I am facinated by your scope of research. As one that work in 5G industry, see how vendors strugle with each antena added to RU(4 unpolarised antennas on 30cm^2 RU due to heat disapation issues, massive MIMO, for sure🤦♂️). All those mechanical beamforming techiques, such as holographic beam forming look like great solution for near future 5G improvments. Hope my colegues understand that as well.
Thank you for great presentation. In the discribed RIS prototype, they used 2 voltages to represent the phase states of -90 and +90 in order to limit the complexity of the prototype ... If each diode can provide 2 phase states , why we need 2 diodes in each reflicting element ?
The RIS elements contain two different substrates, probably to achieve the right kind frequency response where the amplitude is high for both phase states. But I don’t know the exact motivation behind the design. You can find more details in the paper “Wireless Communications through a Simplified Architecture Based on Time-Domain Digital Coding Metasurface” that is mentioned as a reference for the design in the paper: arxiv.org/pdf/2103.00534.pdf
Hi Emil! At 34:30, the optimization problem is for the case when every channel between an i^th IRS element and the transmitter/receiver is a single-tap channel right? In case when every channel is a multipath channel, will the optimization problem be a non-convex problem? There we might have to approximate by choosing the strongest path for both the IRS-transmitter and IRS-receiver channels and align them in a direction that is opposite to the combined phase shifts of the two paths?
Yes, you are perfectly correct! The wideband case is covered in the magazine paper with the same title as this video. Here is a video where we discuss it in more detail: Reconfigurable Intelligent Surfaces for Wideband Communications: Challenges and Possible Solutions ruclips.net/video/3RjlCpvn0Dc/видео.html
@@WirelessFuture Thank you for your reply. I have another question. Usually, the IRS-transmitter and the IRS-receiver channel will always have a direct LOS path. Is it correct if we always tune the IRS towards a direction that depends on the LOS paths of the two channels? I mean, is the LOS path always the dominant path among the rest of the NLOS paths? In Rican, of course it will be but how about in general? Thank you once again!
@@WirelessFuture Watched the video! Got my answer to my second question. IRS is effective only when there exists a LOS path from transmitter to IRS or from IRS to receiver. And usually, this is the strongest path. Thanks!
@@WirelessFuture Btw, can you please share me the link of the paper. I tried searching it on Google scholar but could not find it. Is it not published yet?
Thank you again for this great presentation. 1- Could you please recommend a reference, preferably yours, about direction finding and/or beam tracking in the presence of RSI? 2. Also, does it make sense to implement an RSI surface with uniform linear elements, say lambda/2 spaced? 3. Will this ideal of RSI replace the standard phased array formed by antennas any time soon? Is it already deployed and tested in real wireless communication systems or this is for the future 6G systems? Thank you.
1. I haven’t written so much on beamtracking and channel estimation since there are many other good papers on that, but there is a section in the following paper: arxiv.org/pdf/2102.00742.pdf It also contains references to other works on the topic. 2. Yes, that makes good sense. Regular patterns make it easier to exploit geometrical properties. Smaller spacing lambda/2 is often considered to get more isotropic like radiation patterns. 3. I don’t think this is the goal. RIS is rather a competitor to conventional relays than to MIMO antennas. MIMO antennas are used for spatial multiplexing while RIS/relays are utilized to improve the propagation channel.
Hi. Thank you for such a great tutorial on RIS. I would like to know if there is any Github profile for this work so that we can reproduce these results and analyze the model ourselves.
There is a publication with the same name that contains some of the numerical results. The code from that paper can be found here: github.com/emilbjornson/SPM_RIS If there are a particular result that you are interested in and cannot be found in that paper, then please give me a time stamp from the video and I can guide you to the right publication.
The RIS should be configured to maximize the SNR at the receiver. This is achieved as explained on slides 18 and 20. The solution basically means that every RIS element is delaying the reflected signal so that the signals from all elements reach the receiver in phase. In a pure line-of-sight scenario, one can interpret the optimal operation as illustrated on slide 19. (There is a link to the slides in the description.)
Professor I have a query at 7:39 where the reflected beam is directed to user 1 and user 2 at two different times.. Is it possible to angle the beam to user 1 and user 2 at same time by adjusting the e^j_theta. thank you
The RIS must use the same configuration for all the incident signals. Since the two users want very two different phase shifts in this example, but you can only select one, you direct the beam to one user or find some trade off between them. There exist algorithms for finding such trade offs. I think these methods can be useful when the transmitter has multiple antennas, but less so in single-antenna setups when the base station can only transmit signals with one directivity. With a single-antenna transmitter, we are stuck with NOMA transmission, which isn’t particularly efficient since the degrees-of-freedom is always 1.
That's some great explanation. Thank you very much for putting this video just in time when I needed !! At 23:24 you mention an RIS prototype for WiFi band used 55 columns and 20 rows of reflecting elements. Is there a limit / constraint for the same ?
You can read about the prototype in this paper: arxiv.org/pdf/2103.00534.pdf There is no strict limit on how large it can be or which band that can be used, but you need to design the hardware for that band and the configuration takes more time when the surface is larger.
Thank you for great presentation Sir, At 20:30, I want to ask why do we ingore other possible paths which can affect the received signal? For instance, transmitted wave can achive to the receiver by reflecting from any wall of build. Should we extend direct channel expression for these paths?
These paths are included in the direct channel. It involves all paths except those going via the RIS. I've understood that the term "direct" can be misinterpreted as meaning a single line-of-sight path, but that isn't the intention. In fact, the RIS is mainly useful when the direct path doesn't include any line-of-sight path but is rather weak.
@@recepvural This slide is about a narrowband channel, which implies that the multipaths are not resolvable, but lumped together as one term with a common attenuation and delay. Wideband channels consist of a summation of paths that have this form, but with different attenuations and delays. This is the topic of a new video: “Reconfigurable Intelligent Surfaces for Wideband Communications: Challenges and Possible Solutions”.
How can we know that a wave is from certain direction since angles phi and theta can take continuous set of values between -pi to pi and -pi/2 to pi/2 in 3 dimensions?
This is part of the channel estimation problem. One can either try to estimate the individual elements, using LS methods that ignore the geometrical structure (arxiv.org/pdf/2102.00742), or one can try to identify directions using array signal processing methods. There is a wealth of classical angle-of-arrival estimation methods, such as ESPRIT and MUSIC
Wi-Fi Direct is a way for two devices to connect directly to each other, at high speed. D2D in cellular networks need the assistance of a base station so that the licensed spectrum is used efficiently.
What a great, clear, and informative presentation. Thank you very much for your efforts spreading the knowledge.
Thank you for your sharing and the professor's clear presentation, it's so helpful for the person who just started to study this field like me!
Thanks for this amazing video! Your explanation of RIS really helped clarify some concepts for my msc research
It is really beautiful explanation of how we can think of surface as curved for directivity. (At 37:22)
I am facinated by your scope of research. As one that work in 5G industry, see how vendors strugle with each antena added to RU(4 unpolarised antennas on 30cm^2 RU due to heat disapation issues, massive MIMO, for sure🤦♂️). All those mechanical beamforming techiques, such as holographic beam forming look like great solution for near future 5G improvments.
Hope my colegues understand that as well.
Thank you for your lecture, I would like to ask about slide 38:40 is there any specific article that addresses that issue?
This is a variation of Figure 4 from the first paper mentioned in the description: arxiv.org/pdf/2102.00742.pdf
Thank you for great presentation.
In the discribed RIS prototype, they used 2 voltages to represent the phase states of -90 and +90 in order to limit the complexity of the prototype ... If each diode can provide 2 phase states , why we need 2 diodes in each reflicting element ?
The RIS elements contain two different substrates, probably to achieve the right kind frequency response where the amplitude is high for both phase states. But I don’t know the exact motivation behind the design. You can find more details in the paper “Wireless Communications through a Simplified Architecture Based on Time-Domain Digital Coding Metasurface” that is mentioned as a reference for the design in the paper:
arxiv.org/pdf/2103.00534.pdf
Hi Emil! At 34:30, the optimization problem is for the case when every channel between an i^th IRS element and the transmitter/receiver is a single-tap channel right? In case when every channel is a multipath channel, will the optimization problem be a non-convex problem? There we might have to approximate by choosing the strongest path for both the IRS-transmitter and IRS-receiver channels and align them in a direction that is opposite to the combined phase shifts of the two paths?
Yes, you are perfectly correct! The wideband case is covered in the magazine paper with the same title as this video. Here is a video where we discuss it in more detail: Reconfigurable Intelligent Surfaces for Wideband Communications: Challenges and Possible Solutions
ruclips.net/video/3RjlCpvn0Dc/видео.html
@@WirelessFuture Thank you for your reply. I have another question. Usually, the IRS-transmitter and the IRS-receiver channel will always have a direct LOS path. Is it correct if we always tune the IRS towards a direction that depends on the LOS paths of the two channels? I mean, is the LOS path always the dominant path among the rest of the NLOS paths? In Rican, of course it will be but how about in general? Thank you once again!
@@WirelessFuture Watched the video! Got my answer to my second question. IRS is effective only when there exists a LOS path from transmitter to IRS or from IRS to receiver. And usually, this is the strongest path. Thanks!
@@WirelessFuture Btw, can you please share me the link of the paper. I tried searching it on Google scholar but could not find it. Is it not published yet?
@@amitbora6336 arxiv.org/abs/2102.00742
Thank you again for this great presentation.
1- Could you please recommend a reference, preferably yours, about direction finding and/or beam tracking in the presence of RSI?
2. Also, does it make sense to implement an RSI surface with uniform linear elements, say lambda/2 spaced?
3. Will this ideal of RSI replace the standard phased array formed by antennas any time soon? Is it already deployed and tested in real wireless communication systems or this is for the future 6G systems?
Thank you.
1. I haven’t written so much on beamtracking and channel estimation since there are many other good papers on that, but there is a section in the following paper: arxiv.org/pdf/2102.00742.pdf
It also contains references to other works on the topic.
2. Yes, that makes good sense. Regular patterns make it easier to exploit geometrical properties. Smaller spacing lambda/2 is often considered to get more isotropic like radiation patterns.
3. I don’t think this is the goal. RIS is rather a competitor to conventional relays than to MIMO antennas. MIMO antennas are used for spatial multiplexing while RIS/relays are utilized to improve the propagation channel.
Hi. Thank you for such a great tutorial on RIS. I would like to know if there is any Github profile for this work so that we can reproduce these results and analyze the model ourselves.
There is a publication with the same name that contains some of the numerical results. The code from that paper can be found here: github.com/emilbjornson/SPM_RIS
If there are a particular result that you are interested in and cannot be found in that paper, then please give me a time stamp from the video and I can guide you to the right publication.
@@WirelessFuture Thank You for the quick reply. I really appreciate it.
nice presentation sir. which beamformer technique is based on reflection coefficient of RIS FOR EFFICIENT ENERGY.The reflect beam is from RIS to USER
The RIS should be configured to maximize the SNR at the receiver. This is achieved as explained on slides 18 and 20. The solution basically means that every RIS element is delaying the reflected signal so that the signals from all elements reach the receiver in phase. In a pure line-of-sight scenario, one can interpret the optimal operation as illustrated on slide 19. (There is a link to the slides in the description.)
you are special, Please go on, tanks more and more.
Professor I have a query at 7:39 where the reflected beam is directed to user 1 and user 2 at two different times.. Is it possible to angle the beam to user 1 and user 2 at same time by adjusting the e^j_theta. thank you
The RIS must use the same configuration for all the incident signals. Since the two users want very two different phase shifts in this example, but you can only select one, you direct the beam to one user or find some trade off between them. There exist algorithms for finding such trade offs. I think these methods can be useful when the transmitter has multiple antennas, but less so in single-antenna setups when the base station can only transmit signals with one directivity. With a single-antenna transmitter, we are stuck with NOMA transmission, which isn’t particularly efficient since the degrees-of-freedom is always 1.
That's some great explanation. Thank you very much for putting this video just in time when I needed !!
At 23:24 you mention an RIS prototype for WiFi band used 55 columns and 20 rows of reflecting elements. Is there a limit / constraint for the same ?
You can read about the prototype in this paper: arxiv.org/pdf/2103.00534.pdf
There is no strict limit on how large it can be or which band that can be used, but you need to design the hardware for that band and the configuration takes more time when the surface is larger.
Thank you for great presentation Sir,
At 20:30,
I want to ask why do we ingore other possible paths which can affect the received signal? For instance, transmitted wave can achive to the receiver by reflecting from any wall of build. Should we extend direct channel expression for these paths?
These paths are included in the direct channel. It involves all paths except those going via the RIS. I've understood that the term "direct" can be misinterpreted as meaning a single line-of-sight path, but that isn't the intention. In fact, the RIS is mainly useful when the direct path doesn't include any line-of-sight path but is rather weak.
@@WirelessFuture Then, how could we express multipath channel with single attenuation and phase component (h=sqrt(ρ)𝛿(t-𝜏)) at 31:10 ?
@@recepvural This slide is about a narrowband channel, which implies that the multipaths are not resolvable, but lumped together as one term with a common attenuation and delay. Wideband channels consist of a summation of paths that have this form, but with different attenuations and delays. This is the topic of a new video: “Reconfigurable Intelligent Surfaces for Wideband Communications: Challenges and Possible Solutions”.
How can we know that a wave is from certain direction since angles phi and theta can take continuous set of values between -pi to pi and -pi/2 to pi/2 in 3 dimensions?
This is part of the channel estimation problem. One can either try to estimate the individual elements, using LS methods that ignore the geometrical structure (arxiv.org/pdf/2102.00742), or one can try to identify directions using array signal processing methods. There is a wealth of classical angle-of-arrival estimation methods, such as ESPRIT and MUSIC
Can you show me what the operating frequency is?
The RIS technology can be implemented in many different frequency bands, but the experimental results were conducted in the 5.8 GHz band using WiFi.
In d2d mode not needed bs at slll?
Wi-Fi Direct is a way for two devices to connect directly to each other, at high speed. D2D in cellular networks need the assistance of a base station so that the licensed spectrum is used efficiently.
wow....that's a great presentation ...