Wifi stations only interfere when they're transmitting at the same time, and other stations can hear it. Channels are a very simplistic way of understanding WiFi sharing, time and space are also important dimensions. If Station A and Client A want to communicate, they will not interfere with Station B and Client B if they pick different times to talk, or cannot "hear" each other at all.
There is some good work going on to improve this. Faster WiFi speeds mean that less time is needed to transmit the same number of bytes, meaning that stations are less likely to interfere. (This is why everyone hates your 802.11b-only client that can do a maximum of 11Mbps. It ties up the entire channel for no good reason.)
The default rate control built into most APs is awful, and makes things worse than they need to be. For example, let's say sending at 100Mbps, there's a 50% chance that the message won't be decodable, and sending at 1Mbps, there's a 99.999% chance that the message will get through. APs will try sending at the high rate, notice an error, and then go to the low rate. But that's wrong: it doesn't minimize the use of airtime. Sending 100Mb at 100Mbps twice is much more time-efficient than sending it at 1Mbps once. Some work is under way to improve this algorithm and get it in consumer routers.
There is also work to make APs transmit to stations at a transmit power tuned to be as low as possible. If a station is close, the AP only needs a small amount of power to send a decipherable message to the station. And, because the power is low, it interferes with less space. I believe this is now working under Linux with ath9k chipsets.
Anyway, "channels" is a terrible way to understand WiFi performance, because space and time are the other critical dimensions. Unfortunately, most "WiFi analyzers" only show which channels have APs, which is a meaningless metric.
I wonder how much we're at the limits of Forward Error Correction as well. It was really eye opening to see how much turbo code FEC could improve on total channel bandwidth without taking up any more spectrum.
Unfortunately they are harder to control. My apartment is surrounded by other apartments also transmitting on a polluted spectrum and I can't control when or where they transmit. I also can't control the age or transmitting speed of their devices. Simply opening up more spectrum is a net benefit to both me and them.
You can't actually control which frequencies your neighbors transmit on either. The best you can do is observe which frequencies are the most crowded, and avoid them.
But essentially this is what happens in the time dimension as well. Wi-Fi devices share a the channel across time using a technique called Carrier Sense Multiple Access (CSMA). When a device has something to send, it observes the channel to see if anybody else is transmitting, and if not, it waits a short random back-off time and then transmits. All devices are doing this, and since the back-off time is randomized, it's more or less fair.
Unfortunately that only works if the other APs are using the exact same channel, rather than an adjacent-but-overlapping one. It'd be great if consumer routers would only allow use of channel 1, 6, or 11, to avoid the usual situation where there is APs on every single channel interfering with each other rather than co-operating.
I have a neighbor on channel 5. I have thought of discussing it with them, but being across the street, the signals are so weak that I am not sure if interference is possible.
Could WiFi analyzers effectively determine information about space from a single location? Time is obviously possible, but isn't it inherently providing information about space by providing power readings from a single point?
Using the spatial dimension means directing a WiFi signal in a particular direction, or nulling interference in a particular direction. This can be done adaptively using multiple antennas on a single device.
There are mainly two techniques. The first, beamforming, involves adjusting the signal phase over each antenna such that it all adds up constructively in some direction. The second spatial technique, called spatial multiplexing, involves transmitting a different signal across each antenna, and hoping that each signal travels along a sufficiently different path (reflecting off walls, etc.) that all the signals can all be sorted out at the receiver.
I'm probably wrong but doesn't seeing traffic require monitor mode on whatever chipset? I've never looked into whether the the Nexus 4, 5 or 7 (2013) can do monitor mode but it'd be interesting to find out.
I never mentioned a WiFi analyzer that can see traffic. By "time is obviously possible" I meant "an obvious improvement that could be made based on your description"
This is an awesome and useful feature, but it isn't quite the same as monitor mode packet capturing. The FFT won't do much to help you figure out whether the interference is coming from a really high traffic volume or just a talkative 802.11b IoT PoS.
I think as we are continuing to balloon request sizes, stream more and higher quality video, increasingly populate urban cores, etc. finding more spectrum to pollute is an untenable long-term solution.
I think a more reasonable approach is to start putting wire in house-wrap, and perhaps putting a repeater antenna outside the Faraday cage for cell signals.
Actually, we sort of already solved it. Wifi was never meant to go insanely long distances, and now spamming APs inside of even a smaller home is a cheap and effective way of dealing with signal problems.
60ghz (Added by 802.11ad) can't penetrate sheetrock effectively, let alone bother your neighbors, a much smaller distance than 5ghz (which has half-solved the dense urban environment issue on it's own, but not completely; the article overplays issues significantly).
I support abandoning 2.4ghz for Wifi and anything else that is high bandwidth. The good news is, the 802.11 Working Group and the major Wifi impl manufs agree with me, which is why 802.11ac is 5ghz only, and 802.11ad is 60ghz first, both of them removing 2.4ghz for anything but 802.11 b/g/n legacy connectivity.
Also, as a side note, I want the FCC to ban the LTE-over-5ghz plan. 5ghz is ours; they can have 2.4ghz, we're done with it.
Except when you live in a 200 year old house like I do with 1 foot thick stone walls. I don't have to worry about "polluting" my neighbors because my nearest neighbor is a mile away. Any "standards" ought to not just assume everyone lives in dense, multi-family residences.
Yes, but it's called designing for the most damaging worst case. Both ultra-dense urban and houses like yours are corner cases, but on opposite ends of the spectrum.
For you, I'd suggest looking into Unifi AC APs (the lowest end ones are $99 each, the Pros are extremely nice), and then put them everywhere you want signal.
A ban on LTE over 5 GHz seems a bit heavy-handed and prone to allowing further abuses (as some dude on IRC said about something else, it's like trying to kill a mosquito with a chainsaw: you end up all bloody and probably don't even get the mosquito).
I think the real objectionable part is the requirement that the 5 GHz band be restricted to "licensed-assisted operation", only as a secondary downlink for a system based on a licensed band. A rule that any system using unlicensed spectrum shall not depend on licensed spectrum would block the existing LTE-on-5GHz plan and either get the whole idea scrapped or get mobiles that can work entirely on unlicensed spectrum into the market, opening up the possibility of private LTE networks on unlicensed spectrum.
Given how important centralized choke points seem to be to the 3GPP architecture, I tend to suspect it would just chase licensed users completely off the unlicensed spectrum.
>Also, as a side note, I want the FCC to ban the LTE-over-5ghz plan. 5ghz is ours
agreed, but i think simply banning LTE-over-5Ghz is not going far enough. We need a segment of spectrum reserved for end-user network communication - wifi is so important, shoving it into unlicensed spectrum seems like a bad idea. regulation should codify it as "ours". The cost and ease of use of ubiquiti bridges that operate on unlicensed wifi spectrum is quickly becoming a problem as well, as all our devices have to start competing with the private bridge networks.
Unfortunately, it can't penetrate human bodies or furniture effectively either. It's effectively requires a line of sight to have a good connection. If the line of sight is blocked, the best you can hope for is a one-bounce reflection at a significantly lower signal-to-noise ratio.
Where will you place the 60 GHz access point and where will you place the 60 GHz antenna on the device such that you'll consistently have LOS coverage?
Ceiling. Unless you have a pendant light, residential ceilings are usually entirely visible from all points (well, unless you're under the coffee table...)
2.4GHz works fine in the suburbs. The only downside is that it stops working when someone turns on a microwave oven. If the telecommunications companies listen to your suggestion, they are likely to use both 2.4GHz and 5GHz.
I know you meant that as a joke, but when has "free Wifi" ever resulted in something better 1-2 mbit? I live in an extremely bad cell phone signal area, and not a single "free Wifi" area around here is usable, and I'd still rather use my cell phone signal for data given how bad the "free Wifi" is.
There are plenty of free hotspots in my city that are 20+ Mbps. It probably has more to do with the cost of Internet access in your area than the Wi-Fi connections themselves.
The free wifi at Telstra stores is pretty fast, often over 100Mbps. But then you're standing awkwardly outside a Telstra store with a laptop looking suspicious.
Municipal broadband (wifi) is inevitable. I'd much rather pay the landlord or utility or someone to just handle it. Much as I hate Comcast, I'm loving that XfinityWifi is ever more common.
What does municipal broadband have to do with that connection being worthless? If anything, that would just cause even more pollution.
We need to stop considering Wi-Fi the default mode of connectivity, and go back to it being a stopgap for devices where it's unpractical to have a wire.
Perhaps it's time to start building powerline connectivity into phone and laptop chargers?
As long as cell networks implement data caps, maybe. But I don't see why municipal wifi is inherently inevitable (or even desirable).
New York is starting to roll out combination charging/wifi stations across the city, and I think they're horrible for a whole number of reasons. But the biggest thing is that public wifi isn't really any better in terms of bandwidth or latency than mobile networks to begin with - the only real advantage they have is that they bypass data caps[0].
[0] Oh, and they make it trivial for the city to track your whereabouts at all times.
For those who are unaware, these are the channels that caused the FCC to effectively ban OSS firmware in wifi routers. As far as I know, Ubiquiti and Mikrotik could be configured to use them unconditionally, which caused problems with radar.
There is some good work going on to improve this. Faster WiFi speeds mean that less time is needed to transmit the same number of bytes, meaning that stations are less likely to interfere. (This is why everyone hates your 802.11b-only client that can do a maximum of 11Mbps. It ties up the entire channel for no good reason.)
The default rate control built into most APs is awful, and makes things worse than they need to be. For example, let's say sending at 100Mbps, there's a 50% chance that the message won't be decodable, and sending at 1Mbps, there's a 99.999% chance that the message will get through. APs will try sending at the high rate, notice an error, and then go to the low rate. But that's wrong: it doesn't minimize the use of airtime. Sending 100Mb at 100Mbps twice is much more time-efficient than sending it at 1Mbps once. Some work is under way to improve this algorithm and get it in consumer routers.
There is also work to make APs transmit to stations at a transmit power tuned to be as low as possible. If a station is close, the AP only needs a small amount of power to send a decipherable message to the station. And, because the power is low, it interferes with less space. I believe this is now working under Linux with ath9k chipsets.
Anyway, "channels" is a terrible way to understand WiFi performance, because space and time are the other critical dimensions. Unfortunately, most "WiFi analyzers" only show which channels have APs, which is a meaningless metric.
Here's an interesting talk about how this all works in the real world: http://apenwarr.ca/diary/wifi-data-apenwarr-201602.pdf