Thanks for the video 1- @2:39 you mentioned we can measure radio wave that are on the way to travel far away but from my understanding in the near field the power is reactive meaning imaginary not real so how can we measure that in practice or what do we actually measure ? 2- @3:21 you mentioned something about the radiative near field , it was mentioned that the measurements of the signal strength would vary with angle and distance may i know where is the mathematical formulation for this statement ? bec i found in Balanis book antenna theory analysis and design 2016 eq 4.23 that the radial component of the power is independent of the distance r 3-when we talk about different regions for ex as fraunhofer region in the context of antenna arrays how is the limit defined ? i mean we can draw several lines from the each Tx antenna to the Rx with different lengths 4-may i know how the signal properties (amplitude - power) would vary if we are talking about spherical wave vs plane wave ? 5-may i know how would the beamforming problem be different when we are at the near field ? i mean what should we consider in this case that was ignored in the case of the far field ? 6-if i have an expression for the electric field from maxwell equations , what should be done to get an expression or formulation for the complex channel ? 7- @29:55 it was mentioned that the antenna distance of lambda/2 is the critical distance for antenna coupling, may i know A-what is antenna coupling ? B-why is it something bad ? C-how does it affect MIMO capacity ? D-when does it happen ? E-how to avoid it ? F-why lambda/2 was set as a threshold for mutual coupling ? 8- @41:22 it was mentioned that with spherical wavefront we can send multiple streams from single antenna Tx to multi antenna Rx in LOS channel model and still receive them , can i know this is physically possible ? i mean if Tx sends s=x1+x2+x3 in spherical wave how can the multi antenna Rx resolve the interference ? 10- @45:04 it was mentioned that when number of antennas increases the beamwidth decreases proportional to M but this means that we can move into the beam and get out of it in a time less than the coherence time Tc of the channel thus we cannot do the channel estimation with periodicity of Tc, you mentioned that that beamwidth decreases but the fraounhfer distance increases so this do the needed equalization my questions are A-The fraunhofer distance is 2*d*M^2/lambda where lambda is the wavelength and d is the distance between the element in the array if we do the multiplication we would get something that scales with M not one right ? B-why do we need to multiply the fraunhofer distance by the beamwidth in the first place ? C-i read this paper here arxiv.org/abs/1511.02937 and they clearly states that the coherent time of the beam alignment each beam coherent time not channel coherence time how would this impact the answer to the question above 11- @45:05 do we need to do channel estimation after each beam alignment ? 12- @46:42 you mentioned another aspect regarding the beamforming in the near field in which the beamwidth would be constant and it cannot be smaller than the wavelength A-may i know why would it be constant B-why it cannot be smaller than the wavelength C-where are the proofs of all of that ? D-@47:33 why when we move wavelength divided by 8 we would lose 1/2 of power ? 13- another question in dipole antenna may i know why the lengths of dipole is lambda/2 or lambda/4 ?
1. The electrical field induces a current in the receive antenna that is then measured. We are always using complex numbers when analyzing electrical fields, but the imaginary part should be viewed as a phase-shift. 2. See the paper that we link to in the description. The main thing is whether the distance from any point on the transmitter to the measurement location is approximately the same (when measuring amplitude and phase) or not; that is, if the transmitter can be viewed as a point-source or not. 3. You can read about this in the paper as well. We will also release a more technical video in the coming weeks. 4. Waves are always spherical, but what matters here is whether the part of the wave that reaches the receive antenna "look" plane or not. More precisely, if there are phase and/or amplitude variations over the receive antenna (apart from those caused by the incident angle) or not. 5. We cannot use transmission/reception schemes that are designed for plane waves (i.e., based on array response vectors). Moreover, the shape of the "beam" around the focal point will look very different. See the figure in the background of the video for an example. The paper in the description gives more precise illustrations. 6. I don't understand the question. One can characterize everything from Maxwell's equations, but we generally make approximations to get more intuitive expressions. When moving from the reactive near-field to the radiative near-field to the far-field, we can use more and more approximations. 7. The simplest way to describe it is that mutual coupling occurs when antenna 2 is in the reactive near-field of antenna 1, so that they influence each other. I cannot explain this in detail in the comments field, but it basically means that the signals "leak" between the antennas and therefore the radiated waveform will not be the sum of the individual waveforms that the antennas would have generated in isolation. 8. You will have to transmit the signals in slightly different angular directions. Angel Lozano wrote a nice piece about this here: www.comsoc.org/publications/ctn/old-theory-new-tricks 10. When you beamform at a location in the far-field, then the beam doesn't appear until you reach a distance proportional to the Fraunhofer distance. This is illustrated in the paper mentioned in the description. If d_FA is the Fraunhofer array distance, then the angular beamwidth scales as 1/sqrt(d_FA). Hence, at the point where the beam begins the physical beamwidth grows as sqrt(d_FA). See equation (33) and set F=d_FA. 11. You need to do channel estimation first and then you can compute the optimal beam. 12. The calculation about the minimum beamwidth is provided on Page 4 in this paper: arxiv.org/pdf/1803.11023 13. There are two types of dipole antennas having these two different lengths. In summary, I recommend you read the paper that we link to in the description and in my reply. In coming weeks, we will also release a technical video called "Physically Large Antenna Arrays: When the Near-Field Becomes Far-Reaching".
Thank you for the information. Most recently wondered if in 5G mMTC application there are SIM cards in the devices. It sounds a bit unfitting for these tasks. Do you know that?
i’m fascinated by these videos🙏😎.. & Ur question .. yet ignorant of antenna tech ..but.. last 2 yrs my iPhone emitting bio magnetic pulses.. same smart watch gets Ur pulse & Heart EKG frequencies ..
The network operator needs to know which devices belong to what customers, which is one of the purposes having a SIM card. However, you don't need a physical card for achieving that. Many compact devices are using eSIM these days (e.g., Apple Watch) which is an embedded circuit that can store the same information as a physical SIM card. I believe this is what will be used in mMTC applications.
Thank you. How many maximal beams are possible with an array? I think in one video you said that theoretically infinite beams are. However, I read something about min(m,n) where m = number of transmit antennas and n = number of receive antennas. We speak about digital beamforming.
The maximum number of beams that can be transmitted simultaneously is infinite, but one cannot separate the signals from so many beams. The maximum number of separable beams is min(m,n) as you are pointing out. The intuition behind this number is quite easy: You cannot identify more than one signal per receive antenna and you shouldn’t transmit more signals than you have transmit antennas because then the beams become too wide and partially overlapping.
@@WirelessFuture Thank you, but when on has only a few antennas (e.g. 2), there can only a few beams. Is that correct? One has only limited number of parameters to adjust (1 x phase per antenna and 1x amplitude per antenna).
@@jasminnadic2103 With digital beamforming, you can design the signal that is emitted from each antenna in any way you like at each antenna. It can be the summation of many signals with different amplitudes and phases. But the beams that you can create with two antennas will be wide (the width is roughly proportional to 1/"number of antennas"), so you can only create two non-overlapping beams. This is usually what matters in practice: Don't transmit more beams that you have antennas, and preferably have more antennas than beam so that there is some natural separation between the beams. The fact that one can transmit an infinite number of beams is more a philosophical thing.
I am not sure if it makes so much sense to use DNN or NN in general as Erik suggested. I guess sth that is being ignored here is that for applying most of ML models (RL is an expectation), one needs to collect sufficient data. And this data (or fingerprints) changes based on the location as Erik mentioned again (so many objects in the room), so I guess a paper using DNN for such purpose might end up in the same place with those who ignored coupling :D
Thank you for the information. One question, please. With respect to multiplexing in 5G. 5G uses OFDMA with a flexible numerology that is clear. Do the different users then get different subcarriers, respectively, for the whole time? So that they use a special bandwidth exclusively, or is there a additional a TDMA and within timeslots they share the bandwidth? Many thanks.
5G is packet-switched so the operation focuses on making sure that limited-sized packets of data are delivered as soon as possible. Each user is assigned some of the subcarriers for the time it takes to transmit the current packet. Importantly, multiple users can be assigned to the same subcarrier at the same time. They are then separated using beamforming instead. This is what Massive MIMO is all about.
Thank you, as always it was an interesting episode. In the far field one can find the equation E=1/r and H=1/r and S=1/(r^2) -> There isn't any dependance on the frequency (wavelength). However, one says that transmission depends heavily on the frequency and in the FRIIS Equation there is the frequency (wavelength). To me it looks contradictionary. Have you spoken about that and I have missed it? Thank you very much.
The frequency dependence in Friis equation is due to the size of the receive antenna, which shrinks with the wavelength when considering an isotropic antennas (or any other fixed-gain antenna). We talk about this in the episode about mmWave communications as well as in the episode with Tom Marzetta. I also recommend the following blog post: ma-mimo.ellintech.se/2019/10/29/is-the-pathloss-larger-at-mmwave-frequencies/
![Ep 23. Wireless Localization and Sensing (With Henk Wymeersch) [Wireless Future Podcast]](http://i.ytimg.com/vi/nYxFiNosm0g/mqdefault.jpg)
![Ep 23. Wireless Localization and Sensing (With Henk Wymeersch) [Wireless Future Podcast]](/img/tr.png)
![Ep 21. Wireless Coverage Without Beamforming [Wireless Future Podcast]](http://i.ytimg.com/vi/-V3tQQ8ogLg/mqdefault.jpg)
![Ep 6. Q/A on Massive MIMO [Wireless Future Podcast]](/img/n.gif)
Thanks Prof Emil. I can't stop leaning from you.
Thanks for the video
1- @2:39 you mentioned we can measure radio wave that are on the way to travel far away but from my understanding
in the near field the power is reactive meaning imaginary not real so how can we measure that in practice or what do
we actually measure ?
2- @3:21 you mentioned something about the radiative near field , it was mentioned that the measurements of the
signal strength would vary with angle and distance may i know where is the mathematical formulation for this statement ?
bec i found in Balanis book antenna theory analysis and design 2016 eq 4.23 that the radial component of the power
is independent of the distance r
3-when we talk about different regions for ex as fraunhofer region in the context of antenna arrays how is the
limit defined ? i mean we can draw several lines from the each Tx antenna to the Rx with different lengths
4-may i know how the signal properties (amplitude - power) would vary if we are talking about spherical wave vs plane wave ?
5-may i know how would the beamforming problem be different when we are at the near field ? i mean what should we consider
in this case that was ignored in the case of the far field ?
6-if i have an expression for the electric field from maxwell equations , what should be done to get an expression
or formulation for the complex channel ?
7- @29:55 it was mentioned that the antenna distance of lambda/2 is the critical distance for antenna coupling, may i know
A-what is antenna coupling ?
B-why is it something bad ?
C-how does it affect MIMO capacity ?
D-when does it happen ?
E-how to avoid it ?
F-why lambda/2 was set as a threshold for mutual coupling ?
8- @41:22 it was mentioned that with spherical wavefront we can send multiple streams from single antenna Tx to multi
antenna Rx in LOS channel model and still receive them , can i know this is physically possible ? i mean if Tx sends
s=x1+x2+x3 in spherical wave how can the multi antenna Rx resolve the interference ?
10- @45:04 it was mentioned that when number of antennas increases the beamwidth decreases proportional to M but this means
that we can move into the beam and get out of it in a time less than the coherence time Tc of the channel thus we cannot
do the channel estimation with periodicity of Tc, you mentioned that that beamwidth decreases but the fraounhfer
distance increases so this do the needed equalization my questions are
A-The fraunhofer distance is 2*d*M^2/lambda where lambda is the wavelength and d is the distance between the element in
the array if we do the multiplication we would get something that scales with M not one right ?
B-why do we need to multiply the fraunhofer distance by the beamwidth in the first place ?
C-i read this paper here arxiv.org/abs/1511.02937 and they clearly states that the coherent time of the beam
alignment each beam coherent time not channel coherence time how would this impact the answer to the question above
11- @45:05
do we need to do channel estimation after each beam alignment ?
12- @46:42 you mentioned another aspect regarding the beamforming in the near field in which the beamwidth would
be constant and it cannot be smaller than the wavelength
A-may i know why would it be constant
B-why it cannot be smaller than the wavelength
C-where are the proofs of all of that ?
D-@47:33 why when we move wavelength divided by 8 we would lose 1/2 of power ?
13- another question in dipole antenna may i know why the lengths of dipole is lambda/2 or lambda/4 ?
1. The electrical field induces a current in the receive antenna that is then measured. We are always using complex numbers when analyzing electrical fields, but the imaginary part should be viewed as a phase-shift.
2. See the paper that we link to in the description. The main thing is whether the distance from any point on the transmitter to the measurement location is approximately the same (when measuring amplitude and phase) or not; that is, if the transmitter can be viewed as a point-source or not.
3. You can read about this in the paper as well. We will also release a more technical video in the coming weeks.
4. Waves are always spherical, but what matters here is whether the part of the wave that reaches the receive antenna "look" plane or not. More precisely, if there are phase and/or amplitude variations over the receive antenna (apart from those caused by the incident angle) or not.
5. We cannot use transmission/reception schemes that are designed for plane waves (i.e., based on array response vectors). Moreover, the shape of the "beam" around the focal point will look very different. See the figure in the background of the video for an example. The paper in the description gives more precise illustrations.
6. I don't understand the question. One can characterize everything from Maxwell's equations, but we generally make approximations to get more intuitive expressions. When moving from the reactive near-field to the radiative near-field to the far-field, we can use more and more approximations.
7. The simplest way to describe it is that mutual coupling occurs when antenna 2 is in the reactive near-field of antenna 1, so that they influence each other. I cannot explain this in detail in the comments field, but it basically means that the signals "leak" between the antennas and therefore the radiated waveform will not be the sum of the individual waveforms that the antennas would have generated in isolation.
8. You will have to transmit the signals in slightly different angular directions. Angel Lozano wrote a nice piece about this here: www.comsoc.org/publications/ctn/old-theory-new-tricks
10. When you beamform at a location in the far-field, then the beam doesn't appear until you reach a distance proportional to the Fraunhofer distance. This is illustrated in the paper mentioned in the description. If d_FA is the Fraunhofer array distance, then the angular beamwidth scales as 1/sqrt(d_FA). Hence, at the point where the beam begins the physical beamwidth grows as sqrt(d_FA). See equation (33) and set F=d_FA.
11. You need to do channel estimation first and then you can compute the optimal beam.
12. The calculation about the minimum beamwidth is provided on Page 4 in this paper: arxiv.org/pdf/1803.11023
13. There are two types of dipole antennas having these two different lengths.
In summary, I recommend you read the paper that we link to in the description and in my reply. In coming weeks, we will also release a technical video called "Physically Large Antenna Arrays: When the Near-Field Becomes Far-Reaching".
Thank you for the information. Most recently wondered if in 5G mMTC application there are SIM cards in the devices. It sounds a bit unfitting for these tasks. Do you know that?
i’m fascinated by these videos🙏😎.. & Ur question .. yet ignorant of antenna tech ..but.. last 2 yrs my iPhone emitting bio magnetic pulses.. same smart watch gets Ur pulse & Heart EKG frequencies ..
The network operator needs to know which devices belong to what customers, which is one of the purposes having a SIM card. However, you don't need a physical card for achieving that. Many compact devices are using eSIM these days (e.g., Apple Watch) which is an embedded circuit that can store the same information as a physical SIM card. I believe this is what will be used in mMTC applications.
Thank you. How many maximal beams are possible with an array? I think in one video you said that theoretically infinite beams are. However, I read something about min(m,n) where m = number of transmit antennas and n = number of receive antennas. We speak about digital beamforming.
The maximum number of beams that can be transmitted simultaneously is infinite, but one cannot separate the signals from so many beams. The maximum number of separable beams is min(m,n) as you are pointing out. The intuition behind this number is quite easy: You cannot identify more than one signal per receive antenna and you shouldn’t transmit more signals than you have transmit antennas because then the beams become too wide and partially overlapping.
@@WirelessFuture Thank you, but when on has only a few antennas (e.g. 2), there can only a few beams. Is that correct?
One has only limited number of parameters to adjust (1 x phase per antenna and 1x amplitude per antenna).
@@jasminnadic2103 With digital beamforming, you can design the signal that is emitted from each antenna in any way you like at each antenna. It can be the summation of many signals with different amplitudes and phases. But the beams that you can create with two antennas will be wide (the width is roughly proportional to 1/"number of antennas"), so you can only create two non-overlapping beams. This is usually what matters in practice: Don't transmit more beams that you have antennas, and preferably have more antennas than beam so that there is some natural separation between the beams. The fact that one can transmit an infinite number of beams is more a philosophical thing.
I am not sure if it makes so much sense to use DNN or NN in general as Erik suggested. I guess sth that is being ignored here is that for applying most of ML models (RL is an expectation), one needs to collect sufficient data. And this data (or fingerprints) changes based on the location as Erik mentioned again (so many objects in the room), so I guess a paper using DNN for such purpose might end up in the same place with those who ignored coupling :D
Thank you for the information. One question, please. With respect to multiplexing in 5G. 5G uses OFDMA with a flexible numerology that is clear.
Do the different users then get different subcarriers, respectively, for the whole time? So that they use a special bandwidth exclusively, or is there a additional a TDMA and within timeslots they share the bandwidth?
Many thanks.
5G is packet-switched so the operation focuses on making sure that limited-sized packets of data are delivered as soon as possible. Each user is assigned some of the subcarriers for the time it takes to transmit the current packet.
Importantly, multiple users can be assigned to the same subcarrier at the same time. They are then separated using beamforming instead. This is what Massive MIMO is all about.
Thank you, as always it was an interesting episode. In the far field one can find the equation E=1/r and H=1/r and S=1/(r^2) -> There isn't any dependance on the frequency (wavelength). However, one says that transmission depends heavily on the frequency and in the FRIIS Equation there is the frequency (wavelength). To me it looks contradictionary. Have you spoken about that and I have missed it? Thank you very much.
The frequency dependence in Friis equation is due to the size of the receive antenna, which shrinks with the wavelength when considering an isotropic antennas (or any other fixed-gain antenna). We talk about this in the episode about mmWave communications as well as in the episode with Tom Marzetta. I also recommend the following blog post: ma-mimo.ellintech.se/2019/10/29/is-the-pathloss-larger-at-mmwave-frequencies/