More Pressure on WiFi

As if we really needed more pressure put onto our public WiFi spectrum, both Verizon and AT&T are now launching Licensed Assisted Access (LAA) broadband for smartphones. This is the technology that allows cellular carriers to mix LTE spectrum with the unlicensed 5 GHz spectrum for providing cellular broadband. The LAA technology allows for the creation of ‘fatter’ data pipes by combining multiple frequencies, and the wider the data pipe the more data that makes it to the end-user customer.

When carriers combine frequencies using LAA they can theoretically create a data pipe as large as a gigabit while only using 20 MHz of licensed frequency. The extra bandwidth for this application comes mostly from the unlicensed 5 GHz band and is similar to the fastest speeds that we can experience at home using this same frequency with 802.11AC. However, such high-speed bandwidth is only useful for a short distance of perhaps 150 feet and the most practical use of LAA is to boost cellphone data signals for customers closest to a cell tower. That’s going to make LAA technology most beneficial in dense customer environments like busy downtown areas, stadiums, etc. LAA isn’t going to provide much benefit to rural cellphone towers or those along interstate highways.

Verizon recently did a demonstration of the LAA technology that achieved a data speed of 953 Mbps. They did this using three 5 GHz channels combined with one 20 megahertz channel of AWS spectrum. Verizon used a 4X4 MIMO (multiple input / multiple output) antenna array and 256 QAM modulation to achieve this speed. The industry has coined the new term of four-carrier aggregation for the technology since it combines 4 separate bands of bandwidth into one data pipe. A customer would need a specialized MIMO antenna to receive the signal and also would need to be close to the transmitter to receive this kind of speed.

Verizon is starting to update selected cell sites with the technology this month. AT&T has announced that they are going to start introducing LAA technology along with 4-way carrier aggregation by the end of this year. It’s important to note that there is a big difference between the Verizon test with 953 Mbps speeds and what customers will really achieve in the real world. There are numerous factors that will limit the benefits of the technology. First, there aren’t yet any handsets with the right antenna arrays and it’s going to take a while to introduce them. These antennas look like they will be big power eaters, meaning that handsets that try to use this bandwidth all of the time will have short battery lives. But there are more practical limitations. First is the distance limitation and many customers will be out of range of the strongest LAA signals. A cellular company is also not going to try to make this full data connection using all 4 channels to one customer for several reasons, the primary one being the availability of the 5 GHz frequency.

And that’s where the real rub comes in with this technology. The FCC approved the use of this new technology last year. They essentially gave the carriers access to the WiFi spectrum for free. The whole point of licensed spectrum is to provide data pipes for all of the many uses not made by licensed wireless carriers. WiFi is clearly the most successful achievement of the FCC over the last few decades and providing big data pipes for public use has spawned gigantic industries and it’s hard to find a house these days without a WiFi router.

The cellular carriers have paid billions of dollars for spectrum that only they can use. The rest of the public uses a few bands of ‘free’ spectrum, and uses it very effectively. To allow the cellular carriers to dip into the WiFi spectrum runs the risk of killing that spectrum for all of the other uses. The FCC supposedly is requiring that the cellular carriers not grab the 5 GHz spectrum when it’s already busy in use. But to anybody that understands how WiFi works that seems like an inadequate protection, because any of the use of this spectrum causes interference by definition.

In practical use if a user can see three or more WiFi networks they experience interference, meaning that more than one network is trying to use the same channel at the same time. It is the nature of this interference that causes the most problems with WiFi performance. When two signals are both trying to use the same channel, the WiFi standard causes all competing devices to go quiet for a short period of time, and then both restart and try to grab an open channel. If the two signals continue to interfere with each other, the delay time between restarts increases exponentially in a phenomenon called backoff. As there are more and more collisions between competing networks, the backoff increases and the performance of all devices trying to use the spectrum decays. In a network experiencing backoff the data is transmitted in short bursts between the times that the connection starts and stops from the interference.

And this means that when the cellular companies use the 5 GHz spectrum they will be interfering with the other users of that frequency. That’s what WiFi was designed to do and so the interference is unavoidable. This means other WiFi users in the immediate area around an LAA transmitter will experience more interference and it also means a degraded WiFi signal for the cellular users of the technology – and they reason they won’t get speeds even remotely close to Verizon’s demo speeds. But the spectrum is free for the cellular companies and they are going to use it, to the detriment of all of the other uses of the 5 GHz spectrum. With this decision the FCC might well have nullified the tremendous benefits that we’ve seen from the 5 GHz WiFi band.

The Future of WiFi

There are a lot of near-term improvements planned for WiFi. The IEEE 802.11 Working Group (part of the Wi-Fi Alliance) has a number of improvements being planned. Many, but not all of the improvements, look at the future of using the newly available millimeter wave spectrum.

It’s been twenty years since the first WiFi standard was approved. I remember how great it felt about fifteen years ago when Verizon gave me a WiFi modem as part of my new FiOS service. Up until then my computing had always been tied to cables and it was so freeing to use a laptop anywhere in the house (although that first generation WiFi didn’t do a great job of penetrating the plaster walls in my old house).

Here are some of the improvements being considered:

802.11ax. The goal of this next-gen WiFi is to enable speeds up to 10 Gbps using the 5 GHz band of free WiFi spectrum. The standard also seeks to provide more bandwidth in the 2.4 GHz band. The developing new standard is looking at the use of Orthogonal Frequency Division Multiple Access (OFDMA), multi-user MIMO and other technology improvements to squeeze more bandwidth out of the currently available WiFi frequency.

Interestingly, this standard only calls for an improvement of about 37% over today’s 802.11ac technology, but the various improvement in the way the spectrum is used will hopefully mean about a four times greater delivery of bandwidth.

Probably the biggest improvement with this standard is the ability to connect efficiently to a greater number of devices. At first this will make 802.11ax WiFi more useful in crowded environments like stadiums and other public places. But the real benefit is to make WiFi the go-to spectrum for use for the Internet of Things. There is a huge race going on between WiFi and cellular technologies to grab the majority of that exploding market. For now, for indoor uses WiFi has the lead and most IoT devices today are WiFi connected. But today’s WiFi networks can get bogged down when there are too many simultaneous requests for connections. We’ll have to wait to see if the changes to the standards improve WiFi enough to keep in ahead in the IoT race.

Of course, the 10 GHz speed is somewhat theoretical in it would provide all of the bandwidth to one device that was located close the transmitter – but the overall improvement in bandwidth promises to be dramatic. This new standard is expected to be finalized by 2019, but there will probably be new hardware that incorporates some of the planned upgrades by 2018.

802.11ay. 802.11ay is the successor to 802.11ad, which never got any market traction. These two standards utilize the 60 GHz spectrum and are intended to deliver big amounts of bandwidth for short distances, such as inside a room. This new standard promises to improve short-range bandwidth up to 20 Gbps, about a three times improvement over 802.11ad. The new standard might have the same market acceptance issues if most users are satisfied instead with 802.11ax. The primary improvements over 802.11ad are the addition of MIMO antennas with up to four simultaneous data streams.

802.11az. The earlier two improvements discussed above are aimed at improving bandwidth to WiFi users. The 802.11az standard instead looks at ways to improve the location and positioning of users on a WiFi network. Since many of the improvements in WiFi use MIMO (multiple input multiple output) antennas, system performance is improved significantly if the WiFi router can accurately and quickly keep track of the precise location of each user on the WiFi network. That’s a relatively simple task in a static environment of talking to fixed-location devices like a TV or appliances, but much harder to do with mobile devices like smartphones, tablets, etc. Improvements in locating technology allows a WiFi network to more quickly track and connect to a device without having to waste frequency resources to first find the device before each transmission.

The other big improvement promised by this standard is increased energy efficiency of the network. As the network becomes adroit at identifying and remembering the location of network devices, the standard allows for WiFi devices to shut down and go to sleep and drop off the network when not in use, saving energy for devices like IoT sensors. The WiFi hub and sensor devices can be ‘scheduled’ to connect at fixed times allowing for devices to save power by sleeping in between connections.

These changes are necessary to keep WiFi useful and relevant. The number of devices that are going to be connected to WiFi is expected to continue to grow at exponential rates, and today’s WiFi can bog down under heavy use, as anybody who tries to use WiFi in a business hotel understands. But a lot of the problems with today’s WiFi can be fixed with the combination of faster data throughput along with tweaks that reduce the problems caused by interference among devices trying to gain the attention of the hug modem. The various improvements planned by the IEEE Working Group are addressing all of these issues.

The Customer WiFi Experience

Every broadband provider is familiar with customer complaints about the quality of broadband connections. A lot of these complaints are due to poorly performing WiFi, but I think that a lot of ISPs are providing broadband connections that are inadequate for customer needs. Making customers happy means solving both of these issues.

It’s the rare customer these days that still only has a wired connection to a computer and almost the whole residential market has shifted to WiFi. As I have covered in a number of blogs, there are numerous reasons why WiFi is not the greatest distribution mechanism in many homes. I could probably write three of four pages of ways that WiFi can be a problem, but here are a few examples of WiFi issues:

  • Customers (and even some ISPs) don’t appreciate how quickly a WiFi signal loses strength with distance. And the losses are dramatically increased when the signal has to pass through walls or other impediments.
  • Many homes have barriers that can completely block WiFi. For instance, older homes with plaster walls that contain metal lathe can destroy a WiFi signal. Large heating ducts can kill the signal.
  • Most ISPs place the WiFi router at the most convenient place that is nearest to where their wire enters the home. Most homes would benefit greatly by instead placing the router somewhere near the center of the house (or whatever place makes the most sense with more complicated floor plans). Customers can make things worse by placing the WiFi router in a closet or cupboard (happens far too often).
  • There are a lot of devices today, like your cellphones, that are preset to specific WiFi channels. Too many devices trying to use the same channels can cause havoc even if there is enough overall WiFi bandwidth.
  • A WiFi network can experience the equivalent of a death spiral when multiple devices keep asking to connect at the same time. The WiFi standard causes the transmission to pause when receiving new requests for connection, and with enough devices this can cause frequent stops and starts of the signal which significantly reduces effective bandwidth. Homes are starting to have a lot of WiFi capable devices (and your neighbor’s devices just add to the problem).

A number of ISPs have begun to sell a managed WiFi product that can solve a lot of these WiFi woes. The product often begins by a wireless survey of the home to understand the delivery barriers and to understand the best placement of a router. Sometimes just putting a WiFi router in a better place can fix problems. But there are also new tools available to ISPs to allow the placement of multiple networked WiFi routers around the home, each acting as a fresh and separate hotspot. I live in an old home built in 1923 and I bought networked hotspots from Eero which solved all of my WiFi issues. And there is more help coming in the future, with the next generation of home WiFi routers offering dynamic routing between the 2.4 and 5 GHz WiFi spectrum to better make sure that devices are spread around the usable spectrum.

But managed WiFi alone will not fix all of the customer bandwidth issues. A surprising number of ISPs are not properly sizing bandwidth to meet customer’s needs. Just recently I met with a client who still has over half of their customers on connection speeds of 10 Mbps or slower, even though their network is capable of gigabit speeds. It is a rare home these days that will find 10 Mbps to always be adequate. One of my other clients uses a simple formula to determine the right amount of customer bandwidth. They allow for 4 Mbps download for every major connected device (smart TV, laptop, heavily used cellphone, gaming device, etc). And then they add another 25% to the needed speed to account for interference among devices and for the many smaller use WiFi devices we now have like smart thermostats or smart appliances. Even their formula sometimes underestimates the needed bandwidth. But one thing is obvious, which is that there are very few homes today that don’t need more than 10 Mbps under that kind of bandwidth allowance.

It’s easy to fault the big cable companies for having lousy customer service – because they largely do. But one thing they seem to have figured out is that giving customers faster speeds eliminates a lot of customer complaints. The big cable companies like Comcast, Charter and Cox have unilaterally increased customer data speeds over the past few years. These companies now have base products in most markets of at least 50 Mbps, and that has greatly improved customer performance. Even customers with a lousy WiFi configuration might be happy if a 50 Mbps connection provides enough bandwidth to push some bandwidth into the remote corners of a home.

So my advice to ISPs is to stop being stingy with speeds. An ISP that still keeps the majority of customers on slow data products is their own worst enemy. Slow speeds make it almost impossible to design an adequate WiFi network. And customers will resent the ISP who delivers poor performance. I know that many ISPs are worried that increasing speeds will mean a decrease in revenue – but I find many of those that think this way might be selling six or more speeds. I’ve been recommending to ISPs for years to follow the big cable companies and to set your base speed high enough to satisfy the average home. A few years ago I thought that base speed was at least 25 Mbps, but I’m growing convinced that it’s now more like 50 Mbps. It seems like the big cable companies got this one thing right – while many other ISPs have not.

The Death of WiFI Hotspots?

I’ve been thinking about the new unlimited data plans and wondering what impact they will have on public WiFi. As I wrote in a recent blog, none of the plans from the major cellular carriers are truly unlimited. But they have enough data available that somebody who isn’t trying to use one of these plans for a home landline connection will now have a lot more data available than ever before.

The plans from the big four carriers have soft monthly download caps of 22 Gigabytes or higher, at which point they throttle to slower speeds. But 22 to 30 GB is a huge cap for anybody that’s been living with caps under 5 GB or sharing family plans at 10 GB. And to go along with these bigger caps, the cellular companies are also now offering zero-rated video that customers can watch without touching the data caps. That combination is going to let cellphone users use a mountain of data during a month.

So I wonder how many people who buy these plans will bother to log onto WiFi in coffee shops, airports and hotels any longer? I know I probably will not. For the last few years I’ve seen articles almost weekly warning of the dangers of public WiFi and I’ve become wary of using WiFi in places like Starbucks. And WiFi in other public places has largely grown to be unusable. WiFi can be okay in business hotels in the early afternoon or at 3:00 in the morning, but is largely worthless in the prime time evening hours. And free airport WiFi in the bigger airports is generally already too slow to use.

If you think forward a few years you have to wonder how long it’s going to take before public WiFi wanes as a phenomenon? Huge numbers of restaurants, stores, doctor offices, etc. spend money today on broadband and on WiFi routers for their customers and you have to wonder why they would continue to do that if nobody is asking for it. And that’s going to mean a big decrease in sales of industrial grade WiFi routers and landline broadband connections. Many of these places already buy a second data connection for the public and those connections will probably be canceled in droves.

I wonder how much sense it makes for Comcast and others to keep pouring money into outdoor hotspots if people stop using them? You only have to go back a few years to remember when the concept of building the biggest outdoor hotspot network was the goal for some of the largest cable companies. Already today my wife has to turn off her WiFi when running in the neighborhood since her phone constantly drops her music stream through attempts to change to each Comcast WiFi connection she runs past. How many people with these unlimited plans will even bother to ever turn on their WiFi?

I also wonder if the cellular networks are really ready for this shift. There is a huge amount of data shifted today from cellphones to hotspots. As a business traveler I’m already thinking about how hard it might be soon to get a cellular data connection during the business hours if nobody is using the hotel WiFi. I know that 5G is going to fix this issue by offering many more connections per cell site, but we aren’t going to see widespread 5G cell sites for at least five years and probably a little longer.

I’ve always found it interesting how quickly changes seem to hit and sweep the cellular industry. There was virtually no talk a year ago about unlimited data plans. In fact, at that time both AT&T and Verizon were punishing those with legacy unlimited plans to try to drive them to some other plan. But the industry has finally plateaued on customer growth and cellular service is quickly becoming a commodity. I think a lot of us saw that coming, but I never suspected that the way it would manifest would be with competition of unlimited calling and the possible death of public WiFi. I don’t know if this industry will ever stop surprising us at times.

I guess a day could come soon when kids will have no memory of public hotspots. I can remember fondly when traveling to places like Puerto Rico or the Caribbean that the first thing you did on landing was find the locations of the Internet cafes. I remember back when our company decided to move out of our offices that one of my partners practically lived in a Starbucks for the next year. It was an interesting phase of our industry, but one whose days are probably now numbered.

A New WiFi Standard

Wi-FiThere is a new version of WiFi coming soon that ought to solve some of the major problems with using WiFi in the home and in busy environments. The new standard has been labeled as 802.11ax and should start shipping in new routers by the end of this year and start appearing in devices in early 2018.

It’s the expected time for a new standard since there has been a new one every four or five years. 802.11a hit the market in 1999, 802.11g in 2003, 802.11n in 2009 and 802.11ac in 2013.

One of the most interesting thing about this new standard is that it’s a hardware upgrade and not a real change in the standards. It will be backwards compatible with earlier versions of 802.11, but both the router and the end devices must be upgraded to use the new standard. This means that business travelers are going to get frustrated when visiting hotels without the new routers.

One improvement is that the new routers will treat the 2.4 GHz and 5 GHz spectrums as one big block of spectrum, making it more likely to find an open channel. Most of today’s routers make you pick one band or the other.

Another improvement in 801.11ax is that the routers will have more antennas in the array, making it possible to connect with more devices at the same time. It’s also going to use MIMO (multiple input, multiple output) antenna arrays, allowing it to identify individual users and to establish fixed links to them. A lot of the problems in current WiFi routers come when routers get overwhelmed with more requests for service than the number of antennas that are available.

In addition to more signal paths the biggest improvement will be that the new 801.22ax routers will be able to better handle simultaneous requests for use of a single channel. The existing 802.11 standards are designed to share spectrum and when a second request is made to use a busy channel, the first transmission is stopped while the router decides which stream to satisfy – and this keep repeating as the router bounces back and forth between the two users. This is not a problem when there are only a few requests for simultaneous use, but in a crowded environment the constant stopping and starting of the router results in a lot of the available spectrum going unused and in nobody receiving a sustained signal.

The new 802.11ax routers will use OFDMA (orthogonal frequency division multiplying) to allow multiple users to simultaneously use the same channel without the constant stopping and starting at each new request for service. A hotel with a 100 Mbps backbone might theoretically be able to allow 20 users to each receive a 5 Mbps stream from a single WiFi channel. No wireless system will be quite that efficient, but you get the idea. A router with 802.11ax can still get overwhelmed, but it takes a lot more users to get to that condition.

We’ll have to wait and see how that works in practice. Today, if you visit a busy business hotel where there might be dozens of devices trying to use the bandwidth, the constant stopping and starting of the WiFi signal usually results in a large percentage of the bandwidth not being given to any user – it’s lost during the on/off sequences. But the new standard will give everybody an equal share of the bandwidth until all of the bandwidth is used or until it runs out of transmitter antennas.

The new standard also allows for scheduling connections between the router and client devices. This means more efficient use of spectrum since the devices will be ready to burst data when scheduled. This will allow devices like cellphones to save battery power by ‘resting’ when not transmitting since they save on making unneeded requests for connection.

All these various changes also mean that the new routers will use only about one-third the energy of current routers. Because the router can establish fixed streams with a given user it can avoid the constant off/off sequences.

The most interesting downside to the new devices will be that their biggest benefits only kick in when most of the connected devices are using the new standard. This means that the benefits on public networks might not be noticeable for the first few years until a significant percentage of cellphones, tablets, and laptops have been upgraded to the new standard.

The Internet of Everywhere

tostitos-logoForget the Internet of Things. That is already passé and I saw something yesterday that made me realize we have now moved on to the Internet of Everywhere.

Tostitos has put out a special ‘party bag’ of chips for the SuperBowl. The bag contains a chip and a tiny sensor that can detect traces of alcohol on your breath when you breathe on it. If you test as intoxicated the bag will light up red and flash “Don’t Drink and Drive.” But that’s only the beginning. If you set off the red flash you can tap the bag against your smartphone and it will automatically call Uber and give you a $10 discount on your ride.

This is obviously a super-cool marketing idea and I expect the company to sell lots of bags of chips and will get a lot of positive press. And I would expect a lot of people will strive to make the bag flash red. But this demonstrates how cheap computer chips have become when a company can design a campaign using millions of chips in throwaway bags for a one-time promotion. This goes to show how amazingly small sensors have become that this bag can give you a mini-breathalyzer. I’m sure the test is not super-accurate, but the very fact that it can do this and still be affordable is amazing.

Engineers have been predicting this sort of technology for a few years. For example, there are now chips that can be printed onto human skin and can act as a keypad for your smartphone. We are not far away from having chips printed on every grocery item in the store, which will simplify checkout and will fully automate inventory control. With cheap chips we can literally sprinkle sensors throughout a farm field to cheaply monitor for the localized need for water, fertilizer or the presence of pests.

The real Internet of Things isn’t going to be unleashed until we can develop affordable swarms of sensors and also provide a way for them to communicate with each other. Today the IoT is being used mainly to monitor factory production and in homes for alarm monitoring and other similar functions. But the revolutionary value of IoT will come when it can grow to be the Internet of Everywhere.

Then we can have constant monitors inside our blood stream to sense for diseases and to fight them off early. We can monitor sensitive environmental areas to protect endangered wildlife. We can monitor our homes to a degree never done before – want to know if a mouse just snuck in – done!

There have been a lot of breakthroughs in creating small, low-power sensors. But the real challenge is still to find a way to communicate easily and reliably with a cloud of sensors. We are not going to be able to create a WiFi path with a thousand different home sensors but will need some sort of mesh technology that can first collect and make sense of what the sensors are telling us.

But I have no doubt that if a potato chip bag can tell me if I’ve had too much to drink and can then call for a ride to take me home, that we are making great progress.

And as I write this blog I’m sitting here thinking of if only I could show this bag to one of my long-passed grandparents. What would they ever make of this flashing chip bag, of my smartphone and of Uber? But then again, perhaps their biggest question might first be, “What is a Tostito?”

The Battle for IoT Connectivity

Amazon EchoThere is a major battle brewing for control of the connections that control the Internet of Things. Today in the early stage of home IoT most devices are being connected using WiFi. But there is going to be a huge push to have connection instead made through 5G cellular.

I saw an article this week where Qualcomm said that they were excited about 5G and that it would be a world-changing technology. The part of 5G that they are most excited about is the possibility of using 5G to connect IoT devises together. Qualcomm’s CEO Stephen Mollenkopf talked about 5G at the recent CES show and talked about a future where 5G is used for live-streaming virtual reality, autonomous cars and connected cities where street lamps are networked together.

Of course, Qualcomm and the cellular vendors are most interested in the potential for making money using 5G technology. Qualcomm wants to make the hundreds of millions of chips they envision in a 5G connected world. And Verizon and AT&T want to sell data connections to all of the 5G connected devices. It’s an interesting vision of the world. Some of that vision makes sense and 5G is the obvious way to connect outdoors for things like street lights.

But it’s not obvious to me at this early stage of IoT that either 5G or WiFi are the obvious winner of the battle for IoT connectivity in the home. There are pros and cons for each technology.

WiFi has an upper hand today because it’s already in almost every home. People are comfortable using WiFi because it doesn’t cost anything extra to connect an IoT device. But WiFi has some natural limitations that might make it a harder choice in the future if our homes get filled with IoT devices. As I’ve discussed in some recent blogs, the way that WiFi shares data can be a big problem when there is a lot of steady and continuous demand for the bandwidth. WiFi is probably a great choice for IoT devices that only occasionally need to make a connection or that need short-burst connections to share information.

But the WiFi standard doesn’t include quality of service and any prioritization of which connections are the most important. WiFi instead always does its best to share bandwidth, regardless of the number of devices that are asking to connect to it. When a WiFi router gets multiple demands it shuts down for a short period and then tries to reinitiate connections again. If too many devices are demanding connection, a WiFi system goes into a mode of continuously stopping and restarting and none of the connections get a satisfactory connection. Even if there is enough bandwidth in the network to handle most of the requests, too many simultaneous requests simply blows the brains out of WiFi. The consequence for this is that having a lot of small and inconsequential connections can ruin the important connections like video streaming or gaming.

But cellular data is also not an automatic answer. Certainly today there is no way to cope with IoT using 4G cellular networks. Each cell site has a limited number of connections. A great example of this is that I often talk to a buddy of mine in DC while he commutes, and he usually loses his cellular signal when crossing the between Maryland and Virginia. This is due to there not being enough cellular connections available in the limited area of the American Legion bridge. 5G will supposedly solve this problem and promises to expand the number of connections from a cell site by a factor of 50 times or so – meaning that there will be a lot more possible connections. But you still have to wonder if that will be sufficient in a world when every IoT device wants a connection. LG just announced that every appliance it sells will now come with an IoT connection, and I imagine this will soon be true of all appliances, toys and almost anything else you buy in the future that has any electronics.

Of a bigger concern to me is that 5G connections are not going to be free. With WiFi, once I’ve bought my home broadband connection I can add devices at will (until I overload my router). But I think Verizon and AT&T are excited about IoT because they want to charge a small monthly fee for every device you connect through them. It may not be a lot – perhaps a dollar per device per month – but the next thing you know every home will be sending then an additional $50 or more per month to keep IoT devices connected. It’s no wonder they are salivating at the possibility. And it’s no wonder that the big cable companies are talking about buying T-Mobile.

I’m also concerned from a security perspective of sending the data from all of my IoT devices to the same core routers at Verizon or AT&T. Since it’s likely that the recent privacy rules for broadband will be overturned or weakened, I am concerned about having one company know so much about me. If I use a WiFi network my feeds will still go out through my data ISP, but if I’m concerned about security I can encrypt my network and make it harder for them to know what I’m doing. That is going to be impossible to do with a cellular connection.

But one thing is for sure and this is going to be a huge battle. And it’s likely to be fought behind the scenes as the cellular companies try to make deals with device manufacturers to use 5G instead of WiFi. WiFi has the early lead today and it’s still going to be a while until there are functional 5G cellular networks. But once those are in place it’s going to be a war worth watching.

Why is my WiFi Slow?

Wi-FiOne of the universal complaints in the broadband world is that WiFi networks operate poorly. So today I thought I’d talk a bit about how WiFi functions. I think it’s probably different than what most people expect.

Most people know that there are two frequencies used for WiFi today – 2.4 GHz and 5 GHz. The 2.4 GHz band covers 80 megahertz of total bandwidth and is divided into 11 channels in the US. That may sound like a lot, but one 802.11 connection requires five consecutive channels. In practical terms this means that almost all WiFi gear in the US is preset to only offer channels 1, 6, and 11 and that means that only three non-overlapping transmissions can occur at the same time. The WiFi in Japan covers a wider spectrum footprint, up to channel 14, meaning they can use four non-overlapping signals simultaneously.

In practical use if you can see three or more WiFi networks you are experiencing interference, meaning that more than one network is trying to use the same channel at the same time. It is the nature of this interference that causes the most problems with WiFi performance. When two signals are both trying to use the same channel, the WiFi standard causes all competing devices to go quiet for a short period of time, and then both restart and try to grab an open channel. If the two signals continue to interfere with each other, the delay time between restarts increases exponentially in a phenomenon called backoff. As there are more and more collisions between competing networks, the backoff increases and the performance of all devices trying to use the spectrum decays. Your data is transmitted in short bursts each time you make a connection and before the restart cycle repeats.

If you’ve ever been in a hotel where you can see ten or more other WiFi signals, the reason for slow speeds is that there are huge conflicts between competing devices. People generally assume that the hotel has a poor Internet connection, but they could have a fast connection and the slo speeds are due to so many devices trying to connect simultaneously. Each WiFi device is rapidly turning on and off repeatedly trying to get open access to a channel. Your device will grab a channel for a short time and then get kicked off due to interference. Congestion has become so bad on the 2.4 GHz band that AT&T and Comcast no longer use 2.4 GHz for video or voice. Almost all smartphone makers no longer recommend using their smartphones at 2.4 GHz.

WiFi has improved dramatically with the introduction of the 5 GHz spectrum. In North America this spectrum swath has 24 non-overlapping channels. However, more than half of these channels are reserved for weather and military radar. However, this still provides a lot more potential paths to add to the three paths provided by the 2.4 GHz spectrum. Unfortunately the 5 GHz band shares the same WiFi characteristics as the 2.4 GHz spectrum and has the identical interference issues. But with more open channels there is still an increased chance of finding a free channel to use.

And interference between devices is not the only culprit of poor WiFi speeds. The network configuration can also contribute to poor performance. Some of the biggest sources of interference are range extenders or mesh networks that are used to try to get better signals. Range extenders listen to all WiFi transmissions and then retransmit them at a higher power level, and usually using a different channel. This creates even more WiFi signals in the intermediate environment competing for an open channel. When you can see your neighbor’s WiFi network, if they are using range extenders they might be always trying to use most of the available WiFi channels.

In a lot of the US we now also see a lot of public hotspots. For example, Comcast is in my neighborhood and I can walk and maintain a WiFi signal is most places from WiFi public signals that are transmitted from every Comcast home WiFi router. These public signals are always on, meaning that the WiFi router is using at least one channel at all times.

Probably the biggest new culprit for poor WiFi performance comes from our quest for greater speeds. The 802.11ac standard operates by merging together a lot of WiFi channels, and divides the whole WiFi spectrum into just two 160 MHz-wide channels. This means that only two devices using this 802.11ac can use up all of your home WiFi bandwidth. This standard was intended to be used to operate in short high-bandwidth bursts, but as people use this for gaming or watching 4K video the channels stay occupied all of the time.

Unfortunately the demands for WiFi are only increasing. The cellular carriers are still pestering the FCC to allow LTE-U, which would using WiFi to complete cellular calls. There are currently tests underway of the technology. We can also expect an increasing demand for WiFi from IoT devices. While most WiFi devices won’t use spectrum continuously, they still place demands on the channels and cause interference. There are also increasing use of devices that are always on, such as video surveillance cameras or smart home controllers like the Amazon Echo. A lot of experts look out five or ten years and expect WiFi to be unusable in a lot of places.

How We Use Cellphone Data

HTC-Incredible-S-SmartphoneNielsen recently took a look at how we use cellphone data. They installed apps on people’s phones that tracked data usage on both cellular networks and WiFi. The data comes from a massive study on the usage of 45,000 Android users in August. Nielsen also continues to study the usage of 30,000 cellular customers every month using the same app.

What Nielsen found wasn’t surprising in that they found that younger people use cellular data the most. They also found that Hispanics are the largest data users among various ethnic groups.

Here are the average monthly usage by age:

‘                      Cell Data         WiFi Data

18 – 24            3.2 GB            14.1 GB

25 – 34            3.6 GB            11.2 GB

35 – 44            2.9 GB              9.3 GB

45 – 54            2.1 GB              7.5 GB

55 – 64            1.4 GB              6.4 GB

65+                  0.9 GB              4.8 GB

This study quantifies a lot of things that we already knew about cellular usage. We know, for example, that younger people use their cellphones to watch video more than older people. I have anecdotal evidence of that by watching my 17-year old. If she’s representative of her age group then they are using cellular data even more than the 18-24 year olds. They communicate with pictures and videos where older generations use email, chat, and text messaging.

These numbers also show that most people are not yet using their cellphones as a substitute for landline data usage. Certainly there are many individuals for whom the cellphone is their only source for data, but these numbers show average cellphone data usage far below average landline usage. I have a number of clients that track landline customer data usage and most of them are reporting average monthly downloads somewhere between 100 GB and 150 GB per household. Comcast recently reported that their 6-month rolling median data usage is 75 GB – meaning half of their customers use less than that, and half use more. All of the numbers in the above charts, while representing individuals and not families, are still far below those numbers.

Nielsen also tracked data usage by ethnicity, as follows:

‘                                  Cell Data        WiFi Data

Hispanic                       3.8 GB          10.1 GB

Native American         3.5 GB            7.3 GB

African-American        3.3 GB            9.1 GB

Asian                            2.3 GB            9.9 GB

White                           2.2 GB             8.6 GB

This shows that Hispanics, on average, are the largest users of data, both cellular and WiFi. Whites are at the bottom of the average usage chart.

Nielsen also was able to look into usage by geography. They didn’t publish all of the results, but did provide some interesting statistics. For example, they have some strong evidence now that cities with widespread WiFi networks can save customers money on their cellphone plans. For example, New York City has a lot of public WiFi and users in the city use WiFi 14% more than the national average while using cellular data 12% less. Contrast this with a city like Los Angeles with little public WiFi, and citizens there use WiFi 9% less than the national average and use cellular data 13% more. This kind of study can provide the basis for a city to quantify the benefits to the public for building a public WiFi network.

Some Relief for WiFi?

Wi-FiThe FCC is currently considering a proposal by Globalstar to open up a fourth and private WiFi channel. It looks like the vote is going to be close with Commissioners Rosenworcel and Pai saying they oppose the idea.

Globalstar, based in Covington, Louisiana, is a provider of satellite-based telephone systems, but has been dwarfed in that part of the industry by the much larger Iridium. Globalstar was awarded a swath of spectrum in the high 2.4 GHz bandwidth to use for its satellite phones. The Globalstar bandwidth sits next to the part of the WiFi spectrum used for Bluetooth – but there is such a small amount of satellite phone usage that interference has never been an issue.

Globalstar made a proposal to make their spectrum available for WiFi, but with the twist that the want their slice of spectrum to be private and licensed by them. This differs from the rest of the WiFi spectrum that is free and open for anybody to use. Globalstar argues that allowing some large users, such as AT&T, to use their spectrum will take a lot of the pressure off of existing WiFi.

There are places today where WiFi interference is noticeable, and it is likely to get worse. Cisco projects that the amount of data carried by WiFi will triple in the next three years – a growth rate 50% greater than data usage overall. There is expected to be a lot of demand put onto WiFi from the Internet of Things. And the cellular companies have a proposal called LTE-U that would let them dip into the WiFi spectrum for cellular data.

But as might be imagined there is a lot of opposition to the Globalstar plan. One of the major objections is that this would be a private use of the spectrum while the rest of the WiFi is available to everybody. Globalstar could license this to a handful of companies and give them an advantage over other WiFi users by giving them access to a largely empty swath of spectrum that wouldn’t have many users. Having a few companies willing to pay the price for Globalstar’s spectrum flies against the whole concept of making WiFi available to everybody.

But the primary concern about the idea is that it will cause interference with existing WiFi. Today the normal WiFi antennas used to send and receive data are not very expensive, and they routinely broadcast signals outside of the range of the narrow WiFi channels. This creates a condition called adjacent channel interference where WiFi interferes with adjacent bands of spectrum. The FCC has handled this by creating buffers around each WiFi channel that allows for the bleed-over signals.

The Globalstar spectrum sits in one of those adjacent buffer zones and critics say that heavy use of the Globalstar spectrum would directly then interfere with existing WiFi that already bleeds into the Globalstar spectrum. In general it’s never been a good idea to place two heavily used slices of spectrum next to each other without buffers, and the proposal would jam Globalstar spectrum next to existing WiFi. On the other side of the Globalstar spectrum is the part of WiFi reserved for Bluetooth, and again use of the spectrum would eliminate any buffer.

The opponents to the idea have been very vocal. They don’t think the FCC should allow for the risk that Globalstar will create a clear channel for a few carriers while interfering with everybody else trying to use WiFI. The industry as a whole says this is an overall losing idea.

The issue has been in front of the FCC for a few years and looks like it will come to a vote soon. Chairman Wheeler is for the Globalstar plan with two other Commissioners already against it. It will be up to the final two commissioners to decide if this is a go or not.