A part of the signal will be reflected, a part will be absorbed, and a part will penetrate the object. The balance between these effects depends on the material. The reflected signal normally follows Snell’s law, in terms of the main direction of the reflected signal. But there is also metamaterials that reflect signals in “weird” ways and rough material that scatter signals diffusely.
Yet there's a problem how would the antenna estimates the phaseshift I think we need a couple of omnidirectional Antennas just to detect the rotation of the user and then calculating the phase shift
Yes, in this setup, we are only maximizing the amplitude of a sinusoid without caring about its phase. In practice, one would modulate information symbols onto the sinusoid and then some of them must be predefined "pilot" symbols that enable the receiver to estimate the phase. These symbols are called "demodulation reference signals" in 5G.
What if there were two mirrors? One on the left of line-of-sight and the other on the right. And line-of-sight being blocked by an obstacle. Would it be possible to emit two beams for the two mirrors simultaneously using the current setup? If not, then would additional gain control on the signal paths help in any way to achieve multiple beams?
Yes, that is possible. What one would do in practice is to measure the channel’s phase shift between each transmit antenna and the receiver, and then compensate for them so that all antennas emit signals that reach the receiver in phase. This will lead to a combined radiation pattern that direct power towards both mirrors. There will be less power reaching each mirror, but the total received power becomes larger than if one only transmits to one of them.
Thank you for the interesting video. You illustrated it in a way that analog beamforming and digital beamfoming look similar. Could you add some clarification about the difference between the two. Thanks
I sent a single data signal at a single frequency, so then analog and digital beamforming are basically equivalent. The benefit of digital beamforming is that we can vary the beamforming freely over subcarriers and can send multiple data signal per subcarriers.
Think it is more about implementation: in analog BF the phase shifting is applied after digital to analog conversion and you can have some steps like as in this experiment so in result you can get some finite preset of beams you can choose from. While in digital beamforming you will see a precoding before D/A and there is the access to each individual antenna element so this allows different powers and phases to different antennas elements making MU-MIMO possible and more flexible. Of course it goes with the cost of having digital chain for each antenna element.
Interesting as always, thanks. Not directly related to that video, I read recently about the channel bandwidth of WLAN in the 2.4 GHz band. Is it wrong to say, doubling the bandwidth from 20 to 40 MHz, doubles the data rate? With regard to the well known SHANNON formular, we raise the noise term (B*N0) - that can leads to a decline of the SNR (P remains equal). In the nominator we have still the same value with P and the SNR= P/B*N0 gets smaller with higher B. So we don't exactly double the data rate by doubling B. Is that conclusion right?
Yes, that is correct. One needs to also double P to get twice the capacity. I don’t know the exact details related to WLAN, but I think there is a spectral power limitation of a certain number of dBm per Hz of spectrum, so if we increase the bandwidth, we are also allowed to increase the transmit power proportionally. There is likely a maximum total transmit power per device.
Thank you for the interesting video. This is for the single user receiver. Could you please give some ideas if there are multiple receiver how the beamforming is formed for individual user. Thank you.
You can compute the maximum ratio transmission beamforming for each user, as if the user was alone, and then transmit all the signals simultaneously. The issue with this approach is that you don't know how much interference the users will cause to each other - it will depend on how different the channels happen to be. That is why zero-forcing and regularized zero-forcing are better choices, since these adapt the transmission to reduce the interference. We describe this in the following paper: arxiv.org/pdf/1404.0408.pdf Here is a video that also discuss these things at a high level: th-cam.com/video/OkTHsTHPTQA/w-d-xo.html
How frequency flexible would this be? Because in real world, the tx freq would not be at a fixed 36.5GHz frequency. Also how does RF bandwidth effect this (for example, a 250MHz wide transmit signal). I know with 60GHz RF BW is 2.16GHz wide, which seems like it would be an issue. Would like to know your thoughts. Thanks!
The channel response varies with the frequency, particularly, in multipath scenarios. The coherence bandwidth is a rough measure of the width of the frequency interval for which the channel response is approximately constant. If the signal bandwidth is larger than the coherence bandwidth, it is preferable to use digital/hybrid beamforming so that the directivity of the transmission can be adapted to the channel variations. This was not considered in this video.
Yes, both terms have been used over the years. People often associate zero-forcing with a situation where you get zero interference between a set of users that are served simultaneously, but one can put the zeros/nulls anywhere.
great video, Professor as usual ... I have a question if you allow me. Maximum ratio beamforming is a diversity technique where a multipath prorogation is needed to add different paths SNRs together to increase the overall received signal (using MRC technique) ... however, mm Wave propagation is mostly LOS by nature. thus, is it a must to have blockers (to create multipath) to use this diversity technique?
You are right that MRC was originally developed to provide spatial diversity. However, MRT/MRC are techniques that give both a power gain (beamforming gain) and a diversity gain. If there is no fading, there is no diversity gain, but still a power/beamforming gain. It is not bad to lose the diversity gain in this way - a channel without fading is better than a channel with fading (if the average power is the same), but the difference vanishes when diversity is achieved.
If the antennas are phase shifted for maximum constructive interference, then can you achieve maximum destructive interference by simply inverting all antennas except the non-shifted one?
More or less. Suppose you have four antennas that are phase-aligned and contribute with equally much power. You can then achieve destructive interference by turning the phases of half the antennas by 180 degrees. Two antennas will then cancel the other two antennas.
Thanks Prof, for this insightful presentation. How to calculate ZF and MRT theoretical? And what this the difference between these two beamforming techniques? Thanks.
I describe the setup in the first video of the series. There is a link to the hardware in the description. I’m not sure if this answers your question, or if you wonder something more specific.
Yes, reflection can turn the signal up-side-down, which is basically a phase-shift by 180 degrees. In addition to that, the propagation delay also creates a phase-shift. The receiver needs to know the phase-shift if it should be able to demodulate data (except for amplitude-shift-keying that isn't using the phase). This is normally achieved by sending known signal before the data transmission begins, called a reference signal or pilot signal.
@@WirelessFuture Will the receiver interpret an inverted signal as simply 180 deg phase shifted? Wouldn't the waveform be wrong, and therefore impossible for the receiver to interpret without inverting it back to normal?
Yes, only a 180 degrees phase shift. The reason is that we transmit data with a bandwidth that is much smaller than the carrier frequency. For example, if you use 100 MHz of spectrum at 10 GHz, then each data symbol oscillates 100 times so a phase shift of half a period is not a big deal. But the receiver needs to know what the shift is so it can “rotate” the phase correctly. If you think about a 16-QAM signal, the phase shift will be like rotating the constellation diagram, and the receiver must know how to rotate it back.
![Beamwidth, beam pattern, and sensing [Part 5, Fundamentals of mmWave communication]](http://i.ytimg.com/vi/UbA4IbFXTSU/mqdefault.jpg)
![Beamwidth, beam pattern, and sensing [Part 5, Fundamentals of mmWave communication]](/img/tr.png)
![Constructive and destructive interference [Part 3, Fundamentals of mmWave communication]](http://i.ytimg.com/vi/hxkojUEKAR0/mqdefault.jpg)
![Beamforming directivity [Part 1, Fundamentals of mmWave communication]](http://i.ytimg.com/vi/WOE_nKBKUsE/mqdefault.jpg)
Really nice content. Any plan to show a RIS related experiment
Not right now, but if I could get my hands on a RIS for mmWave frequencies, I would love to continue this video series with an episode on that
thanks, it was quite informative
Super
Thank you Professor, I enjoy watching your videos. what would happen If you used something other than mirror ?
A part of the signal will be reflected, a part will be absorbed, and a part will penetrate the object. The balance between these effects depends on the material. The reflected signal normally follows Snell’s law, in terms of the main direction of the reflected signal. But there is also metamaterials that reflect signals in “weird” ways and rough material that scatter signals diffusely.
@@WirelessFuture thanks for complete responding.
Excelente explicação
so cool
Yet there's a problem how would the antenna estimates the phaseshift I think we need a couple of omnidirectional Antennas just to detect the rotation of the user and then calculating the phase shift
Yes, in this setup, we are only maximizing the amplitude of a sinusoid without caring about its phase. In practice, one would modulate information symbols onto the sinusoid and then some of them must be predefined "pilot" symbols that enable the receiver to estimate the phase. These symbols are called "demodulation reference signals" in 5G.
What if there were two mirrors? One on the left of line-of-sight and the other on the right. And line-of-sight being blocked by an obstacle. Would it be possible to emit two beams for the two mirrors simultaneously using the current setup? If not, then would additional gain control on the signal paths help in any way to achieve multiple beams?
Yes, that is possible. What one would do in practice is to measure the channel’s phase shift between each transmit antenna and the receiver, and then compensate for them so that all antennas emit signals that reach the receiver in phase. This will lead to a combined radiation pattern that direct power towards both mirrors. There will be less power reaching each mirror, but the total received power becomes larger than if one only transmits to one of them.
@@WirelessFuture Thank you for taking time to answer this.
Thank you for the interesting video. You illustrated it in a way that analog beamforming and digital beamfoming look similar. Could you add some clarification about the difference between the two. Thanks
I sent a single data signal at a single frequency, so then analog and digital beamforming are basically equivalent. The benefit of digital beamforming is that we can vary the beamforming freely over subcarriers and can send multiple data signal per subcarriers.
Think it is more about implementation: in analog BF the phase shifting is applied after digital to analog conversion and you can have some steps like as in this experiment so in result you can get some finite preset of beams you can choose from. While in digital beamforming you will see a precoding before D/A and there is the access to each individual antenna element so this allows different powers and phases to different antennas elements making MU-MIMO possible and more flexible. Of course it goes with the cost of having digital chain for each antenna element.
Interesting as always, thanks.
Not directly related to that video, I read recently about the channel bandwidth of WLAN in the 2.4 GHz band. Is it wrong to say, doubling the bandwidth from 20 to 40 MHz, doubles the data rate?
With regard to the well known SHANNON formular, we raise the noise term (B*N0) - that can leads to a decline of the SNR (P remains equal). In the nominator we have still the same value with P and the SNR= P/B*N0 gets smaller with higher B. So we don't exactly double the data rate by doubling B.
Is that conclusion right?
Yes, that is correct. One needs to also double P to get twice the capacity. I don’t know the exact details related to WLAN, but I think there is a spectral power limitation of a certain number of dBm per Hz of spectrum, so if we increase the bandwidth, we are also allowed to increase the transmit power proportionally. There is likely a maximum total transmit power per device.
Thank you for the interesting video. This is for the single user receiver. Could you please give some ideas if there are multiple receiver how the beamforming is formed for individual user. Thank you.
You can compute the maximum ratio transmission beamforming for each user, as if the user was alone, and then transmit all the signals simultaneously. The issue with this approach is that you don't know how much interference the users will cause to each other - it will depend on how different the channels happen to be. That is why zero-forcing and regularized zero-forcing are better choices, since these adapt the transmission to reduce the interference. We describe this in the following paper: arxiv.org/pdf/1404.0408.pdf
Here is a video that also discuss these things at a high level:
th-cam.com/video/OkTHsTHPTQA/w-d-xo.html
How frequency flexible would this be? Because in real world, the tx freq would not be at a fixed 36.5GHz frequency. Also how does RF bandwidth effect this (for example, a 250MHz wide transmit signal). I know with 60GHz RF BW is 2.16GHz wide, which seems like it would be an issue. Would like to know your thoughts. Thanks!
The channel response varies with the frequency, particularly, in multipath scenarios. The coherence bandwidth is a rough measure of the width of the frequency interval for which the channel response is approximately constant. If the signal bandwidth is larger than the coherence bandwidth, it is preferable to use digital/hybrid beamforming so that the directivity of the transmission can be adapted to the channel variations. This was not considered in this video.
Thanks, the term null steering can be seen as Zero Forcing - is that right?
Yes, both terms have been used over the years. People often associate zero-forcing with a situation where you get zero interference between a set of users that are served simultaneously, but one can put the zeros/nulls anywhere.
great video, Professor as usual ... I have a question if you allow me.
Maximum ratio beamforming is a diversity technique where a multipath prorogation is needed to add different paths SNRs together to increase the overall received signal (using MRC technique) ... however, mm Wave propagation is mostly LOS by nature. thus, is it a must to have blockers (to create multipath) to use this diversity technique?
You are right that MRC was originally developed to provide spatial diversity. However, MRT/MRC are techniques that give both a power gain (beamforming gain) and a diversity gain. If there is no fading, there is no diversity gain, but still a power/beamforming gain. It is not bad to lose the diversity gain in this way - a channel without fading is better than a channel with fading (if the average power is the same), but the difference vanishes when diversity is achieved.
If the antennas are phase shifted for maximum constructive interference, then can you achieve maximum destructive interference by simply inverting all antennas except the non-shifted one?
More or less. Suppose you have four antennas that are phase-aligned and contribute with equally much power. You can then achieve destructive interference by turning the phases of half the antennas by 180 degrees. Two antennas will then cancel the other two antennas.
@@WirelessFuture Love those intuitive tips. But which half? Any half?
Any half. In the video, I inverted every other antenna, but any two would have worked out.
Thanks Prof, for this insightful presentation.
How to calculate ZF and MRT theoretical?
And what this the difference between these two beamforming techniques?
Thanks.
Here is a paper where we explain these things in a rather self-contained way: arxiv.org/pdf/1404.0408.pdf
Great video! Can you please share the whole setup (parts). I’d like to replicate but using a real beamformer
I describe the setup in the first video of the series. There is a link to the hardware in the description. I’m not sure if this answers your question, or if you wonder something more specific.
Thank you. Yes, that answered my question. I didn’t realize that it was in the description.
Is the next video published or not yet?
It has finally been published: th-cam.com/video/UbA4IbFXTSU/w-d-xo.html
@@WirelessFuture 🙂🙂Thank you
The reflected transmission will be inverted at the receiver, correct? Can the receiver demodulate that?
Yes, reflection can turn the signal up-side-down, which is basically a phase-shift by 180 degrees. In addition to that, the propagation delay also creates a phase-shift. The receiver needs to know the phase-shift if it should be able to demodulate data (except for amplitude-shift-keying that isn't using the phase). This is normally achieved by sending known signal before the data transmission begins, called a reference signal or pilot signal.
@@WirelessFuture Will the receiver interpret an inverted signal as simply 180 deg phase shifted? Wouldn't the waveform be wrong, and therefore impossible for the receiver to interpret without inverting it back to normal?
Yes, only a 180 degrees phase shift. The reason is that we transmit data with a bandwidth that is much smaller than the carrier frequency. For example, if you use 100 MHz of spectrum at 10 GHz, then each data symbol oscillates 100 times so a phase shift of half a period is not a big deal. But the receiver needs to know what the shift is so it can “rotate” the phase correctly. If you think about a 16-QAM signal, the phase shift will be like rotating the constellation diagram, and the receiver must know how to rotate it back.
TY.
310324
Dear Professor
I sent you an invitation via LinkedIn