802.eleventy what? Why Wi-Fi kind of sucks
arstechnica.co.uk
arstechnica.co.uk
- It needs to work on unlicensed spectrum, which means that it has to play well with all manner of devices that contend for this spectrum (e.g. other WiFi devices, Bluetooth, IEEE802.15.4). In practice this means that it cannot do much beyond CSMA/CA (i.e. the 'listen before talking' thing). CSMA/CA is a terrible contention mechanism for high density scenarios, and before long much of the air-time is taken by collisions. LTE does not have this problem, it works on licensed spectrum, as such, an LTE base station can just divide the time/spectrum blocks and allocate them to the various contending devices as it pleases (as it owns the spectrum), making almost optimal use of the spectrum that is available to it. 802.11ax will improve on this a bit (e.g. it will have OFDMA, which reduces the collision domain; it will allow the AP do to some coordination, via 'trigger' frames)
- Wifi has a lot of luggage; IEEE 801.11ax will be backward compatible with tens of billions of devices going all the way to IEEE 802.11b, which came out in 1999.
- Costumers don't like spending all that much money on Wifi. This cost-pressure means that we don't have as many people looking into WiFi as we should (people writing drivers; people debugging problems; radio engineers; investment in testing equipment).
- MIMO (introduced in 802.11n), downstream MU-MIMO (introduced in 802.11ac), and upstream MU-MIMO (to be introduced in 802.11ax) are all technically impressive, but also very hard to implement well. (But we are now starting to see the benefits of this, particularly the 802.11ac wave2 devices.)
Anyway, I have high hopes for WiFi, well beyond a billion WiFi chips are sold every year, and it is getting better all the time.
If you detect anything older than 802.11n, change channels or start knocking on doors.
IMO the entire 2.4ghz public spectrum block should be viewed as for legacy support. However far more spectrum should go to limited range use. (Also, building walls should have filter meshes that absorb frequencies not designed for use with mobile computer to cell towers / GPS).
You neighbour's 802.11b will never see that another Wi-fi network piggybacks on him and goes 100 times faster. There is even not the need to use the cumbersome CSMA/CA.
Indeed it is not an existing amendment, but if you are interested I would be happy to help someone to present a few ideas at IEEE 802.11.
What is a problem is that the preambule time is wasted, but it would not be too hard to imagine that even this preambule could be used for newer amendments after some homework.
How do you expect the "knocking on doors" part to go?
"Excuse me, something using an old wifi standard in your home is stomping on my speeds. Could you either turn it off or replace it?"
But if you did want to force someone to upgrade, you could always jam them... pretty easy to selectively jam a radio. Or crack their WEP (most 11b)... So many bad things.
-First I did my own tests of 802.11ac in 2014 and the manufacturers were correct in their claims at that time. You have to understand that the best speed is when you are in ideal radio conditions and simply you are never in ideal conditions and most of the time you are even far from the ideal case.
- Second, 802.11 sucks but not about raw speed, the MAC layer of most chip implementations is often ultra simplified and the outcome is that it is difficult to be authentified. This is strange as the Wi-Fi chip most often is a little computer and the MAC is implemented in software.
- Third, there are unreasonable economic expectations by users as well as the article's author: Wait you want gigabit speeds, ultra-reliability in challenging radio conditions, and that at a tenth of the cost of a 3G mobile radio?
- Fourth: Your phone has more hard time to cope with that throughput, than the Wi-Fi chip has. Android and Linux in general have many internal buffers because there are layers in charge of different features. The usable throughput is the raw radio throughput divided by the number of buffers. There are research OSes which use pointers instead of buffers, but Linux and Windows use buffers.
This is not true. Buffers can under some circumstances increase latency, or cause a TCP connection to take longer to get to its steady-state speed, but they don't hurt steady-state throughput unless something else is very wrong.
What I discussed was the way the buffers are in series inside the OS, this is known since a long time. A modern operating system throughput could be 5 to 9 faster than Linux: https://www.usenix.org/node/186147
Honestly the big reason mobile radios are more reliable is that they're allowed to transmit at like 20x the power of devices in the ISM band. That's not a cost thing, it's just a regulatory issue steming from (some rather outdated) concerns about interference in the shared environment.
I mean, it's true that wifi is sort of a mess. Marketing of the standards has gotten way ahead of real world hardware capability (frankly MIMO as it stands is basically voodoo snakeoil that provides no consumer benefit whatsoever). But it's no less complicated than all the junk going on in LTE either. Everything's a mess.
Wifi continues to have the extraordinary advantages of being privately deployable, performant, pervasive, and standard (i.e. ethernet-framed, so everything Just Works the way everyone expects). It's not going anywhere.
Curious why you say this. In the last few years, I've seen a significant increase in PHY data rate when using multiple spatial streams.
The snakeoil comment was about the MIMO functionality which I can only really test via my iPhone7 at the moment with the 2x2 MIMO client to the 4x4 MIMO AP (UAP-AC-HD). Once I get the MacBook Pro I'll have a 3x3 MIMO client.
I bought primarily because if this animation explaining the benefits of MIMO. https://unifi-hd.ubnt.com/
The marketing worked on me!
There's no magic bullet, and buying a bigger more expensive AP is, on its own, likely to make no real difference to any wifi benchmarks you might throw at it.
However, what you do get is the ability to easily place several APs in your house, know they will work together seamlessly, as you would expect from enterprise-y gear, and end with a more reliable and consistent wifi network.
(The article goes into detail why raw speed won't be gained by buying the bigger AP, so I won't repeat it here...)
MIMO works fine for me. The jump from 802.11g to 802.11n with two or more spatial streams was a significant bump in speed. And yes, I mean speed as in TCP, not PHY rate.
Yes and no. You can always go faster with more TX power, but if your neighbor can do the same then you get interference.
I can see 20 networks around my house. If everyone 20xs their power it would probably be thousands competing on a few narrow bands.
Third, there are unreasonable economic expectations by
users as well as the article's author: Wait you want
gigabit speeds, ultra-reliability in challenging radio
conditions, and that at a tenth of the cost of a 3G
mobile radio?
Imagine you're planning a trip to City X, you look online and four-star hotels with brands you've heard of quote $75 a night. If you turn up and it's not a four-star hotel, is it your fault because you had "unreasonable economic expectations"? Should you somehow have known that the going rate for a four-star hotel room was $500 a night, when half a dozen reputable suppliers were offering the same thing at a fraction of the cost? Seems to me it's not your fault - you've been a victim of fraud.
Of course many issues like that could be solved if APs shipped with their signal strength on low rather than blasted to the FCC limits. But I don't think we'll see that changed any time soon.
All the average person sees is the number of bars in their wifi connection UI, which AFAIK doesn't take contention into account, only signal strength.
I would be happy with any kind of reliability for a mobile chipset in reasonable radio conditions. Too often, I have to turn off WiFi to get basic connectivity happening. 4G/LTE connections are way better. It would be nice if Apple detected a bad WiFi connection and fell back to cell, but it seems they haven't figured out that trick yet.
As far as I can tell, this comes down to two problems: the most obvious technical challenge is the difficulty of detecting soft failures rather than hard failures. What they need to implement is a hard timeout which resets the connection state if the remote end fails to respond correctly within a set interval. I encounter this regularly commuting on the subway or taking underground tunnels between buildings; toggling airplane mode is the only way to get it to accept that the base station (WiFi or cellular) it was talking to 1500 feet back is never going to start responding.
The social problem appears to be that nobody considers this a keynote demo feature and so it hasn't advanced beyond iOS 1.0 despite years of bug reports.
I recently swapped out the piece-of-junk router from Comcast with a decent router, and it made a tremendous difference. Doesn't solve the issue when using other people's Wi-Fi networks though.
Apple introduced this very feature in iOS 9 https://support.apple.com/en-us/HT205296
I've never really had time or interest to investigate how well it's implemented, admittedly. It caused some consternation among those with crap data plans at the time, who where understandably upset when the iOS 9 update enabled the feature and lead to unexpected bills.
Does anything else do the same kind of mesh dynamic frequency allocation, but without requiring any kind of cloud service?
The rest of us live in a world with a couple of computers and a couple of phones and maybe a game console or TV connected and everything is fine.
"AD" i think is 60ghz, which doesn't penetrate walls too well in comparison.
Who knows, maybe we'll all use special wifi wall paint in the future, just to make the entire inside of the house or business an antenna.