Boffin breakthrough doubles Wi-Fi speed A team of researchers has developed a technology that has the potential to double the speed of wireless radios – such as those used by Wi-Fi – by allowing signals to be received and transmitted simultaneously. "Textbooks say you can't do it," said Stanford University assistant professor Philip Levis when announcing the …
It's of course a hallowed sport here to always know better than anyone else but that shouldn't be reason a reason to stamp on a willingness to learn.
I was recalling my physics lessons, imagining the boffins were doing something analogous to this:
http://en.wikipedia.org/wiki/Interference_%28wave_propagation%29
If the one tuned "dead zone" where the one local receiver sits is big enough for that local receiver, then it doesn't seem unlikely that there'll be dead zones elsewhere big enough for the one remote receiver. I imagine that would cause trouble. If not, why not?
This works with the direct signals from the Tx aerials, close to the wi-fi box, which are also the strongest signals. A few metres away something is reflecting the two signals, fillinf in some of the gaps in the pattern. Yes, that can of cola on my desk is necessary to provide a good signal on the other side of the office,
I used to be able to tell if a truck was coming down the road because of the effect on the digital TV signal. I got a better signal by pointing the aerial at the church tower.
Yes, probably, if my (and your) understanding of what they are doing is correct. But, since the zone for destructive interference is as large as the space between the antennae, all you would have to do is more the receiver that distance to get out of the "dead spot".
What is the ratio of up/down traffic for multiple pages? It matters not whether you have one page open, multiple pages or are downloading files, your http request takes sod all bandwidth and would have little impact on whatever you are downloading. Most wi-fi traffic is one way and full-duplex will have little real world advantage.
What would applications running in the backgorund have to do with anything? Unless you're running bittorrent, in which case you'd be rather eccentric not to be using a wired connection. Or possible running Adobe Photoshop which is constantly calling home (sorry couldn't resist the dig)
Consider a further thing, the speed of you ISP connection, let's be generous say you get a constant guarenteed 20Mb, the bottleneck will be there not with the wi-fi link.
Hotdesking? Sure you need full-duplex wi-fi to write documents and read a few emails </sarcasm>
Please give a real workd example were full-duplex wi-fi will be of benefit
Why not have a dedicated send and a dedicated receive antenna. You know what's being transmitted already without having to rely on it being re-received and nulled. Why not just subtract that from the incoming signal from the receive antenna, therefore cancelling the transmitted signal onboard at it's highest quality? Using 3 antenna and a null point restricts antenna orientation, the use of high gain antennas and I would imagine wouldn't work in a repeater situation.
I'd have thought the point here is that the noise-reduction doesn't have to be done by the hardware, but rather takes advantage of a natural phenomenon so that work is done for it. That has to be an economical approach, dunnit?
As for receiver positioning, the unit's own receiver has to be precisely positioned (at the, uh, nexus(?) of the two signals) to take advantage of the destructive interference, so surely that's true throughout the broadcast area: there's just a precise beam (or is it a plane?) in either direction perpendicular to the two transmit antennae where the same effect occurs (i.e. where the wavefronts hit each other). Much easier to avoid than to hit, I would think, but I took 2 years to get my Higher Physics, so I'm ready to be re-educated ;)
Well, I'm using a point charge as my "model" of the antennas. Due to the apparent fact that the two signal are 180 degrees out of phase, for distances much larger than the separation between the two antennas, there wouldn't be any electrical field to receive.
Yay! You can "transmit" and receive on the same frequency simultaneously.... except no one can receive what you "transmitted". Brilliant!!
Too lazy to grab my EM textbook to look at the actual mathematics.
"Due to the apparent fact that the two signals are 180 degrees out of phase..."
Don't think they are. They're IN-phase, so no loss of transmit, but depending on the positioning of the RX antenna, they're out of phase to THAT one. 'course, local reflections, a pair of dangly earrings, finger over the ant. etc. may nadger that. But, if we don't think phones for a while (after last Friday, hard not to!) but small basestations, talking to lappies, it may have merit. But, yeah, it's phones.
Personally, I'd have gone for polarisation diversity, which seems simpler (Tx. horizontal, RX vertical) or circular, Tx clock, Rx anti-clock.
But, hell, I'm an unemployed fartie. What do I know?
you only get some 3dB loss - unless you use (highly directional) helical antennae - which you wouldn't want to do as you need omni directional antennae - even then the cross polarisation loss is (IIRC) only some 9dB.
The null point interference pattern is hyperbolic between any two antennae - and this property is the basis of how DECCA; LORAN and OMEGA radio navigation systems to work (all based on OBOE used by the Germans in WWII)
The problem isn't a case of just subtracting the transmitting signal from the receiving antenna. The problem lies in the fact that the two antennas are so close and the transmitting signal is so powerful at that range, it swamps the receiver and completely overloads it, making the subtraction of any signal impossible.
There is nothing new about this. Analogue phone lines do it for voice and 1Gbps Ethernet on CAT5e does it on each pair.
I'd say vanishingly impossible on RF. You need over 90dB of isolation. On a Cable the characteristics are 1000s of times more stable than RF in a room and the Transmit and Receive signal levels not that hugely different. On a Wireless Set, the Transmit can be +18dBm and receive at -85dBm.
Mobile phone networks mostly use separate channels for Transmit and Receive. The main application would be WiFi and it's not going to work in the real world. You would be doing fantastic to get 18dB isolation, which would make for VERY short range WiFi.
They clearly note they are approaching this from an 802.15.4 network perspective, not 802.11 They have spelled out the fact that they have difficult in applications that have high transmit power (they tested at 0dBm), and large bandwidth (100 MHz and above is noted as being very difficult to handle). They're achieving significant reduction (on the order of 20dB to 30dB) from the antenna cancellation alone, and as much as 50dB when including a digital noise cancellation circuit.
They would have to improve the noise cancellation circuit, and have very precise antenna placement in order to have this technique work reliability at 802.11 power levels, yes, but they note that both are possible in the text.
Putting the third antenna between the first two, in a null spot, would mean that the signals on the first two sending antennas are 180 degrees out of phase. Otherwise they'd constructively interfere with each other.
This may then also cancel out the sending signal at various other locations. Conceivably this would mean that only near the transmitter would other devices be able to receive. Far away there would be no signal.
Whilst I do believe it is possible, I do not think this method will be viable. At least not from what I gather from the explanation in the article ( or my very crude mental model of their system).
I suppose the antenna is one of the better places to do the cancelling. However there are other places one could conceivably do it as well.
And it isn't that textbooks say you can't do it. They just highlight the fact that it is extremely difficult due to vast differences in signal magnitude, making the linear assumption engineers love to make invalid. If we lived in a perfectly linear world, this wouldn't be such a big problem.
The key point is is that this has to be done in analogue, as digital systems simply don't have the requisite range nor resolution to perform the subtraction. Though the matching will have to be damn near perfect since the signals can differ in magnitude by about 10^10. Telephone signals are pretty much evenly balanced, making the matching less critical.
Now they also need to put in a second receiving antenna. Otherwise the whole MIMO advantage is lost.
I started my working life as a ship's radio officer in the 1960s. All the gear was based on thermionic valves. Now most of the traffic was still WT and 'keying' used to close down the receiver to avoid overload. But we had these exotic things called RT that allowed telephony. And the gear allowed for simplex, half-duplex, & full duplex working. These folks are talking about full duplex.
The trick then was frequency domain separation. Any decent communications receiver had an IF strip with very steep 'sides' to the selectivity. A few tens of KHz separation on a 16MHz base worked just fine. People could talk & listen at the same time without messing about with pressels.
Now if they are trying to achieve full duplex on the same frequency, yes that is clever but I don't think even the proposed trick is novel. I think it was called 'RF bucking'
And it suggests that the people who did the original wifi band planning were twits.
And the idea of it doubling throughput is doubtful. Traffic is rarely symmetrical. Hardly likely to make the iPlayer stream 'Top Gear' any quicker is it?
(pressel == PTT)
yes similar background - I have been wondering what all the fuss is about; it's not as if frequency diversity is very hard; and a good comb filter if you want to go for spread spectrum is pretty easy to do now days. Just thinking ship board we had - 1k5Watt tx signal; aerial diversity at less than 0.1 of a wavelength (so effectively not separated...) and incoming signal on the rx aerial of a few micro watts can't be bothered to do the dB calculation but its 10^-6 to 10^3 (10^9) order of magnitude
For the null to work they are going to have to be single channel boxes - rigid aerials in epoxy ? or little aerial plug ins for each channel - admin nightmare...
I imagine with a reasonably accurate aerial alignment plus modern DSP chips this may work - but a good DSP and properly designed RF circuits would probably result in the aerial geometry only adding a couple of dB improvement - may be it is enough to make it just about viable if your RF designer works to standard IT specs (/joke)
Ah Robert - the days of de GKA qrj4 qry5 GKG up k (but your name suggest WCC or WSL not GKA ..) - 73's
My coats got the Handy Talky in the pocket
Sounds like a good idea, the simplest ideas are often the best ones, but i guess its still early days to be asking for specs and proof?
When they can show some data on real life improvements over similar equipment in exact same environments i'll be interested to see what this actually delivers. As someone said above, its not going to double up capacity just by becoming full duplex surely?
Im no RF engineer but surely the signal would still be affected by the usual factors that make WiFI in busy locations so tempremental?
Internet radio.
Of course, a complete misnomer. But I was born in the '50's. Then, there was no such thing in the general populace as "radio". It was wireless. Same as there was no such time as "25 past 5", or, God Forbid, "5-25". It was, as every old git knows, "Five-and-twenty past five. Pip emma.".
Plus, the Archers was always promulgated on the "Home Service", which is strange, implying London was "Home", but the Archers were in "Dork-etshire", or nearby.
So maybe "Wireless Radio" is an attempt to embrace the entire populace, young and old.
"one researcher told the team it couldn't work because the idea was so obvious that that someone else must have already tried it"
"the team – which has a provisional patent on the technology"
Hang on, I thought there were rules against obviousness in the patent system.......
...double check...
OK, I see what happened... they used the American patent system - geared more around making lawyers money than protecting inventiveness.
I was doing an experiment of exactly this setup last year. Only difference was that it was not for a communications system. Worked but did not take it iuofurther as it was much too easy to unbalanced the system. Something can be so obvious in theory but not done because it is obvious, nut actually doesn't work
No such thing as a 'provisional patent' only a provisional patent application which is just a placeholder while you decide whether to file a real application. For a provisional application, no-one looks at whether it is new or obvious, and it can't be used to stop someone else doing the same thing (unless they actually file a real application and get it granted).
Basically sounds good but is of pretty limited value other than getting a filing date.
The idea of 'nulling' at the central receive antenna sounds good in an ideal enviroment, but each transmitting antennae will generate reflections of varying intensity, so there will not be a perfect null. Nulling will help, and may be part of a further range of measures. But in a reflective environment I wonder how much attenuation will really be achieved ?
They call out in the paper real world nulls that achieve reductions on the order of 20-30 dB. Granted the testing occured in a fairly controlled office environment (a hall at the Uni), but they did note its a fairly active radio environment, (802.11, Bluetooth and so on). I was hoping they'd provide some information on actual background levels in the text, but they did not. At any rate, they did move the location of the access points around quite a bit (see the text for the map), which should have resulted in several fairly different reflective environments.
I am looking at this from a Wireless device (smart phone / laptop etc..) connected to an access point.
So, you have your new full duplex access point, but unless the device can also do the full duplex, then there is no benefit, right?
Other wise your send/receive WAP, will be receiving (while sending) while your device is only sending (because it can't receive at the same time)
I assume that you will get immediate benefit with base stations that need to communicate with multiple devices, so you will see the technology appear here first. That will improve speeds to all clients sharing the otherwise contended airspace.
Then, once you have that kind of base-station installed, you will see further benefit if you update your client hardware.
Dam I thought this was already inplace, oh well. Least when i looked at it I thought it's only good for 2 way communications and actualy by design increases security as makes eavesdropping a little harder due to the 2 channels bleeding over but each end knowing what to bleed out.
Of note to do this I also concluded that a form of calabration would be needed to accomodate ariel movement and reflections recieved by the areil.
But hey - next time I have a idea I thinks already invented I'll just oh wait I dont have a lab or any finance to make things happen so I'll just read about it a few year later on.
Anybody suggest a good place to submit great idea's? Say like a way for this to scale to more than 2 radios as I did work out a way of doing it but again thought its so simple sombody already done it :\
Anon as its the only way to travel
My experience with this work was that the swamping wasn't a problem with the antenna so much as leakage in the curcuitry. This was a big issue with putting multiple radios in one chassis, say to manage sectors. The problemst can be overcome with careful design but its non-trivial -- by comparason positioning the antennas is almost an afterthought.
Err, so we now have transmitters turning on for very long periods of time while running in FDX mode.
What about the other poor buggers that would like to use one of the 13 channels which are now continuously active because of continuous TX.
This is a recipe for 100% channel congestion. Just take a look at how many active WiFi signals there are in YOUR area. Now imagine trying to get a word in edgewise.
(A radio "ham").
I also come from a radio background, mainly working on telemetry, including a lot of antenna design. While I have no doubt you can get this to work in a lab, as for the real world forget it!!
As soon as your your own transmission is reflected by any nearby object, the reflected signal will swamp the null point you are using to receive in, Fail!!!
We did full duplex when I was involved... Albeit using frequency separation.
ZLB, STC DS12s (4ISB 5kW PEP) and DS13s (100kW PEP, anodes water cooled by direct contact!)
(And a few more "modern" bits of kit, but when it comes to transcontinental HF telephone circuits, nothing beats raw power.)
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
