The FCC announced a New Year's Wi-Fi gift during the International CES show earlier this month: a proposal to dramatically expand the unlicensed spectrum in the 5GHz frequency band for use by Wi-Fi devices. The announcement comes as a growing number of vendors are announcing products that will support the "Gigabit Wi-Fi" 802.11ac standard in 2013.
To find out the implications of the FCC's plan, we talked with Matthew Gast, director of product management for Aerohive Networks (responsible for the software powering Aerohive's controllerless access points, and author of "802.11n: A Survival Guide").
Gast blogged, enthusiastically and gratefully, after FCC Chairman Julius Genachowski announced the spectrum move, even admitting he had an "engineer-crush" on the chairman as a result.
[For more information on this band, and 11ac, check these links:
So, does Genachowski know you have an engineer crush on him?
I don't think so. What I didn't put in the blog is that I was [once] at the FCC supporting a technical board meeting. The chairman came in and spent two hours with us, an amazing amount of time. I introduced myself to him, and I'm sure he has no idea who I am. When I saw him, I thought "I hope I look that good when I get to his age." [Genachowski is 50.]
In your blogpost, you say the FCC is adding 195MHz of spectrum for Wi-Fi. For us non-spectrum-geeks, is that a lot?
The FCC is saying that it's 35% more spectrum. That sounds like a nice number. What I then did was made some educated guesses about the number of 20MHz channels this would add to the 5GHz band. And if you look at the channels, it's actually 60% more spectrum.
How do you figure?
The "existing" 5GHz channels - before the new FCC announcement -- were these: 22 20MHz-wide channels in three bands.
[Gast's technical breakdown is as follows:
5.17-5.33MHz: 36, 40, 44, 48, 52, 56, 60, 64
5.49-5.725MHz (these are the Dynamic Frequency Selection channels, which Wi-Fi radios must avoid if they're used by weather or other radars): 100, 104, 108, 112, 116, 132, 136, 140, 144
5.735-5.835MHz: 149, 153, 157, 161, 165]
In that second band, FCC removed three channels between 116 and 123 - 120, 124 and 128. This gap was for [protecting] a new wind shear radar [in effect, a special kind of weather radar] being developed by the Federal Aviation Administration for airports. The FCC concluded that outdoor Wi-Fi was interfering with this radar. That decision blew a hole in the middle of the available spectrum in this band.
The newest FCC announcement adds back the 3 channels in the DFS. It also adds new spectrum, so you have two [entirely] new bands. My best guess is that this new spectrum is good for 10 totally new channels. My guess at channel numbering is:
5.350-5.470MHz: 72, 76, 80, 84, 88, 92
5.850-5.925MHz: 170, 174, 178, 182
So, pre-announcement you had 22. Post announcement, you have 22+3+10=35, an increase of 62%. It's a much bigger improvement than just the [added] spectrum, and that's way more important to me as Wi-Fi engineer.
Why is that important?
First of all, we now have 60% more channels for high capacity networks. I once did a Wi-Fi network for a tradeshow, when everyone had 11b and 11g devices on the 2.4GHz band, in a football-field sized exhibition hall, with in effect just three channels. With Wi-Fi today, if any one person is transmitting on a channel, no one else can "say" anything. [So] it's basically impossible to lay out any kind of high capacity network with just three channels. That's why the 5GHz band is so important.
Secondly, you increase the speed available. The next generation of Wi-Fi, 802.11ac, uses wider channels than 11n [11n can combine two 20MHz channels into a "fatter" 40MHz channel; 11ac can combine 40MHz channels into 80MHz channels]. The table in my blog post shows the higher capacity of the wider 80 and 160MHz channels. With the existing spectrum, we don't have that many 80MHz channels. That lack of wider channels will take the big speed increase off the table for 11ac users. But with the new spectrum [and added channels], 11ac networking just got the ability to carry much larger amounts of data.
How else does this affect Wi-Fi networking?
There's another interesting but a bit subtle effect. In Wi-Fi, the big factor is the signal-to-noise ratio [SNR], which has to do with how well you [meaning "a radio"] can "talk" above the ambient electrical noise in your area. As SNR improves, you can perceive finer differences in the signal, making it easier to talk above that background noise. The more SNR improves, the faster you can go.
And that has another related effect, dealing with the higher, 256 QAM modulation [introduced] in 11ac. To do 256 QAM, you have to pick out very fine distinctions. It's like throwing darts at a dart board that has more, but finer concentric circles: the more accurately you're able to throw, and the less air disturbance between you and your dart board, the better it is.
The No. 1 cause of "noise" in Wi-Fi is other Wi-Fi devices: we "hear" each other transmitting. If you transmit, I can't. It's like being in a party that's in a small room, where you can hear this burble of noise from one or more groups talking around you. To be heard, you have to "talk over" that burble.
But with the new spectrum and more channels, it's like moving this party into a much larger room. When that happens, everyone is [in effect] spread out. So all the groups can talk more quietly, and you can recapture the subtlety in what you're talking about: you don't have to talk louder. And less interference means better speed.
So how will this change the Wi-Fi experience for end users?
Right now, these products exist only in laboratories. Imagine you have two 11ac products. One supports these new 5GHz Wi-Fi channels and one doesn't. For the radio that doesn't, it will have a choice of 20 channels, which also will be sought by 11a and 11n clients. It will auto-select a data rate and it will not be able to use 256 QAM.
But the radio that can use the new channels will have a better signal-to-noise ratio. It can pick a different data rate, and make use of 256 QAM. It will be 25% faster. In a hospital with big medical images or a classroom with streaming video, these will all load faster.
Actually, the new channels will also help other Wi-Fi clients even if they can't use them.
All kinds of networks just trundle along, with their load increasing, and they work until they collapse. You can see this in highway commuting at rush hour: if I leave home too late, the traffic load gets too much and I spend way too much time in second gear.
The new channels remove some of that load. If just you and I are sharing a Wi-Fi network and you run on 11ac and I'm not, I get faster data rates. You take less time to do what you do. And it reduces traffic congestion. It's like carpooling: even though I don't carpool, those who do carpool mean there are fewer cars on the highway when I'm driving. So the new spectrum improves the Wi-Fi experiences even for those with older devices.
So what happens next?
A press release is not an official commission action. Those FCC actions are based on rules, under what's called Part 15. There are rules you need to operate in this spectrum: you can use this much power, use a modulation like this; you can't interfere with weather radar, and so on.
What matters is the precise technical rules that will be put in place. The FCC issues a "notice of proposed rulemaking" or NPR, with a public comment period. This formal rule-making process by the FCC starts Feb. 20.
What will Wi-Fi vendors do? Can they modify existing products to use the new channels, or do they have to build new ones?
That depends on the nature of the rules. What Wi-Fi chip vendors realized years ago was they could build very flexible radio chips, in effect software-defined radios, and reprogram them to do different things. For a 5 GHz radio, I might [be able to] load software to support new 20MHz channels or other features.
If the new rules are simple enough, it's possible to just update the software on the chip and an access point radio, for example, can access the new channels. This depends on a whole bunch of things. For example there will be some kind of testing procedures for the chips, which in effect will be asking something like "can I use channel 72?" Now we already do this with DFS channels. So in the best case, there will be a very small change or even none at all. Worst case: completely new software to implement and pass these new tests.
Then systems vendors like Aerohive will have to take our products to an FCC-authorized testing lab, pass the prescribed tests and get an FCC number that indicates we passed.
What kind of time frame are we talking about?
I have no idea. The FCC is very, very careful. They're aware that a lot of their work has safety implications. The hard part is figuring out, in their engineering division, the tests that will have to be run. Unfortunately that is not an overnight process.
But it's worth waiting for, apparently?
Genachowski gets the idea of unlicensed spectrum. It's a band that anyone can use as long as you play nice. And you just let things arise from that basic principle. Today, for example, no one under 30 even knows what an Ethernet cable looks like. Wi-Fi depends on spectrum. And all these new interesting devices like the ones at CES need Wi-Fi because the wide-area networks can't take the load. Wi-Fi is going to be everywhere.
How are you using Wi-Fi?
I have a collection of iPads: an iPad 1, 2, and 3. I switch between an Android phone and for overseas travel a BlackBerry phone. I have a travel router [multi-device Wi-Fi connectivity] for the T-Mobile network. I have an iPad touch, but that works with my car's Bluetooth radio; a laptop, and a Nook ebook reader. And inside my house, we use Wi-Fi for audio-visual distribution. That's it for now.
John Cox covers wireless networking and mobile computing for Network World.