Dev.SN ♥ developers
Sections
Dev.SN
AnonTechie writes:
According to an article from The Register, a team from Stanford University has patented technology that could halve the bandwidth that a mobile provider needs.
Operating under the name Kumu Networks, they are showcasing tech which they claim would exactly double throughput. Radio equipment (such as mobile phones) would be able to send and receive on the same frequency through a process similar to noise-cancelling headphones; by knowing what a base station is transmitting it can cancel out the information from the very faint signal it receives.
This discussion has been archived.
No new comments can be posted.
Radio Tech Could Halve Bandwidth Needs
|
Log In/Create an Account
| Top
| 1000 moderator points
| 36 comments
| Search Discussion
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
I'm not a robot like you. I don't like having disks crammed into me... unless they're Oreos, and then only in the mouth.
-- Fry
(Score: 5, Insightful) by FatPhil on Thursday February 27 2014, @05:38PM
Making a public pledge to no longer contribute to slashdot
(Score: 5, Interesting) by FatPhil on Thursday February 27 2014, @05:47PM
By transmitting at the same frequency, they increase the noise, and so reduce the SNR, and so Shannon says they reduce the maximum available bandwidth.
They're not approaching the Shannon limit - they're pulling the Shannon limit closer to where they are!
This smacks too much of the compression snakeoil that those with a background in information theory get so annoyed by. If they know what the transmitter is sending, so that they can subtract it from what they're receiving, then it must already have been transmitted to them, which used bandwidth that they're not tallying. Like an OTP, it seems like they're transmitting bandwidth through time.
Making a public pledge to no longer contribute to slashdot
(Score: 2, Insightful) by Desler on Thursday February 27 2014, @05:49PM
And even if it was real why would any carrier care? They make gobs of profits off of scarcity not a glut of available bandwidth.
(Score: 4, Interesting) by bd on Thursday February 27 2014, @06:02PM
The summary is written from the point of view of the base station.
The base station transmits a strong signal and wants to receive a very weak signal on the same frequency. But it obviously knows exactly what it is sending right now, so it can subtract that from the signal it receives.
That is how I understand the summary. I am not an RF expert, but that is obviously something you would want to do in radar applications. It is astonishing that nobody would have done something like that up until now. Maybe the article is not very accurate?
(Score: 4, Informative) by adolf on Thursday February 27 2014, @07:31PM
The summary is misleading, at best. Your description actually makes sense (I'm no expert in RF, either, but I do butter my bread with it), and it would apply equally to field radios (ie: phones).
In all cases, a transmitter knows exactly what it it transmitting. A receiver co-located with that transmitter (perhaps even integrated within the same IC) can therefore subtract the transmitter's signal, leaving only other signals (ie: a far-away transmitter).
Simple -- at least on the face of it. It could even be analog and self-adjusting without drama (just rotate phase and amplitude of the receive antenna until the amplitude of the mixed RX-TX signal is at minimum).
I can see complications, though: Reflections of the local transmitter from nearby objects or buildings will also be received, and must be mitigated somehow because they will often dwarf the signal received from the remote transmitter. The reflections will not be anywhere near linear in nature, and may continuously and unpredictably vary in amplitude and delay and other characteristics. Echo cancellation? Maaaaaybe.
I'm wasting my days as I've wasted my nights and I've wasted my youth
(Score: 3, Interesting) by tftp on Thursday February 27 2014, @10:01PM
It will be very computationally intensive to recognize all the reflections - there will be more than one, all superimposed.
But that's at least is theoretically possible. What is less possible is to subtract your own signal (at +33 dBm) from an incoming signal that is at, say, -100 dBm. You need at least 130 dB of dynamic range. Traditional RF systems use analog components (circulators, filters, separate antennas) to suppress unwanted signals and to bring them into the range that the receiver can handle. A directional coupler is one such possibility... but it is not going to be good enough because the TX signal reflects from the antenna, and sometimes those are very strong reflections (if you are holding it wrong.)
The easiest way to implement full duplex today is by packetizing data that comes in and out, and use only a fraction of time to receive and transmit those packets. Then you don't have to receive while you are transmitting.
(Score: 1) by adolf on Thursday February 27 2014, @10:50PM
Eh? That's the easy part:
If A is our locally-transmitted and very strong signal, and B is a combination of whatever is coming into the RX antenna, then:
Transmit signal A. Invert it (rotate phase 180 degrees...give or take), attenuate it as appropriate, and mix it with the input to the receiver (B). We A minus B equals C, which is everything but the local transmitter.
Done.
Meanwhile, TDMA isn't full-duplex. It's just two or more transmitters taking turns. Normally, they take turns very quickly, but they're still not talking at the same time. Meanwhile, the spectral efficiency of TDMA is approximately the same as simplex: Nothing is gained.
I'm wasting my days as I've wasted my nights and I've wasted my youth
(Score: 2, Informative) by tftp on Thursday February 27 2014, @11:16PM
It's easy only until you look closer. Then you discover that the antenna mismatch results in a frequency-dependent and environment-dependent phase shift. Also you notice that the electrical length between all components becomes critical - and that is temperature-dependent. Even a "simple" 180 degree phase shift has to be done with a transformer or a balun that is not ideal. However a tiny disbalance in the feedback path will cause not only a failure to receive, but perhaps a damage to the LNA (often they can't take more than 1 mW.) It's all doable... but it's much harder to do in a cell phone, when you have a whopping $0.50 in parts costs to spend on everything that it takes.
This whole approach has roots that go into early telephony. Every landline telephone that uses a SINGLE copper pair has a hybrid circuit [wikipedia.org] that separates, as well as it can, the incoming and the outgoing voice. Traditional hybrids [wikipedia.org] use exactly the scheme that you described. It's easy to do in 300-3400 Hz, compared to 1-2 GHz.
I will make no claims with regard to {T,C}DMA and their spectral efficiency.
(Score: 1) by adolf on Friday February 28 2014, @01:41AM
In terms of phasing and attenuation, I'm pretty sure I mentioned that already. As I've said before, that's the easy part.
Getting things synchronized and keeping there is also easy. You act as if the concept of a PLL hasn't been around since forever.
Regarding telephones: Yes, that's a similar problem. It's also a solved problem using rudimentary parts, whereas we have ridiculously-fast DSPs these days. *shrug* (And in other news, AMPS is dead, and cell phones have been much fancier than an FM transceiver for just a little while now...)
Regarding cost: Sheesh. With an attitude like that, it'll be a wonder if this color television thing ever takes off -- the sets are just so expensive.
Needing a transformer to rotate phase? Puh-leeze.
And if you don't want to comment about TDMA and spectral efficiency, why did you bring up TDMA in a discussion about spectral efficiency?
Aaand. Yep, that's enough for me on this thread.
Cheers.
I'm wasting my days as I've wasted my nights and I've wasted my youth
(Score: 2) by FatPhil on Friday February 28 2014, @06:33AM
However, that wouldn't make it just a badly written article, it would make the article just plain wrong:
"radio equipment – such as that used by mobile telephones "
So we're talking about handsets, not base stations. And by knowing about base stations, *it* (the handset).
Were it talking about base stations, the 2nd sentence would be "By knowing what it is transmitting, a base station can cancel out
I can't believe The Reg, even if it's fallen a long way from where it once was, could fuck up things so badly. (Then again, I've not read it for half a decade, it fell so far.)
So I'm still no wiser. Let's just look for NSF grant applications, and see if they're just after free money.
Making a public pledge to no longer contribute to slashdot
(Score: 2) by bd on Friday February 28 2014, @07:33AM
The article is indeed badly written up. See: http://kumunetworks.com/ [kumunetworks.com]. They have a paper linked directly on their homepage describing what they want to do.
To quote the paper:
See figure 2 for an overview, they insert an analog circuit between PA and LNA that cancels interference from the transmitter. While the concept is certainly not novel, their particular interference cancellation circuit may very well be.
(Score: 3, Interesting) by frojack on Thursday February 27 2014, @08:00PM
Well, I read a paper some time ago that sometimes adding noise helps detecting a signal. It might have been this one [ieee.org] or maybe it was this one [ieee.org] but I can't recall exactly, and its entirely possible I misunderstand the entire issue.
Discussion should abhor vacuity, as space does a vacuum.
(Score: 1) by Cactus on Thursday February 27 2014, @05:48PM
Yeah, but they achieve the result a different way, I guess. Just imagine it's the 80's, and they added a clock/radio in to something new.
(Score: 4, Funny) by Sir Garlon on Thursday February 27 2014, @06:11PM
But this time it's on mobile devices!
Just like, those of us who are old enough to remember the dot-com bubble remember, all those patents for obvious prior art, but on the Internet!
Next it will be a wave of patents leveraging big data!
Yes, my country has a stupid patent system. I'm kind of embarrassed about it. I wrote to my Senator, but she hasn't got around to fixing it yet.
[Sir Garlon] is the marvellest knight who is now living, for he destroyeth many good knights, for he goeth invisible.
(Score: 2) by maxwell demon on Saturday March 01 2014, @03:57AM
It may be almost forgotten due to the flood of bad patents, but a patent is not meant to protect a concept, but a specific invention, that is a non-obvious implementation of the concept.
The Tao of math: The numbers you can count are not the real numbers.