"Levis said a researcher even told the students their idea was "so simple and effective, it won't work," because something that obvious must have already been tried unsuccessfully."
The problem with full duplex wireless comm. on the same frequency is that the transmitted signal is much stronger than the received signal. Therefore, echo cancelling is usually impractical.
Not sure how these guys overcame the difficulties with transmitter nonlinearities and other unknown impairments.
EDIT: they claim to do partial cancellation by placing the receiver antenna at the null of the two transmitting antennas (getting 50 dB reduction in echo power). Whether this is practical in a commercial low cost radio remains to be seen.
it may be practical in the office setup where 6+ antenna phased arrays Wi-Fi switches are "cost-practical". With phased array you have much more geometrical flexibility - you may "narrow beam" transmission to a client while better adjusting the transmitter's null onto the receiver's physical location to the current conditions, reflections, etc...
No, I think that the parent post was referring to the transmitters on the radio device. The transmitting and receiving antennas are different, and they place the receiving antenna in a null of the transmitters.
This is easy and practical because they're in a location which is fixed by the PCB/housing. There's no phased array required.
No, reflections can have any phase, so you can't possibly cancel them out with a fixed geometry. But given the 1/r^2 falloff, unless the reflections are from very nearby, they will be a lot weaker than the direct path.
Don't be trite. Receiving is completely different from sending. "You" doesn't include the other guy, in my dictionary.
It may raise the amount that can be sent over the network in toto, assuming that data demand is the same in all directions. Which it almost never is.
So what they've done is, they allow the data uplink to occur simultaneously with the downlink.
A valid claim is, the latency of data is improved - you don't have to wait/schedule your sends around the receive traffic. Which can matter quite a bit in the bulk of cases e.g. downloading http while jabbing buttons/sending commands up to the server.
it's fairly standard terminology...you've doubled the amount of data that can be sent over a given chunk of spectrum per unit time, without changing the modulation/encoding scheme (bits/hz). A normal full duplex FDD system takes twice the bandwidth to pass the same amount of information as this theoretically can.
Rather than think about bursty, asymmetric Ethernet bits, consider passing something like a fully allocated T1 over it. Symmetric TDM links aren't totally dead, there are a lot of T1 radios out there. What used to take X Hz of bandwidth, theoretically can be done in X/2...without changing the modulation.
I don't think it will actually be used that way, but it is a breakthrough of sorts to be able to do so.
Some people don't know what they're going to say until they say it. If such people could send and receive at the same time, they would be getting more information per unit time.
I don't think that is the case. Suppose a given channel (e.g. a frequency) has a maximum information per second it can carry X, this system will not allow 2X information to be carried. It will allow the information carrying capacity of the channel to be split, such that each side can carry a maximum amount of information A and B respectively such that A+B<=X.
"this system will not allow 2X information to be carried"
I believe that is the entire point of the article. To allow double the capacity between two devices. The original capacity would be the amount of data that could be sent by a single device. However if both devices can send at the same time that is double capacity.
The way "full-duplex" fixed wireless gear often works is by splitting the spectrum in two halves (plus a sometimes small guard band: frequency division). Each end uses one half for TX. Alternatively, you use the full band but switch off which end is transmitting (time division). So it really does double (or even a little more, if eliminating a guard band) the amount each side can send.
If I'm understanding correctly this does double the amount of information sent in a chunk of time.
Example:
You send 1Mbit over a sec. Then you pause for 1 sec to receive 1Mbit. Effectively, you've only sent 1Mbit over 2 seconds.
Now, due to being in full duplex mode, you can send 1Mbit over 1 sec while simultaneously receiving 1Mbit over 1 sec. Thus, allowing you to send 2Mbit over 2 seconds. 1Mbit/sec is double .5Mbit/sec.
Not to belittle their announcement, but what textbook says you can't filter a signal? Or is the novelty in the degree to which they had to filter the outgoing signal?
They put the receiver in a "node" of destructive interference between two transmitting antennas. That way, the transmitting signal doesn't swamp the receiver.
> ... but what textbook says you can't filter a signal ...
i don't think any textbooks makes that claim. from the article: ...a researcher even told the students their idea was "so simple and effective, it won't work," because something that obvious must have already been tried unsuccessfully.
The much larger signal (local TX) would always capture your RF chain, being that using standard selectivity techniques in the analog domain the unwanted looks exactly like the desired. Being so large, it also would generally saturate your receiver.
Can't each transmitter/radio add some type of a 2-nd order frequency onto the transmitting signal (imagine a sin-wave, except it's line is not smooth, the line has its own "frequency") and detect that to filter the incoming signal?
what you proposing is equivalent to increasing the frequency of the channel. If both do the same "2-nd" order frequency - we're back to the same "1-frequency" channel problem. If different - that's outside of the "1-frequency" channel problem.
Each radio would have a unique 2-nd order frequency for it's own filtering use, as a "tag" of it's signal. The primary frequency is known and used for the transmition, and the 2-nd order one is piggybacked on it. I'm not saying there are 2 transmiting frequencies.
> I'm not saying there are 2 transmiting frequencies.
200 years ago, people might have believed you. Then Fourier showed that your statements contradict each other. He said a lot of other interesting things - Bracewell's book is a good introduction.
That's going up on the wall above my monitor.