Your explanation is right to the point and include lots of subtle point. I hope that you could so something about parallel inteference cancellation in NOMA. Thank you !
Thank you sir, what a clear explaination! I did some research and realized that in addition to the power domain, Noma also has a code domain. Hope you make a video to explain more about the code domain.
Glad it was helpful. You can find a full categorised list of my videos at iaincollings.com including "What is Massive MIMO?" th-cam.com/video/39CZyDq3vSw/w-d-xo.html
Thanks for explaining NOMA simply. NOMA makes sense if compared against TDMA (i.e., shared resources), but can we compare it with SDMA? Defenetly no, because with SDMA the users are separated in space via beamforming, where each user has its own direction relative to the BTS. Thus, we can assign the whole system resources (time and bandwidth for example) to each user in the coverage area simultaneously. So no need for SIC at all, especially if large number of antennas is used (massive MIMO).
Yes, you're right, SDMA divides the users spatially, but it's still possible to have multiple users in the same beam direction sharing a beam. But as you say, the beams can be quite narrow at the higher carrier frequencies.
What a clear explaination sir! I have one question: What is your opinion about the possibility of NOMA applying in the 5G and beyond systems? In the case where NOMA is combined with antenna techiniques such as massive MIMO.
Добрый день! Как вы думаете, насчет использование метода NOMA в системах сотовой связи и телевидения? Какой метод может быть общим для систем сотовой связи и телевидения?
Thanks for the great video. Definitely helped me get insight into NOMA where from research papers I couldn’t get it. One question though: In the equation for y1 where x2 hat is subtracted, how do you know the power p2 magnitude with which to multiply in order for the x2 components to cancel? You say that channel information has to be known at all times and I get there are methods for channel estimation to be used for that. But even then, using the wrong power magnitude would mess up the whole method. So again, how would receiver 1 know p2? I can’t imagine a method to estimate it. If RSSI where it where part of the demodulation process, it would always give the total power from p1 and p2. Only the base station could send the information of how much of a fraction p2 is with regard of it’s total power output in an orthonormal approach.
Great question. Channel estimation is done using different resources, for example at a different time, using training sequences, and then the channel estimates are communicated to the users using a control channel (often at a low rate on a different carrier frequency).
@@iain_explains Thanks. So I'll assume the information is passed with the training sequence. Thanks for the quick reply! And thanks again for the great video.
Thanks Lain for the nice explanation. But why do we decode the user with the smaller rate and not the user with the higher SINR regardless of the rate? Or (probably) both the rate and SINR dictate which use to decode first? Am I missing something
You want to decode the one that's easiest to decode first. This is not simply a function of the SINR - it also depends on the rate. For example, a BPSK signal is much easier to detect and decode than a 256-QAM signal. These videos might help: "What is Successive Interference Cancellation (SIC)?" th-cam.com/video/xFaRMnzxX4I/w-d-xo.html and "Successive Interference Cancellation (SIC) with Unequal Power Users" th-cam.com/video/_1RfL-Vsg_Y/w-d-xo.html
hi sir, thank you for the perfect explanation. I have just one question. Can't we use the same constellation for both users in NOMA? for example using BPSK for both?
Hi professor Iain, thanks for the video. Great great explanations. I am wondering about is there any security issue when y1 performs successive interference cancellations by subtracting the decoded version of h1p2x2 ? Thank you.
Great question. Don't forget though, in multi-access wireless channels everyone hears everyone else's signals, whether it's TDMA, FDMA, NOMA, ... whatever. Yes, NOMA actually involves detecting/decoding other users' signals, in order to help with the detection/decoding of your own signal. The other multi-access schemes don't do this, but it doesn't mean they couldn't do it if they wanted to. Don't forget though, just because you might be able to detect another user's symbols, and decode their forward-error-correction code, it doesn't mean you will know what the data means, as long as they are using encryption on their data stream.
Hi Iain. Thanks for good explanation of the NOMA concept. I have one question (just to confirm my understanding): shall BTS apply different modulation scheme for different user (etc. QPSK for user with lower channel gain, and 64QAM for user with better channel gain) in order to allow receiver on user side to detect and decode signals from other users?
It is not necessary for the users to have different modulation schemes, since the signal(s) for/from the other user(s) are either treated as noise, or are fully decoded and removed/cancelled (depending on their place in the sequence). For each user, the modulation scheme and the error-control-coding scheme should be matched to the SINR, to maximise the achievable rate. In other words it is not a "joint" detection scheme.
Hello Lain,. I have a question which I put hère as it might be related. 1) Rx1 in your example first decodes RX2's signal. To do that, user 2's rate should be such that log2(1+|p2x2h1|sq/|p1x1h1|sq+sigma1 sq) is larger than that, no? 2) After decoding user 2's rate, user 1 does not see any interference. There fore, to enable user 1 to decode its signal without error, is not it possible to encode user 1's signal upto log2(1+SNR) instead of SINR, no?
Hi sir, thank you so much for the great explanations. Your video save me a lot of time in studying NOMA in power domain. Do you any videos for explaining about coding domain of NOMA? Some kinds like Sparse Code Multiple Access (SCMA), LDS OFDM, etc. I am really grateful for your videos.
Great question. It's on my "to do" list to make a video about it. I've just moved it up the priority order. Briefly though, modulation waveforms can be represented (at baseband) as vectors in cartesian coordinates. The locations of the ends of the vectors, for a given modulation format, are called "constellation points". I'm not sure if that helped or not, but keep an eye out for a video coming up on the channel.
I have a question in the tdma scenario. We turn on P in the first time slot and then we turn on P in the second time slot. Does that mean the total amount of power in the first and second time slot is 2P. If yes then this give me a strange impression that tdma use double the power but offer less sum rate then NOMA ?
I think you're mixing up Power and Energy. Don't forget, Power is instantaneous. Energy is the integral of Power over time. If the power is constant, then E = P t. So, if you send at a given power level for half the time, then you are only sending half the energy.
Yes, certainly. Each subcarrier can be treated separately and use successive interference cancellation (ie. NOMA). This is similar to using MU-MIMO on a per-subcarrier basis.
Consultant, technical documents writer, tutor, inventor, researcher, or entrepreneur, which is the best if can come up with a practical technical solution for a serious problem in wireless.
Although the users have different powers, they are not "divided" in power. TDMA divides users into different (orthogonal) time slots. FDMA divides users into different (orthogonal) frequency channels. CDMA divides users using different (orthogonal) spreading codes.
Thanks for your efforts and care For the second scenario if we allocated more power to the high gain user then this will not achieve fairness bec from the beginning we could have allocated all power to the high gain user all the time I mean in Noma we can not reverse the order for power sharing ? Other thing how can user of stronger channel gain know the channel of the weak user ? I believe the base station has access to this info but is it practical to convey it to the strong user ? I appreciate the time and efforts you are trying to put in the videos but your videos make it rather more interesting to find some sources that can can elaborate more could you please share the research papers books lectures that you found interesting let us start for example with the materials used for this video preparation ?
It depends on what you mean by fairness. If the low gain user is an IoT device that only needs small packets of data at a low rate, and the high gain user requires a live video stream, then the high gain user needs more connection resources than the low gain user. The "fair" thing would be to give them both the same _proportion_ of their requested throughput. This can be done by appropriate allocation of power between the two. It certainly wouldn't be "fair" to simply allocate power on the basis of channel gain alone. In terms of your question about how the users can know each others channel gains. Yes, that's exactly what I pointed out near the end of the video. It is an important question. The communication of the channel state information is done over control channels/packets, but the accuracy and timelines requirements will limit the application of NOMA in fast fading environments. In terms of the references for more details, this topic is still new, and the references are mostly technical research papers. If you've got specific questions, then please do ask them in the comments, and I'll try to answer them with another video.
@@iain_explains For high gain user it would be enough and "fair" to allocate smaller power ratio and his channel gain would be helping him on his data rate demand As said we could have scheduled him alone with all its power needs from the beginning without giving anything to the low channel gain user The low gain user treat interference as noise now if you allocated less power to him and more power to its interferer this shall not be fair I believe since the low channel gain user now has less channel bgain and more interference thus its sinr would be severely degraded and it's rate wouldn't be greater than the oma case as you plotted Regarding the references can you for example mention the reference where you have seen that it is the case the we can allocate more power to higher rate user ?
OK, I'm not sure how to respond. I'm sorry if you don't believe me, but I don't know what else to say. I'm not going to debate it on this discussion forum. I am a Fellow of the IEEE, with over 25 years of research track record in wireless communications, so I'm comfortable with the accuracy of what I said above, and in the video. Thanks for your interest in my videos, I really do appreciate your passion for the topic.
![[#2024MAMA] ROSÉ (로제), Bruno Mars - APT. | Mnet 241122 방송](http://i.ytimg.com/vi/dgGqD28J6aQ/mqdefault.jpg)
![BOWKYLION - วิงวอน (ex-change) [Official MV]](http://i.ytimg.com/vi/cLdnZWDaO-w/mqdefault.jpg)
Great explanation.I have been searching for a video on NOMA since 2017.Thanks
Glad it was helpful!
Your explanation is right to the point and include lots of subtle point. I hope that you could so something about parallel inteference cancellation in NOMA. Thank you !
Thanks, yes that topic is on my to-do list (but it's getting to be quite a long list).
Thank you sir, what a clear explaination! I did some research and realized that in addition to the power domain, Noma also has a code domain. Hope you make a video to explain more about the code domain.
Thanks for the suggestion. I'll add it to my "to do" list.
Thank you for the explanation. Liked and subscribed.
Looking for massive Mimo explanation.
Glad it was helpful. You can find a full categorised list of my videos at iaincollings.com including "What is Massive MIMO?" th-cam.com/video/39CZyDq3vSw/w-d-xo.html
Thanks for explaining NOMA simply.
NOMA makes sense if compared against TDMA (i.e., shared resources), but can we compare it with SDMA? Defenetly no, because with SDMA the users are separated in space via beamforming, where each user has its own direction relative to the BTS. Thus, we can assign the whole system resources (time and bandwidth for example) to each user in the coverage area simultaneously. So no need for SIC at all, especially if large number of antennas is used (massive MIMO).
Yes, you're right, SDMA divides the users spatially, but it's still possible to have multiple users in the same beam direction sharing a beam. But as you say, the beams can be quite narrow at the higher carrier frequencies.
Great video. Could you give a lecture talking about tensor algorithms used wireless communication systems?
Thanks for the topic suggestion. I've added it to my "to do" list.
Great Video Sir!
Can you please explain RSMA in any upcoming videos??
Thanks
Thanks for the suggestion. It's on my "to do" list (... it's a long list).
Thank you sir. Great explanation.
Glad you liked it
Hi sir, thank you so much. It was a great explanations
Glad you liked it
What a clear explaination sir! I have one question: What is your opinion about the possibility of NOMA applying in the 5G and beyond systems? In the case where NOMA is combined with antenna techiniques such as massive MIMO.
Could you please explain the uplink scenario?
thankyou so much...this is the most clear explanation..
🙂🙂
Glad it was helpful!
excellent explanation! Thank you!
Glad it was helpful!
Добрый день! Как вы думаете, насчет использование метода NOMA в системах сотовой связи и телевидения? Какой метод может быть общим для систем сотовой связи и телевидения?
Thanks for the great video. Definitely helped me get insight into NOMA where from research papers I couldn’t get it. One question though: In the equation for y1 where x2 hat is subtracted, how do you know the power p2 magnitude with which to multiply in order for the x2 components to cancel? You say that channel information has to be known at all times and I get there are methods for channel estimation to be used for that. But even then, using the wrong power magnitude would mess up the whole method. So again, how would receiver 1 know p2? I can’t imagine a method to estimate it. If RSSI where it where part of the demodulation process, it would always give the total power from p1 and p2. Only the base station could send the information of how much of a fraction p2 is with regard of it’s total power output in an orthonormal approach.
Great question. Channel estimation is done using different resources, for example at a different time, using training sequences, and then the channel estimates are communicated to the users using a control channel (often at a low rate on a different carrier frequency).
@@iain_explains Thanks. So I'll assume the information is passed with the training sequence. Thanks for the quick reply! And thanks again for the great video.
Perfect
Subscribed, thanks to this video!!
Welcome aboard!
Modern tech puts Nyquist and Shannon to shame!
Perfect!
Glad you liked it.
I have a question: How does user 1 know that he has to change the type of demodulation e.g. from BPSK to decode x2 to 256-PSK to decode x1 after SIC.
It requires a control channel over which all this sort of information is shared.
Thanks Lain for the nice explanation. But why do we decode the user with the smaller rate and not the user with the higher SINR regardless of the rate? Or (probably) both the rate and SINR dictate which use to decode first? Am I missing something
You want to decode the one that's easiest to decode first. This is not simply a function of the SINR - it also depends on the rate. For example, a BPSK signal is much easier to detect and decode than a 256-QAM signal. These videos might help: "What is Successive Interference Cancellation (SIC)?" th-cam.com/video/xFaRMnzxX4I/w-d-xo.html and "Successive Interference Cancellation (SIC) with Unequal Power Users" th-cam.com/video/_1RfL-Vsg_Y/w-d-xo.html
@@iain_explains thanks a lot... much appreciated
good exp
thank you professor
You're welcome. I'm glad my video has helped.
Sir could you please talk about successive interference cancellation
Thanks for the suggestion. Yes, it's already on my "to-do" list.
hi sir, thank you for the perfect explanation. I have just one question. Can't we use the same constellation for both users in NOMA? for example using BPSK for both?
Yes, sure.
can you do about SWIPT Sir? Simultanous wireless information and power transfer?
Thanks for the suggestion. I've put it on my "to do" list.
u r my guru
Nice one! I'm glad you're finding the videos helpful.
Hi professor Iain, thanks for the video. Great great explanations. I am wondering about is there any security issue when y1 performs successive interference cancellations by subtracting the decoded version of h1p2x2 ? Thank you.
Great question. Don't forget though, in multi-access wireless channels everyone hears everyone else's signals, whether it's TDMA, FDMA, NOMA, ... whatever. Yes, NOMA actually involves detecting/decoding other users' signals, in order to help with the detection/decoding of your own signal. The other multi-access schemes don't do this, but it doesn't mean they couldn't do it if they wanted to. Don't forget though, just because you might be able to detect another user's symbols, and decode their forward-error-correction code, it doesn't mean you will know what the data means, as long as they are using encryption on their data stream.
Thanks for your reply.@@iain_explains
Hi Iain. Thanks for good explanation of the NOMA concept. I have one question (just to confirm my understanding): shall BTS apply different modulation scheme for different user (etc. QPSK for user with lower channel gain, and 64QAM for user with better channel gain) in order to allow receiver on user side to detect and decode signals from other users?
It is not necessary for the users to have different modulation schemes, since the signal(s) for/from the other user(s) are either treated as noise, or are fully decoded and removed/cancelled (depending on their place in the sequence). For each user, the modulation scheme and the error-control-coding scheme should be matched to the SINR, to maximise the achievable rate. In other words it is not a "joint" detection scheme.
Hello Lain,. I have a question which I put hère as it might be related.
1) Rx1 in your example first decodes RX2's signal. To do that, user 2's rate should be such that log2(1+|p2x2h1|sq/|p1x1h1|sq+sigma1 sq) is larger than that, no?
2) After decoding user 2's rate, user 1 does not see any interference. There fore, to enable user 1 to decode its signal without error, is not it possible to encode user 1's signal upto log2(1+SNR) instead of SINR, no?
Did you want to Say SNR ? SINR makes me confused
I think it would be better if you multiply the transmitted symbols x times the square root of the power, not the power itself
Ahhh, yes you're right. I must have had a mind-blank. In the equations for y_1 and y_2, all the p's should be sqrt(p)'s. Thanks for pointing it out.
Hi. Would any of you kindly explain why the square root should be used? What difference does it make? Thanks much 🥰😍
For now, NOMA is adopted for 5G standard?
Please talk about resource allocation in noma
Thanks for the suggestion. I've added it to my "to do" list.
Hi sir, thank you so much for the great explanations. Your video save me a lot of time in studying NOMA in power domain. Do you any videos for explaining about coding domain of NOMA? Some kinds like Sparse Code Multiple Access (SCMA), LDS OFDM, etc. I am really grateful for your videos.
Thanks for those suggestions. I'll give then some thought.
Pardon me if I sound dumb.. but whats a constellation point? Thank you professor for your nice tutorial.
Great question. It's on my "to do" list to make a video about it. I've just moved it up the priority order. Briefly though, modulation waveforms can be represented (at baseband) as vectors in cartesian coordinates. The locations of the ends of the vectors, for a given modulation format, are called "constellation points". I'm not sure if that helped or not, but keep an eye out for a video coming up on the channel.
I have a question in the tdma scenario. We turn on P in the first time slot and then we turn on P in the second time slot. Does that mean the total amount of power in the first and second time slot is 2P.
If yes then this give me a strange impression that tdma use double the power but offer less sum rate then NOMA ?
I think you're mixing up Power and Energy. Don't forget, Power is instantaneous. Energy is the integral of Power over time. If the power is constant, then E = P t. So, if you send at a given power level for half the time, then you are only sending half the energy.
@@iain_explains Thank you very much Professor ! So this is just good old conservation of energy
in OFDM case, can we implement NOMA
?
Yes, certainly. Each subcarrier can be treated separately and use successive interference cancellation (ie. NOMA). This is similar to using MU-MIMO on a per-subcarrier basis.
How do we reallocate power efficiently?
We allocate power based on the distance of users from the base station.
Hi Professor, thank you for your lecture, but in some papers, they use the square root of p, I don't know what is the difference.
I guess you didn't read the note I added below the video. They should all be sqrt(P).
@@iain_explains I'm sorry, next time I will read the note first.
anyone know what jobs to go for with this research background and a EE degree
Consultant, technical documents writer, tutor, inventor, researcher, or entrepreneur, which is the best if can come up with a practical technical solution for a serious problem in wireless.
I need video explanation about rate split multiple access if you can
Thanks for the suggestion. I've put it on my "to do" list.
Can we call NOMA power division multiple access?
Although the users have different powers, they are not "divided" in power. TDMA divides users into different (orthogonal) time slots. FDMA divides users into different (orthogonal) frequency channels. CDMA divides users using different (orthogonal) spreading codes.
@@iain_explains thanks for explanation
Thanks for your efforts and care
For the second scenario if we allocated more power to the high gain user then this will not achieve fairness bec from the beginning we could have allocated all power to the high gain user all the time I mean in Noma we can not reverse the order for power sharing ?
Other thing how can user of stronger channel gain know the channel of the weak user ? I believe the base station has access to this info but is it practical to convey it to the strong user ?
I appreciate the time and efforts you are trying to put in the videos but your videos make it rather more interesting to find some sources that can can elaborate more could you please share the research papers books lectures that you found interesting let us start for example with the materials used for this video preparation ?
It depends on what you mean by fairness. If the low gain user is an IoT device that only needs small packets of data at a low rate, and the high gain user requires a live video stream, then the high gain user needs more connection resources than the low gain user. The "fair" thing would be to give them both the same _proportion_ of their requested throughput. This can be done by appropriate allocation of power between the two. It certainly wouldn't be "fair" to simply allocate power on the basis of channel gain alone.
In terms of your question about how the users can know each others channel gains. Yes, that's exactly what I pointed out near the end of the video. It is an important question. The communication of the channel state information is done over control channels/packets, but the accuracy and timelines requirements will limit the application of NOMA in fast fading environments.
In terms of the references for more details, this topic is still new, and the references are mostly technical research papers. If you've got specific questions, then please do ask them in the comments, and I'll try to answer them with another video.
@@iain_explains
For high gain user it would be enough and "fair" to allocate smaller power ratio and his channel gain would be helping him on his data rate demand
As said we could have scheduled him alone with all its power needs from the beginning without giving anything to the low channel gain user
The low gain user treat interference as noise now if you allocated less power to him and more power to its interferer this shall not be fair I believe since the low channel gain user now has less channel bgain and more interference thus its sinr would be severely degraded and it's rate wouldn't be greater than the oma case as you plotted
Regarding the references can you for example mention the reference where you have seen that it is the case the we can allocate more power to higher rate user ?
OK, I'm not sure how to respond. I'm sorry if you don't believe me, but I don't know what else to say. I'm not going to debate it on this discussion forum. I am a Fellow of the IEEE, with over 25 years of research track record in wireless communications, so I'm comfortable with the accuracy of what I said above, and in the video. Thanks for your interest in my videos, I really do appreciate your passion for the topic.
@@iain_explains Thanks