WiFi Evolves For The IoT

WiFi is everywhere, and it’s the most prevalent of the communication protocols that use unlicensed spectrum. But as a common protocol for the Internet of Things (IoT), it faces challenges both because of congestion and the amount of energy it consumes.

Two new approaches aim to address those concerns. One is to use multiple channels at once. The second involves the new 802.11ah HaLow standard. Both will go a long way toward extending a wireless communications technology that is virtually ubiquitous.

“There are 8 billion WiFi devices,” said Andrew Skafel, CEO of Edgewater Wireless. “No one ever envisioned WiFi would be that successful. That success has created a challenge as more devices try to connect through WiFi. If your performance declines, most of that decline is due to interference and contention.”

WiFi will play an important role as part of the new 5G ecosystem. “5G is too expensive for short distances,” said Rita Horner, senior staff product manager at Synopsys. “So you have Bluetooth from your computer or phone to your headset, WiFi beyond that, and then 5G beyond that.”

While WiFi can augment 5G, the opposite is also true if WiFi coverage or capacity lags. “5G millimeter wave in the United States uses the higher S-band, while in other parts of the world they use a lower frequency band for millimeter wave, where that can be used to replace or augment what you have on a cell phone and WiFi,” said Kurt Shuler, vice president of marketing at Arteris IP.

More efficient usage of available WiFi bandwidth can make for less congested WiFi traffic. Meanwhile, the introduction of devices using the new WiFi HaLow (IEEE 802.11ah) standard can transmit farther with less power. Together, they have the potential to serve the IoT market more effectively.

Exploding WiFi
When WiFi first became available, the IoT was not a consideration. “WiFi was the poor cousin to cellular,” said Skafel. The expectation was, “You’re going to connect your big clunky laptop to your modem in the basement.”

Since then, the number of devices looking to jump onto WiFi for communication has ballooned. “All of a sudden, WiFi has just increased and increased, and now it’s in every laptop and smartphone, and it’s the primary mechanism for the Internet of Things,” he said. The existence of other protocols using the same unlicensed industrial, scientific, and medical (ISM) bands may not compete with WiFi outright, but those protocols make reception less clean.

This results in two major challenges — interference and contention. Interference is simply the fact that multiple transmitters are using the same frequency range and degrading the signals, regardless of the nature of the signal. Within WiFi, interference is avoided by checking first to see if anyone is talking before transmitting. If someone is talking, there is contention, and the new transmitter must back off and try again later. The more devices that use the same frequency, the harder it becomes for them to gain access to the channel.

As the number of devices being connected has grown, WiFi has become increasingly crowded. Some of the connections use very little bandwidth, such as the IoT and smart-home devices. Others, like those streaming high-definition video, use a lot of bandwidth — and they’re very sensitive to latency. With too many devices competing for the same frequencies, performance degrades.

The structure of WiFi
Mainstream WiFi — which today would be 802.11n or 802.11ac — uses two main frequency bands. One is clustered around 2.4 GHz, and the other is clustered around 5 GHz. The expectation is that the 5 GHz range provides higher bandwidth, but the tradeoff is range and the ability to penetrate walls.

“The 5-GHz frequency goes half as far as 2.4,” said Skafel. “So propagation stinks. It doesn’t go through walls very well, so your performance falls off a cliff every time. You could put your hand in front of the access point, and you’d see a decline in performance.” Within those bands are channels, and actual transmission occurs on one of those channels for a given connection.

The availability of multiple channels means that multiple devices should be able to use WiFi without necessarily competing with each other. But, in fact, a given WiFi access point (AP) will use one channel. “In the 2.4 gigahertz band, you have 11 channels, maybe up to 13, depending on where you are,” he said. “And traditional WiFi has been able to use only one of those channels at a time.”

Which channel an access point will use is often set at the factory, and it never changes. Most access points just turn on and they go to a channel, although some will sniff to see which one is interference-free at that moment in time and then select that channel. But If an entire neighborhood happened to have fixed-channel access points from the same manufacturer, they would all be contending for the same channel, while all of the other channels would go unused.

“About the worst thing that can happen is that someone fires up another access point in that band,” he said. “And then what happens is both of those radios interfere with one another, and the performance degrades.”

One instinct might be to time-slice access to allow everyone on. But that’s not allowed, according to Skafel. “If you were to time slice, you actually break the WiFi protocol,” he said.

The next obvious answer would be for an access point to use more than one channel — even load-balancing traffic as conditions changed. While such dynamic adaptation may be a thing of the future, for now, just the ability to use more than one channel would be a major step in relieving congestion. But most of the channels overlap, limiting the number of channels that could be simultaneously active. In the 2.4-GHz band, only channels 0, 6, and 11 are non-overlapping.

Fig. 1: The channel structure of the 2.4-GHz WiFi band. Source: Wikimedia Commons By Michael Gauthier

With three active channels, it’s still possible to assign them in a way that avoids adjacent-channel interference in a collection of access points, depending on how they’re laid out. Theoretically, complete coverage is possible with no overlap. This is what Edgewater provides in a technology it refers to as spectrum slicing.

Fig. 2: Each cell uses a different channel, with adjacent cells using a different channel as shown here for the 2.4-GHz band. A similar situation is possible in the 5-GHz band. Source: Edgewater

Spectrum slicing is a simple idea, but implementation is not necessarily easy. The biggest issue for an access point becomes limiting interference from its own transmissions, which can crowd out the signals it is receiving. That’s because a more resource-efficient way to receive signals on multiple channels is to use a single wideband receiver — which will, by its very nature, pick up the very signals that are being transmitted.

“We’ve got to be able to cancel out our transmission,” said Skafel. “So if we’re transmitting on one channel, we’ve got to make sure that that transmission doesn’t swamp out the receive channel.” To help with this, the company uses separate transmit and receive antennas as well as a number of other techniques — many of them patented, he said.

With a wideband receiver (rather than one receiver per channel), one can receive all of the streams on all of the channels being used. That mixed signal becomes a mixed digital baseband RF signal, and the different streams are pulled apart by lowering the noise floor, filtering, and other techniques. The three streams (or however many channels are in play) are then independently processed up the protocol stack. While they use a single processor to do that, they have headroom to process up to 12 channels “simultaneously”.

The 5 GHz band is trickier. “[In the US], the 5 GHz band is divided into U-NII-1, -2 and -3,” said Skafel. “In U-NII-2, WiFi must avoid interference with weather-radar and military applications. As a result, before using any of the channels in U-NII-2, the AP must scan to detect any of these applications. If interference is found, then the AP must shut down for a specified period of time. Depending on the regulator, this could be 1 minute or 30 minutes. The net result is that, in many places, 80%+ of the time U-NII-2 is not usable. That is, 26 channels are not usable.”

There have been attempts to use more than one channel in the 5-GHz band. The higher end of the WiFi application processors now support two radios in the 5-GHz band. But they do require filters, and they operate at the extreme edge of that band. As a result, the majority of the 5-GHz spectrum is unused.

While reception uses one receiver, transmission must be more precise. Each channel gets its own narrowband transmitter. Because there is space between each pair of channels, it’s possible to transmit on adjacent channels without signals bleeding between channels.

Having multiple channels means that one can assign different traffic to different channels. “[I can] allow one of those channels for IoT device, so my door alarms are all on the same channel as the thermostat,” said Skafel. “They don’t need a lot of bandwidth, they don’t associate particularly often, and they’re slow. And then I maintain a channel for my kids to use for homeschooling, another channel for my wife to use when she’s at work because she works from home. I get one. And then we maintain channels for 4K video transmission. And that 4K video [is] the most latency-sensitive, aside from Fortnite and the games.”

The different channels may or may not be visible to the user, depending on the network operator (in the case of an AP “belonging” to an operator). Some like to keep a single SSID, which means that, at the user level, you can’t see the channels. Others may assign a different SSID to each channel.

In the latter case, it’s easy for a user to pick the desired channel. But if all of the channels are under a single SSID, then channel assignment becomes more challenging, and it’s something that your average user isn’t going to want to have to do. “I am technical,” said Skafel. “I’m definitely not the smartest technical person, but I did associate my alarm system. And it took a bit of work. But is your average Joe going to do that when they get an IoT device? No.” For that reason, he said that further efforts are underway to provide protocols to make assignment easy and seamless.

Cutting power
Part of what’s driving the need for WiFi capacity is the IoT, but that generally means smart-home or industrial applications, where the devices are not battery-powered. That leaves a set of applications that rely on batteries — often non-rechargeable batteries that must be replaced when drained. WiFi has been unable to satisfy the power needs of those applications, and they’ve had to turn to other lower-power protocols for their connections. But many of those protocols have low bandwidth. The new WiFi HaLow standard is intended to satisfy the part of the low-power IoT market that needs higher bandwidth.

Many battery-powered IoT devices transmit bursty, infrequent, small-payload packets. Bluetooth Low Energy (BLE) can do that over short distances; SIGFOX and LoRA can do them over very long distances. Even with HaLow, WiFi will not be able to compete on power in this type of application. But there are applications that require more bandwidth while still sipping energy judiciously, and HaLow can help there. While asleep or idle, HaLow consumes an order of magnitude less energy than traditional WiFi.

“802.11ah has a unique power-saving mode named TWT (target wake time) that allows the access point and [connected device] to negotiate when and how frequently devices will wake up to send or receive data,” said Vahid Manian, COO of Morse Micro. “This allows 802.11ah to effectively increase device sleep time and reduce spectrum contention.” He further noted that, “TWT does not exist in 802.11n radios”

Unlike the more common variants of WiFi, HaLow operates below 1 GHz, from 750 to 950 MHz. This lower frequency means that it can penetrate barriers more easily and travel farther — up to 10 times farther than conventional WiFi, depending on data rate. At 150 kbps, the signal can travel nearly a kilometer. At 80 Mbps, the reach is greatly reduced. “One access point can support 8,000 devices within a 1 km range,” said Manian.

Fig. 3: The different ranges provided by the different versions of WiFi. Source: Methods2Business

With lower-power protocols, because they have lower throughput, the radios have to stay on longer, potentially giving up some of that energy savings. If the power savings is less than the data-rate decline (for example, half the power but 1/10 the data rate), then the higher-powered radio will consume less energy simply because it finishes faster and goes back to sleep.

Fig. 4: HaLow has a higher data rate than other long-range protocols, while having somewhat less range. (Axes are logarithmic.) Source: Bryon Moyer/Semiconductor Engineering

WiFi equipment makers also tout the higher security provided by WiFi in general — and HaLow specifically. “WiFi does have a gold standard on security, [so] with WPA3, [HaLow] is far more secure than other [long-range or low-power] protocols such as Zigbee,” said Manian.

Fig 5: A typical WiFi HaLow system block diagram. Source: Morse Micro

A primary market for HaLow is the surveillance camera market. Line-powered cameras are likely to use standard WiFi, USB, or Ethernet (even receiving power over Ethernet). Battery-powered cameras, however, have a greater challenge. “Battery-operated cameras have a goal of 6-mo battery life,” said Manian. So HaLow becomes attractive for both consumer and industrial versions of such cameras. “Throughput is less than regular WiFi, but better than any other sub-GHz standard,” said Manian.

Other applications include industrial and automotive communications, access control (e.g., door locks) in hotels or other larger installations, and mobile phones. In the latter case, it’s not to replace standard WiFi, but rather, over the long term, to provide a peer-to-peer capability — essentially, a walkie-talkie function – even when there’s no cellular signal.

Some access points may be dedicated solely to HaLow when interconnecting devices that need to send information to each other. Others will integrate both HaLow and traditional WiFi into a single access point. “We believe future APs will adopt HaLow along with current WiFi,” said Manian. Design flexibility can be key here, and software can be a part of that. “People have been looking at the software approach, where you have most of the capabilities on the chip, but you can retarget potentially for different standards or eventually have a dual mode as well,” said Pierre-Xavier Thomas, group director for technical and strategic marketing for Tensilica at Cadence.

Like traditional WiFi, HaLow is subdivided into channels. “WiFi HaLow does have multiple channels: 1, 2, 4, and 8 MHz [wide],” said Manian. The exact number of channels depends on where one is. “The US allows from 902 to 928 MHz, and then you can fit 26 different 1-MHz channels, or you can fit 13 2-megahertz channels,” said Dejan Djumic, WiFi IP and system architect at Methods2Business. “In most of the key geographies, you can find non-overlapping pieces.”

One can send a signal farther using narrower channels. “For longer reach, the standard will use narrower channel bandwidth such as 1 MHz, and that naturally will have lower throughput than wider channel bandwidths such as 8 MHz,” said Manian.

At the least, this allows access points to select a less-busy channel. “We [provide] dynamic selection of a channel,” said Djumic. “[We] monitor multiple channels and then we can give [each one] a grade from 1 to 5. And if you have 5, that is the best channel where the interference is the lowest and you have the biggest capacity available, and then we can move the network traffic to that channel.” Morse Micro confirmed that they can also perform dynamic channel allocation.

With HaLow, you can also divide your attached devices into groups and then time-share the groups – something that would not be possible with traditional WiFi. “There is a mechanism built on top [that acts] like a restricted access window,” said Djumic. “An access point can say, ‘I have 1000 stations and I’m going to divide them into groups of 100 and then assign [the groups] over time.’ Then there are just 100 of them [competing] for [the channel], not thousands of them.”

At present, none of the HaLow manufacturers plans to use more than one channel. But Djumic sees it as a likely thing in the future. “If you take parking, for example, you have all these cameras that will be on one channel — for example, a 4-MHz channel,” he said. “Then you have your parking sensors all around to check which places are occupied. They, for example, come on a different channel that is 1-MHz [wide]. And then you have your payment system… [also] via a WiFi [channel].

WiFi HaLow should become available next year. “2021 is the year that WiFi HaLow will go and products will come,” said Marleen Boonen, CEO and founder of Methods2Business. It will also have a role within the upcoming 5G era.”

Added Skafel: “Part and parcel of this transition to 5G is that, in the background, the phone will be choosing the best-suited network technology, whether it’s WiFi 2.4 or 5 GHz, 6 GHz, whether it’s 1800 MHz, whatever. And the user won’t know.”