A number of recent studies claim that Wi-Fi is on the verge of collapse – or radical slowdown – unless the FCC assigns the 7 GHz band to unlicensed use:

1. “The Near Future Requires Additional Unlicensed Spectrum” – CableLabs
2. “6GHz Wi-Fi: Powering the Future of Enterprise Connectivity” – IDC
3. “Wi-Fi Innovation and Future Spectrum Allocation” – ABI

These papers follow a common pattern, overstating the role of Wi-Fi in homes and offices, exaggerating the adoption of 6 GHz Wi-Fi devices, overstating the value of super-wide 320 MHz Wi-Fi channels, ignoring the value of competing demands for 7 GHz spectrum, and failing to consider the obvious fact that Wi-Fi can benefit more from mmWave spectrum than from mid-band spectrum. They rely on previous speculative work commissioned by the cable and Wi-Fi industries, “Assessing the Economic Value of Wi-Fi” (Telecom Advisory Services LLC) and “Spectrum Needs of Wi-Fi 7” (Intel.)

Remarkably, none of these analyses offers realistic application scenarios for multi-gigabit mobile devices. Astonishingly, the studies fail to realize that their demand for spectrum assignments meant to achieve parity between wireless and wired networks undermine their arguments for locking 5G wireless operators and LEO satellite constellations out of the 7 GHz band.

Most Internet Data is Generated by Desktop Computers

It’s widely understood that mobile devices offload traffic onto Wi-Fi networks inside homes and offices. Open Signal finds that 82 – 86% of the indoor mobile device data is offloaded from the Big Three cellular networks onto Wi-Fi, with seven times as much download as upload.

Cable Labs exaggerates and distorts this finding, claiming: The workhorse of connectivity, Wi-Fi carries more than 90 percent of all consumer internet traffic. This claim is grossly false because it ignores the fact that most consumer Internet traffic is consumed by desktop computers and stationary devices rather than smartphones. Mobile devices aren’t the biggest generators of network traffic, desktop computers, TV sets, and video streamers are.

Cloudflare Radar’s 2024 Year in Review report provides the necessary context. In developed countries such as the United States, desktops are responsible for 67.8% of Internet data compared to 32.2% for mobile devices.

The highest proportion of wireless to desktop Internet are found in the least-developed countries, such as Sudan, Cuba, Syria, and Malawi. These are countries that were probably wireless-only for many years, when the cellular network provided the only path to the Internet.

An Aside on Data Laundering

In the course of examining the references in the three papers, I noticed some amazing transformative work on network traffic estimation. ABI cites Assessing the Economic Value of Wi-Fi for estimates about in-home traffic growth: Indeed, one study has forecast that, between 2024 and 2027, total Wi-Fi traffic in the United States is set to increase by 83%.

Assessing derives its forecast by extrapolating from an already-extrapolated prediction about 2023 US wireless traffic made in Cisco’s 2018 Annual Internet Report. That’s right, Cisco predicted 2023 wireless traffic in a 2018 report, and Assessing’s author Raul Katz extrapolates from Cisco’s prediction all the way to 2027, even throwing in a double-fabricated CAGR.

Wi-Fi Forward passes off this magical number to regulators and policy makers as a perfectly sound basis for proactive assignment of spectrum to forestall an utterly imaginary condition of “spectrum exhaust.” Cable Labs echoes this prediction as well:

The new devices and applications will increasingly require higher speeds and lower latencies to work. A recent ABI Research report projects that current-standard (6 GHz-supported) Wi-Fi devices will grow from 95 million in 2024 to 367 million in 2029 — an increase of 288 percent in just five years — in North America alone.

My goodness, Cisco’s discontinued Annual Internet Reports were great sources, but not this great.

6 GHz Wi-Fi Adoption

Wi-Fi equipment manufacturers are less than enthusiastic about the 6 GHz band. Leading consumer market firms TP-Link and Asus offer dual-band Wi-Fi 7 routers with no support for 6 GHz at all; the semi-pro Wi-Fi 7 access point lines from Unifi and TP-Link Omada also offer dual-band options for both ceiling mount indoor and wall mount outdoor use.

More costly Wi-Fi routers from Asus and TP-Link Deco come in quad-band configurations that split the 6 GHz band into two sub-bands, where one sub-band is dedicated to wireless mesh backhaul. In quad-band configuration, the number of 320 MHz channels is reduced from three to two.

Wi-Fi 7 has to coexist with Wi-Fi 6E in the 6 GHz band. Currently, no Apple iPad or MacBook devices support Wi-Fi 7; it finally broke into the iPhone line with the model 16, however. Most current Apple products do support 6 GHz Wi-Fi 6E, however.

The principal barrier to tri-band Wi-Fi 7 is the cost/benefit tradeoff. Dual-band Wi-Fi 7 routers and ceiling mount access points retail for $100, while tri-band models are at least twice as costly. Effective use of the 6 GHz band also requires Automated Frequency Control (AFC), a feature currently only provided by top-of-the-line products such as the $500 Unifi E7 access point.

While Wi-Fi 7 has significant technical advantages over previous versions, they’re not enough to induce users to abandon their current devices for unnecessary upgrades. Wi-Fi 7 will ultimately be incorporated into smartphones, tablets, and media devices which will reach consumers in the normal replacement cycle. Wi-Fi 7 will not penetrate IoT devices, most of which remain stuck in the 2.4GHz band.

320 MHz Channels

Wi-Fi devices don’t always use the parameters we’ve told them to use. A Wi-Fi 7 device configured to use 320 MHz channels will only do so when it can find enough contiguous spectrum at one of its five preferred locations that is relatively free from interference.

Given that Wi-Fi 6E has been on the market twice as long as Wi-Fi 7, most networks will need to support both standards. When this is the case, the inventory of available wide channels shrinks. Presently, half of 6 GHz Wi-Fi devices use Wi-Fi 6 (with its 160 MHz channels) and half use Wi-Fi 7.

Wide channels don’t propagate as well as narrow ones because of power limits and the chance of encountering interference. Hence, Wi-Fi 7 broadcasters seem to prefer 160 MHz channels in most scenarios. This is actually fine, because well-designed Wi-Fi networks can provide excellent performance – meeting application needs – over the narrower channels.

Competing Demands for 7 GHz

Despite the lukewarm adoption of 6 GHz Wi-Fi products, advocacy groups such as Wi-Fi Forward are already advocating for an additional 875 MHz of unlicensed mid-band spectrum in the 7 GHz band. This band is also coveted by the military for high-resolution radar, by the cellular industry for Fixed Wireless Access and other applications, and by satellite constellations such as Starlink for residential broadband use.

The spectrum mid-band was once defined as 1 – 6 GHz, but is commonly seen as ranging from 3 to 8.5 GHz today. The mid-band is properly seen as providing the best combination of coverage and performance for the current batch of mobile devices using the current generation of signal processing chips.

The low band is best for coverage, while the high band (AKA “mmWave”) provides the best performance. Propagation is obviously more important for residential broadband networks than it is for indoor, 20–25-foot networks such as Wi-Fi and Bluetooth.

mmWave vs. Mid-band Spectrum

Wi-Fi wants to be as fast in its little domain as residential broadband, which means multi-gigabits per second. It will come closer to meeting this goal if it moves to the higher frequency mmWave band, from 9 to 25 GHz, where the interference environment is more compatible with extremely high channel widths.

This observation leads us to wonder whether prominent Wi-Fi advocates such as Cable Labs and its partners Spectrum for the Future and Wi-Fi Forward have Wi-Fi’s best interests at heart. Starlink and FWA have both been successful at attracting residential broadband customers, much to the dismay of the cable industry.

Cable is accustomed to monopoly market share in both linear TV and broadband, two services where market conditions have changed radically in just five years. Both cable TV and cable broadband have lost subscribers since 2023.

Residential Broadband

The fastest growing segment of the broadband market today is FWA, with 12 million subscribers at the end of 2024 and a projected 20 million by 2028. Despite the rapid growth, FWA performance is improving at 50% per year; T-Mobile’s median download speed is over 200 Mbps. For these numbers to continue past 2028, FWA will need more spectrum, preferably in the mid-band. And yes, Wi-Fi will need more spectrum some day.

The mmWave band is the best source for new Wi-Fi spectrum because its limited propagation permits reuse from home to home and apartment to apartment. mmWave Wi-Fi will require buy-in from chip manufacturers and Wi-Fi equipment vendors. They have concerns about the dual-edged sword of propagation (it must be good enough to penetrate interior walls but no exterior ones) which really come down to errors of placement.

Most Wi-Fi users today use all-in-one devices that combine network functions such as firewalls and routing in the same box as Wi-Fi, wide area modems, and Ethernet switches. These devices are rarely well-placed for wireless propagation.

Enterprise Wi-Fi follows a different patten, where Wi-Fi radios are separate from basic network functions. This allows a single ceiling-mounted access point to be placed in the center of the home where it’s relatively immune from neighborhood interference while retaining the ability to serve the entire floor. Prosumer Wi-Fi systems such as Unifi and Omada follow this efficient pattern. Sound installation practices moot the need for more spectrum.

Multi-gigabit Applications

As a way of justifying even more allocations of unlicensed spectrum, advocates point to future applications such as 8K TV, virtual reality, and AI. This group of applications has been touted for several years without much consumer acceptance.

This hasn’t prevented advocates from making wild claims about their bandwidth needs. Intel’s “Spectrum Needs of Wi-Fi 7” paper, cited by ABI as proof that three 320 MHz channels will not meet consumer demand in the near future (the opposite of what Intel claims,) indulges in gross exaggeration of application needs:

…video traffic will continue to be the dominant traffic in many Wi-Fi deployments. With the emergence of 4k and 8k video (uncompressed rate of 20 Gbps), we are experiencing these applications’ ever- increasing throughput and bandwidth requirements.

In fact, 8K video is neither emergent nor as bandwidth-hungry as the researchers claim. Uncompressed video doesn’t exist in networked environments and YouTube recommends coding High Dynamic Range, High Framerate 8K video uploads at 150 to 300 Mbps.

Given that 8K TVs are more common on the floors of consumer electronics shows than in American homes, 4K video at the recommended video bitrate of 12–20 Mbps will do fine for the foreseeable future. While no one knows whether AI will increase consumer demands for high bitrate transfers, current experience with LLMs embedded in search results, synthetic images, and synthetic video doesn’t suggest a need for higher resolution than HD or 4K.

Wired and Wireless Parity

 Without meaning to, advocates for more unlicensed spectrum make arguments that are applicable to network operator needs for more licensed spectrum. We have traditionally accepted asymmetries in network performance for different roles: uploads don’t need to be fast as downloads, and networks that serve groups of users need to be more capacious than any one user’s feed.

Wi-Fi and CBRS advocates (often the same people) now seem to be demanding parity for connections into and out of the home and those within the home. We’ve accepted that gigabit speeds to the home may be necessary to serve the needs of homes full of video-addicted teenagers. But we’re now being asked to believe that homes need multi-gigabit speeds to each individual user.

This can only be true if multiple super bandwidth-hungry wireless devices were in play at the same time within large numbers of homes. The only example we’re yet seen of such a device is the virtual reality headset. As ABI acknowledges:

The devices that will exert the heaviest strain on the new 6 GHz band will be emerging consumer devices like smart glasses VR HMDs, which, although not shipping in the same volume as the con­sumer devices above, will cause an outsized impact on home networks due to their high-performance demands.

ABI forecasts exponential growth for headsets, as many other prognosticators have done for three decades. It’s an exciting technology, but I’m no longer holding my breath for its wide adoption.

Parts of the Whole

Most of the data Wi-Fi touches comes from or to Internet service providers using a variety of technologies such as fiber optics, ground-based FWA, space networks such as Starlink, and rusty old cable modem. Even within the home, Wi-Fi is dependent on Ethernet as the proximate source and destination of all of its data. If ISP networks can’t keep up with ever-more powerful Wi-Fi, the capacity of Wi-Fi and all of the equipment and electricity that powers it goes to waste.

In fact, Wi-Fi is a supporting actor in the Internet movie. Like all supporting actors, its range is severely limited; 20 feet in most cases. Most data is consumed by desktop computers, wired video cameras, and stationary TV sets. Homes and offices connect to ISPs over wired and licensed wireless connections.

Wi-Fi’s role is small because unlicensed spectrum can’t be scaled to the distances needed to connect neighborhoods, let alone cities, states, nations, and continents. The genuine spectrum needs of in-home networks can easily be met without cannibalizing the services that connect homes to the Internet as a whole.

Wi-Fi’s current ask would create a 3000 MHz block of contiguous spectrum in the band most suited for wide area military and civilian use. Simply put, this demand is outrageous.