Like the 3GPP’s 5G New Radio, the WiFi community’s first ‘5G’ standard, 802.11ax, is close enough to finalization that companies can design ‘pre-standard’ chips, with reasonable confidence that they will only take minor tweaks to comply with the eventual specs. Broadcom announced its first 11ax offerings last week, joining Qualcomm and Quantenna.
Although successive attempts have failed to bring the IEEE and its 802.11 standards, the basis of WiFi, directly into the 5G process, it is probable that 5G networks will rely on a combination of the two technology strands, to an even greater extent than current wireless services. And the WiFi roadmap has many aims in common with the 3GPP’s – increasingly so, as cellular standards focus more on higher frequency spectrum (including WiFi’s ‘own’ 5 GHz band) and on shorter-range connections.
So while an LTE or 5G cellular network in 3.5 GHz or 5 GHz is effectively a wireless LAN (albeit with wide area cellular connectivity via the MNO), WiFi can promise much of the density, low latency and data rates targeted by 5G NR. Some of that promise is enshrined in 802.11ax, which is particularly geared to locations with high device density, such as stadiums. It is designed to support peak speeds of 10Gbps using both 2.4 and 5 GHz spectrum (real world top speeds will be closer to 4.8 Gbps).
Enablers include Multiuser MIMO with four-stream downlink and uplink; simultaneous transmit/receive (STR); and OFDMA signalling, which allows it to move up to 1024 QAM modulation. The extra capacity, then, is achieved by counting in multiple bands; having both ends speak at the same time (STR); and supporting MU-MIMO on the uplink as well as the downlink. Overall, speeds should be four times those of the current top end spec, 802.11ac.
The first company to publicly launch an 802.11ax chip was Quantenna last October, and that product sampled in the first quarter of this year. Qualcomm Atheros followed in February and is sampling around now; Broadcom is only announcing its silicon now, while some Taiwanese, Chinese and Korean companies have done demonstrations of some of the technologies, though they have not actually announced chips yet.
Broadcom is calling its 11ax offering Max WiFi, claiming four times faster download, six times faster upload, four times better coverage, and seven times better battery life than 11ac. The chips include the BCM43684 for home routers, the 43694 for enterprise routers, and the 4375, a two-antenna design for smartphones.
Other improvements include the ability to switch off WiFi automatically when it is not being used and then wake it up again rapidly; and spatial reuse which lets access points share channel capacity and notify one another about empty channels. It also features zero-wait DFS (dynamic frequency selection). As non-WiFi technologies, like LTE-LAA and MulteFire, move into 5 GHz, more devices will need to rely on jumping out of WiFi channels which sense interference, to one where there is none.
The release was accompanied by a whole string of equipment vendors singing the praises of both Broadcom and the 802.11ax technology including Altice, Arris, Aerohive, ASUS, D-Link, Microsoft, Netgear, Sagemcom and Technicolor.
The chip provider talked up virtual reality as one of the data-hungry use cases which will require 11ax, and also predicts that the average family of four will have 50 connected devices by 2022 (the US is currently on about 12 devices per home). Though many of the additional gadgets will be for IoT applications, and will connect via ZigBee or Bluetooth, they can still add to the overall traffic demands when they link to a WiFi-enabled gateway or hub.
The 802.11ax effort has not escaped one of the voting plots beloved of IEEE attendees. An investigation, conducted in the latter part of 2016 following a complaint from a member of the 11ax taskgroup, found that 18 firms had created the DensiFi SIG (Special Interest Group) to push its preferred specs into the standard and illegally exclude others. In the past, these splinter groups have sometimes split a standards effort completely, as happened with the WiMedia development for UltraWideBand personal area networks. However, on this occasion, the members of DensiFi (Intel, LGE, Broadcom, Marvell, MediaTek, Qualcomm, Huawei, Orange, NTT, NTT Docomo, Samsung, ZTE, Apple, Cisco, Sony, Toshiba, Newracom and Quantenna) walked away rather than risk losing their votes and see smaller companies take control of a key standard.