timing from external reference sources, such as GNSS.
Then the 802.1Qbv queuing protocol defines how to transmit the high priority TSN Ethernet frames within their schedule, within which less critical ‘non-TSN’ Ethernet frames are transmitted on a best effort basis as usual. The difference is that the best effort may be worse than usual because of the priority given to the urgent TSN frames. It is like how rail tracks can be shared between fast scheduled passenger trains and less urgent, best effort freight trains, which have to fill in the gaps and wait in sidings when necessary to avoid causing delays. At least that is how it is supposed to work.
In the case of WiFi, TSN is being bolted onto other dedicated mechanisms in a process that should be completed with version 7. The key mechanisms, which also increase capacity, are multi-link operation (MLO), access point coordination and MIMO enhancements. There are also some other improvements in the wings that improve capacity and potentially reduce latency through more efficient spectrum utilization, especially use of higher modulation orders and allocation of multiple resource units.
Modulation is governed by QAM technologies, which are employed in fixed cable TV and telco DSL networks as well as wireless. QAM combines two carriers or bit streams by modulating, that is varying, the amplitude of the signal wave. As QAM has advanced, more and more bits can be encoded by increasing the density of this process and currently WiFi 6 and 6E support 1024QAM, which can carry about 25% more raw data than the 256QAM of WiFi 5. WiFi 7 is expected to support 4096QAM, increasing capacity by another 20% (the data rate is proportional to QAM level expressed as a power of two).
As for multiple resource units, WiFi 6 supports use of OFDMA, which already allows several users to transmit data simultaneously by dividing WiFi channels into hundreds of smaller sub-channels, each operating at a different frequency. WiFi 7 will make this process more flexible by supporting allocation of multiple resource units, that is groups of OFMDA tones, for each device.
Then native support for multilink operation, one of the major enhancements, is designed to increase bit rates and reduce latency by combining the 2.4 GHz, 5 GHz and 6 GHz bands supported under WiFi 6E. It is true the latest WiFi equipment can already exploit several links simultaneously, but they are still independent of each other. Under WiFi 7 it will be possible to coordinate them for a single stream in a flexible manner to take advantage of varying traffic and demand, which will help reduce latency for high priority processes, as well as boost capacity.
AP coordination is already supported under the WiFi Alliance’s EasyMesh certification program for providers of WiFi products, with the first version 1.0 published in June 2018. This allows for optimization of the WiFi backhaul, as well as client steering by 802.11k/v standards that move traffic to whichever of the bands is least congested. Essentially it defines a multi-AP WiFi network comprising a single controller managing one or more agents.
There were already proprietary versions from vendors such as AirTies, Cisco, Amazon’s Eero and Netgear, in some cases offering more efficient AP coordination by avoiding need for a central controller and allowing all other nodes to participate in a mesh with greater all round capacity.
But the industry deemed AP coordination too important to be left to proprietary implementations and so acted first to standardize it at a basic level for multivendor operation. Now it is looking ahead to enable ever more sophisticated AP coordination under WiFi 7 with two levels.
The first level will be basic, or low level, AP coordination, which will improve cooperation between multiple APs compared with the current EasyMesh. It will to some extent merely enable functions currently supported by proprietary versions for multivendor operation, for example allowing APs to advertise their capabilities in beacons or management frames to reduce the number of collisions and allow greater spatial reuse to boost aggregate network capacity, which again would help bare down on latency as well.
That will be achieved via a feature known as coordinated spatial reuse (CSR), which has already been earmarked for the first Release 1 of WiFi 7, even though the standard is unlikely to be out before 2024. Under CSR, an AP can recruit other APs within the same WiFi hotspot, in practice usually two, three or four, to share in the simultaneous transmission of the data, with appropriate control over power output and link adaptation. This will create more opportunities to reuse the space and cut the number of collisions compared with the current spatial reuse schemes under 802.11ax.
There is, however, potential to exploit multi-AP coordination further and towards this end the WiFi 7 Task Group will support three new advanced schemes:
- Coordinated OFDMA, allowing neighboring APs to share their frequency resources with each other in multiples of 20 MHz channels.
- Joint single- and multi-user transmissions, where transmissions are coordinated in time to reduce impact of phase synchronization errors and timing offsets, providing enough backhaul capacity is available.
- Coordinated beamforming will be employed to exploit the growing capability of modern multi-antenna APs to multiplex their transmissions to associated clients or APs referred to collectively as stations (STAs). At the same time beams are directed so as to cancel out at STAs they are not transmitting to. This reduces backhauling significantly for joint transmissions and so increases overall capacity.
These advanced capabilities are still being defined and refined, with the final standard not likely to take full shape until 2023. It represents an attempt to get WiFi fully up to speed with not just contemporary but forthcoming and evolving wireless technologies.