Li-Fi to replace Wi-Fi? Sounds like Sci-Fi or bullcrap

[BrianHG]

This will be the future of WiFi : http://www.newelectronics.co.uk/electronics-news/full-duplex-radio-ics-transmit-and-receive-at-the-same-frequency/75059/

The document underestimates the speed increases.  A whole lot of handshaking & overhead will be lost once you sync 2 radios on the same frequency where both sides transmits continuously simultaneously, not to mention looking for clean alternate open channels.

I'm not sure why they didn't think of it the day dual band Wifi became commonplace. Negotiate with the device to transmit on one band while receiving with the other, full duplex.

When transmitting, you cannot receive on the same frequency at the same time.  The outgoing signal is 100 trillion times stronger than the receive signal making it impossible.  Don't get confused, our current WiFi do transmit and receive on the same frequencies, it's just that on side transmits, the other side shuts down the transmitter to listen, then, they swap.  This is called Half-duplex.  It only looks full to the PCs since this is going on so rapidly.

Prior to this IC, only military radar with reciprocal antennas, where they had special structures on the antenna so that the out-going signal was not seen by the receiving antenna in the same scanning dish.  By strategically designing the structure so that the RX part of the antenna was positioned not to see the XT output (signal was canceled out by the orientation and shape of the structure), yet still be in the same focal point of the dish antenna making the radar as accurate as possible.  It was the physical antenna which allowed the powerful transmit outgoing signal to always be on, yet, the RX antenna would not see that signal at all, just the super-weak reflections coming back.

[BrianHG]

Another advance in full duplex, this one explains how it was done they cancelled out that 100 trillion times stronger transmit while receiving at the same time:
https://phys.org/news/2017-03-two-way-radio-chip.html

It's coming....

[Marco]

With highly asymmetric data the impact will be unlikely to reach 2x or higher total throughput though. In fact I doubt it's even worth the headaches in cellular, where contention isn't an issue, the dynamic range and distances higher and the IC's already being complex enough as is.

[BrianHG]

PS. oh, it learns the multipath response during the preamble .... that's smart. The hand shaking is kind of an essential part of what makes it work, CSMA isn't going away either.

.

[BrianHG]

With highly asymmetric data the impact will be unlikely to reach 2x or higher total throughput though. In fact I doubt it's even worth the headaches in cellular, where contention isn't an issue, the dynamic range and distances higher and the IC's already being complex enough as is.

Try continuous realtime flow control and error correction without waiting for larger blocks to be transferred to attain speed.  It will get closer to Ethernet performance.  As for complexity, that's a load of BS.  Take a look at a modern cell phone CPU, even a bottom end one, it roasts the complexity of even this radio chip.  In the future, complexity and sophistication is going up, everything is not stooping at today's level, and we will call what we have today the definitive end of our radio technology.

[Marco]

.

It uses the pre-amble as a ping to measure and be able to correct for the reflections, which it corrects for digitally after solving the 100 trillion problem in analogue.

[David Hess]

Another advance in full duplex, this one explains how it was done they cancelled out that 100 trillion times stronger transmit while receiving at the same time:
https://phys.org/news/2017-03-two-way-radio-chip.html

I remember looking into this 20+ years ago.  The problem then was the shot noise from any active elements outside the path between the sampling point and cancelling point was greater than the received signal.  It worked fine if the received signal strength was large compared to the transmitter noise but that is a pretty big restriction.

This problem applies even if frequency-division multiplexing is used which explains why transmitters require passive filtering on their output whether they meet distortion requirements or not.  If transmit and receive are in the same band, then it is not sufficient to have only high Q bandpass filtering on only the receiver because it will pass the shot noise from the transmitter.

[NiHaoMike]

This will be the future of WiFi : http://www.newelectronics.co.uk/electronics-news/full-duplex-radio-ics-transmit-and-receive-at-the-same-frequency/75059/

The document underestimates the speed increases.  A whole lot of handshaking & overhead will be lost once you sync 2 radios on the same frequency where both sides transmits continuously simultaneously, not to mention looking for clean alternate open channels.

I'm not sure why they didn't think of it the day dual band Wifi became commonplace. Negotiate with the device to transmit on one band while receiving with the other, full duplex.

When transmitting, you cannot receive on the same frequency at the same time.  The outgoing signal is 100 trillion times stronger than the receive signal making it impossible.  Don't get confused, our current WiFi do transmit and receive on the same frequencies, it's just that on side transmits, the other side shuts down the transmitter to listen, then, they swap.  This is called Half-duplex.  It only looks full to the PCs since this is going on so rapidly.

Prior to this IC, only military radar with reciprocal antennas, where they had special structures on the antenna so that the out-going signal was not seen by the receiving antenna in the same scanning dish.  By strategically designing the structure so that the RX part of the antenna was positioned not to see the XT output (signal was canceled out by the orientation and shape of the structure), yet still be in the same focal point of the dish antenna making the radar as accurate as possible.  It was the physical antenna which allowed the powerful transmit outgoing signal to always be on, yet, the RX antenna would not see that signal at all, just the super-weak reflections coming back.

Why can't, for example, the AP transmit on 5GHz and the client transmit on 2.4GHz? Kind of like what drone pilots do for remote control with real time video.

[BrianHG]

This will be the future of WiFi : http://www.newelectronics.co.uk/electronics-news/full-duplex-radio-ics-transmit-and-receive-at-the-same-frequency/75059/

The document underestimates the speed increases.  A whole lot of handshaking & overhead will be lost once you sync 2 radios on the same frequency where both sides transmits continuously simultaneously, not to mention looking for clean alternate open channels.

I'm not sure why they didn't think of it the day dual band Wifi became commonplace. Negotiate with the device to transmit on one band while receiving with the other, full duplex.

When transmitting, you cannot receive on the same frequency at the same time.  The outgoing signal is 100 trillion times stronger than the receive signal making it impossible.  Don't get confused, our current WiFi do transmit and receive on the same frequencies, it's just that on side transmits, the other side shuts down the transmitter to listen, then, they swap.  This is called Half-duplex.  It only looks full to the PCs since this is going on so rapidly.

Prior to this IC, only military radar with reciprocal antennas, where they had special structures on the antenna so that the out-going signal was not seen by the receiving antenna in the same scanning dish.  By strategically designing the structure so that the RX part of the antenna was positioned not to see the XT output (signal was canceled out by the orientation and shape of the structure), yet still be in the same focal point of the dish antenna making the radar as accurate as possible.  It was the physical antenna which allowed the powerful transmit outgoing signal to always be on, yet, the RX antenna would not see that signal at all, just the super-weak reflections coming back.

Why can't, for example, the AP transmit on 5GHz and the client transmit on 2.4GHz? Kind of like what drone pilots do for remote control with real time video.

You get more throughput when both transmit on the higher bandwidth channels.  And, why not have both transmit and receive on both bands simultaneously is you want 4x speed.  Though, with the allowed power for consumer devices, you might get better distance or bandwidth on one band versus the other.  Also, what if one band is occupied by other devices.  At least, if your sender is already transmitting, you can send data back on the same channel without eating up a second band which may have other traffic.  Freeing up channels means lest wasted bandwidth/congestion.  I know my WiFi suffers from all the other new routers my neighbors have installed recently.  I also know going from my laptop to my main WiFi router is twice as fast as the other direction on channel 2, yet, on channel 1, this difference in UL/DL performance is swapped.  With this new tech, the fastest channel I choose will be identical in both directions.

[NiHaoMike]

Why can't, for example, the AP transmit on 5GHz and the client transmit on 2.4GHz? Kind of like what drone pilots do for remote control with real time video.

You get more throughput when both transmit on the higher bandwidth channels.  And, why not have both transmit and receive on both bands simultaneously is you want 4x speed.  Though, with the allowed power for consumer devices, you might get better distance or bandwidth on one band versus the other.  Also, what if one band is occupied by other devices.  At least, if your sender is already transmitting, you can send data back on the same channel without eating up a second band which may have other traffic.  Freeing up channels means lest wasted bandwidth/congestion.  I know my WiFi suffers from all the other new routers my neighbors have installed recently.  I also know going from my laptop to my main WiFi router is twice as fast as the other direction on channel 2, yet, on channel 1, this difference in UL/DL performance is swapped.  With this new tech, the fastest channel I choose will be identical in both directions.

What about when latency is the priority rather than bandwidth? I also think that for most real world bulk traffic that is mostly in one direction, putting the reverse traffic on a different band would improve overall performance.

[BrianHG]

What about when latency is the priority rather than bandwidth? I also think that for most real world bulk traffic that is mostly in one direction, putting the reverse traffic on a different band would improve overall performance.

You obviously ignored to the 2 linked document I posted.  The whole purpose of the FULL DUPLEX chip is when it transmits, it can receive external broadcasts on the same frequency coming from the other side while it's transmitting on the exact same frequency as the FULL DUPLEX radio on the other side is doing the exact same thing at the same time.

Question, do you know understand meaning of full duplex communication?  This new IC isn't emulating full duplex by rapidly switching between transmit and receive, this is what we have now.  This full duplex radio IC is doing it for real.

There would be no purpose in developing such an IC if we already had it.  Right now we have half duplex and it sucks.  Just using 2 different bands means either wasting bandwidth, having to use 2 very spaced apart bands, one which might not have the same speed at different distances and may have different speeds.  Also, if since both sides are always transmitting at the same frequency at the same time, you know your channel is clear both ways.  With full duplex, not switching delays since you can transmit while receiving right on the exact same frequency.

[NiHaoMike]

Doing full duplex across two separate bands is far easier to implement. I would like to see how that single band full duplex chip really performs in the real world. As a related example, MU-MIMO sounds good on paper but doesn't always work well in the real world.

[BrianHG]

Doing full duplex across two separate bands is far easier to implement. I would like to see how that single band full duplex chip really performs in the real world. As a related example, MU-MIMO sounds good on paper but doesn't always work well in the real world.

Ok, yes, this is a legitimate concern.

[f4eru]

PS. oh, it learns the multipath response during the preamble .... that's smart. The hand shaking is kind of an essential part of what makes it work, CSMA isn't going away either.

Not a very convincing system...
Won't do more than 2-3 meters, at most, then you're submerged in your TX noise.
Also, for "hand shaking" don't shake your hands nearby, you'll change the multipath propagation, and the handshake has to happen again.

[Marco]

Isn't there a preamble for every frame?

[borjam]

The problem with the 2.4 and 5.8 GHz bands is, of course, interference. It's especially strong in high density dwellings such as appartment buildings.

The 60 GHz (802.11ad) band would be a good candidate because it won't cross a wall, which means neighbor interference will be non existant. Even windows will attenuate it a lot. The problem, of course, you would need an access point per room. But that means clean spectrum for everyone.

As for lasers, I remember that in the early 90's there was a product called Terabeam that transmitted 10 Mbps Ethernet using lasers. It worked pretty well although it had a weakness, the lasers had to be replaced periodically.

I am now trying a 60 GHz link (Mikrotik Wireless Wire) just to cross a street (20 - 30 metres) and the performance is incredible. Almost the same as digging a trench and laying a fibre. And it's not expensive, less than  €200 a pair. Using a phased array with some degree of beam forming you don't even need to aim them very carefully. just use a rule of thumb.

[Marco]

Millimeter wave is almost as pure line of sight as light. Indoor it doesn't really have any advantage to light, outdoor it's better with fog.