Slashdot Mirror
Huawei Successfully Tests New 802.11ax WiFi Standard At 10.53Gbps
Mark.JUK (1222360) writes "Chinese ICT developer Huawei has confirmed that it was able to achieve a record transmission data rate of 10.53Gbps on 5GHz frequency bands in laboratory trials of their new 802.11ax WiFi (WLAN) wireless networking standard. The testing, which was conducted at Huawei's campus in Shenzhen, used a mix of MIMO-OFDA, intelligence spectrum allocation, interference coordination and hybrid access to achieve the result and the new technology could hit the market during 2018."
Better, faster ways to access inept content.
"A door is what a dog is perpetually on the wrong side of" - Ogden Nash
When if ever will 10gigE be affordable. Is the asic design on those switches really that insane?
laboratory setting missing real world issues with wifi that will slow it down.
So I'll be able to hit my monthly Comcast cap in 60 secs?
SUPER!!
Why?? Standards from the 50s supports multiple symbols per period?
http://en.wikipedia.org/wiki/Phase-shift_keying
Nothing limits you to one bit-per-second per baud. 9600 bps modems were, IIRC, 2400 baud with 4 bits per Hz. (Higher than that it got a bit shady because they started optimizing for being encoded in a digital phone line).
VDSL2 goes up to 32768-QAM, which is 15 bits per symbol. I do not know whether any actual phone lines exist with a sufficient signal-to-noise-ratio to make that coding useful.
Finally! A year of moderation! Ready for 2019?
Information Capacity Frequency Bandwidth * log2(1+signal-to-noise ratio) "In the lab" typically means "BNC cables" which give you very high signal-to-noise ratios. And in MIMO you can potentially treat each pair of antennas a separate channel. You use fancy techniques to increase the SNR for each channel. The nice thing about the 5GHz spectrum is that the frequency band is pretty large up there, and not as much interference with other unlicensed things (phones, microwaves, other wireless communication users... though as it gets more popular that will change) as the lower bands. The not-so-nice thing about the higher frequencies is that they tend to attenuate rapidly when the signal is going through something thicker than air.
-- Erich
Slashdot reader since 1997
For data transmission rates, you'll want Shannon's channel capacity, which is not contradicted:
(a) SNR is a factor of channel capacity
(b) It applies for a single channel. With MIMO you have multiple channels (not independent from each other, but with smart channel coding you get gains over SISO).
Doesn't matter how fast you do that, you won't sell it to gamers just through added latency, control (how the hell are you going to game via a tablet?) and screen-size. Plus who the hell wants to buy two home computers just so they can use one of them from a distance? Look at Steam Home Streaming if you want to do this - I assure you, it has a multitude of limitations even with beefy PC's at both ends.
The thin-client problem is one that solves only a handful of the problems people have with larger systems (not home installs) and has limitations that see everyone go through "thin-client / fat client / thin client" switches endlessly.
Fact is, if you have to have two machines to game on one, you're causing yourself problems.
As someone who's just ripped out a thin-client install in a school (where it was slow, un-updateable, had lots of limitations, etc. and where it's actually been cheaper all along just to put "real" machines into the rooms) I assure you that it's something we would all like the idea of until we tried to use it.
Not that I can't find a use case for faster Wifi. But, as you point out, 99% of people won't need it until it becomes almost obsolete.
Hey, if this is really that fast - I wonder if it could make mesh networking a viable alternative to the current (centralized) form of internet access? After all, why should all of those OLPC recipients be the only beneficiaries of mesh network technology?
Yes. And no. At least, there's no technical reason why not. 5 GHz is attenuated by most residential structural materials by only 1 dB more than 2.4 GHz and there are no microwave ovens and very few cordless phones to contend with in that spectrum. Range and throughput for non-line-of-sight is better than for 802.11a and 802.11b. People in fancy houses would probably want a roof-mounted antennae—red brick attenuates 5 GHz 10.1dB more than 2.4 GHz. Of course, if everybody had an antennae in their attic, everybody would benefit. And therein lies the rub.
The throughput is possible, the range is there, the compatibility with suburban realities is there, the mesh-compatible spanning tree algorithms are there, but the public will to buy a product that incorporates mesh networking is very nearly nonexistent. It spells doom for a product that depends on the network effect to have zero network effect.
The problem is connectivity to the rest of the internet.
Suppose we assume that a given neighborhood has nothing but ancient DSL1 available hardwired. Suppose we further assume that the majority of people in the neighborhood want something better. Suppose we get really generous and assume this device that enables mesh networking is affordable to a presumably somewhat lower income neighborhood (because the cable and phone companies only ignore low income neighborhoods). So all these neighbors buy the device and successfully cover the entire region. Congratulations, they can now talk... to each other.
Somebody, somewhere, has to connect their device to "the other networks," which we call the Internet, and it had better be a very high bandwidth connection because an entire neighborhood is going to funnel through it. 10 Gbit would be ideal. No individual can pay for that, so everybody in the neighborhood has to chip in every month and oh look, you just created an ISP. Or not, because nobody is going to actually take that last step to provide the required organization and get that connection established. Leastwise, not in most places.
Meanwhile the adjacent, probably more affluent neighborhood didn't even look up from their lattés—they already have acceptable hardwired connections and inaction is always easier than action and there's very little incentive for them to enable their own mesh. They already have tens of megabits and 10 gigabits shared out 100 ways is... what they already have, but they'd have to actually do something and once again inaction wins.
Device manufacturers have already followed this line of reasoning from beginning to end and won't even bother to take the first step of manufacturing and affordably pricing a device that can do off-the-shelf mesh networking.
This is why we can't have nice things.