Thanks prof.Emil for your brilliant explanation. I have some questions. Why the capacity with same SNR applied Rayleigh fading channel model(with CN(0,1)) can almost be greater than the capacity applied the estimated channel in real-world scenario? As a result, the capacity applied Rayleigh fading channel model can be seen as the upper bound of spatial multiplexing MIMO? I'm just wondering if it is possible to get larger capacity in channel of real-world scenario than Rayleigh fading channel model.
The best channel matrix from a spatial multiplexing perspective is a (semi-)unitary matrix; that is, all its singular values are equal so one can multiplex many signals that experience equally good channel conditions. An iid Rayleigh fading channel is likely to give you a channel where all singular values are of decent size, but they won't be equally large. One can deploy antennas in real-world line-of-sight scenarios to get a "perfect" channel matrix. Section 4.4.3 in my recent book describes how to do that by carefully selecting the antenna locations: www.nowpublishers.com/article/BookDetails/9781638283140
Sir, the code for Capacity Limits and Multiplexing Gains of MIMO Channels with Transceiver Impairments paper, your honor, you published it on a website, I took it, but it no work, for example, the code waterfilling and the code figure 2, if you allow me to give the code exactly?
The code on GitHub is precisely the code that was used to generate the figures in the paper. I just tried to generate Figure 2 and it worked. What error message did you get?
The standard deviation isn't related to interference (which are other signals), but the average attenuation that the channel creates. The variance is typically -70 dB (near transmitter) to -130 dB (1 kilometer away or more). You can have a look at the following video: ruclips.net/video/OA4viERrlzA/видео.html
@@WirelessFutureSir my professor asked me that... For highly dense area, what value of variance should we take for modelling the channel parameter relative to less dense area. He gave me two values 1 and 3
@@mdraqibkhan8785 One can never receive more power than was transmitted so the attenuation will always be much smaller than 1. However, it is possible to normalize variables to get different scales. The important thing is the SNR makes sense for the setup at hand.
thank you prof , in order to work in WISP who provide 4G LTE and in not distant future 5G what is the best books to read to fully understand wireless world (wave , frequency , modulation , MIMO , P2P microwave links) if you already have a good background about optic network thanks
This is a hard question to answer. It depends a lot on what kind of book you are looking for. Should it be theoretic and scientifically rigor? Should it be explaining the features of the 4G and 5G standards on a practical level? Should it provide basic insights, without deep theory? One suggestion is "Wireless Connectivity: An Intuitive and Fundamental Guide" by Petar Popovski. It covers a large number of topics and provide the core properties in a fairly simple manner. So I would recommend reading it first.
The videos of the course TSKS14 Multiple Antenna Communications are covering most of the book "Fundamentals of Massive MIMO" but not everything. The book "Massive MIMO Networks" is going much deeper into the theory and only a few parts of it are covered by videos.
The pathloss between each transmit antenna and each receive antenna can be modeled in the same way as in single-antenna systems. Typical numbers in wireless communications are -70 dB to -150 dB. A classical pathloss model is the Hata Model: en.wikipedia.org/wiki/Hata_model
@@WirelessFuture if I want to simulate channel estimation so in channel model I would use H=Ar*(capital lambda)*At^H Where (capital lambda) depends on pathloss so for sparse channel what value should I choose
You can compute the pathloss for each diagonal element as in a SISO system, as long as you make sure that each column of At and Ar have a squared norm equal to the number of antennas that exist at that side.
The more advanced/refined systems we want to build, the more detailed and accurate the fading models must become. The problem is that one can either build a model that is accurate for one particular setup but not for other setups, or statistical models that are not entirely accurate but gives you the possibility to quickly simulate many different setups. So the answer is that no model will be accurate for all purposes. There are general models such as spatially correlated Rayleigh fading that are fairly good, as well as 3GPP models. But we always need to make measurements in different setups to find suitable model parameters. The more advanced models are used, the more parameters need to be selected, and the more measurements are needed...
![Slow Fading and Outage Probability [Video 9]](http://i.ytimg.com/vi/9ceFAkafBNw/mqdefault.jpg)
![Slow Fading and Outage Probability [Video 9]](/img/tr.png)
![Capacity of Point-to-point SIMO and MISO Channels [Video 5]](http://i.ytimg.com/vi/WJ1e_0sIJRk/mqdefault.jpg)
Can't thank this person enough! I've been learning from him for a decade and always find profound remarks in what he says. God bless you, Prof. Emil
Thanks prof.Emil for this amazing tutorial and valuable presentation.
Thanks professor for this great presentation!
Thanks prof.Emil for your brilliant explanation.
I have some questions.
Why the capacity with same SNR applied Rayleigh fading channel model(with CN(0,1)) can almost be greater than the capacity applied the estimated channel in real-world scenario?
As a result, the capacity applied Rayleigh fading channel model can be seen as the upper bound of spatial multiplexing MIMO?
I'm just wondering if it is possible to get larger capacity in channel of real-world scenario than Rayleigh fading channel model.
The best channel matrix from a spatial multiplexing perspective is a (semi-)unitary matrix; that is, all its singular values are equal so one can multiplex many signals that experience equally good channel conditions. An iid Rayleigh fading channel is likely to give you a channel where all singular values are of decent size, but they won't be equally large. One can deploy antennas in real-world line-of-sight scenarios to get a "perfect" channel matrix. Section 4.4.3 in my recent book describes how to do that by carefully selecting the antenna locations: www.nowpublishers.com/article/BookDetails/9781638283140
@@WirelessFuture Thank you for your reply. I sincerely appreciate it.
Sir, the code for Capacity Limits and Multiplexing Gains of MIMO Channels with Transceiver Impairments paper, your honor, you published it on a website, I took it, but it no work, for example, the code waterfilling and the code figure 2, if you allow me to give the code exactly?
The code on GitHub is precisely the code that was used to generate the figures in the paper. I just tried to generate Figure 2 and it worked. What error message did you get?
sir what should be the value of standard deviation in Rayleigh distribution for highly interference area .
high value or small ?
The standard deviation isn't related to interference (which are other signals), but the average attenuation that the channel creates. The variance is typically -70 dB (near transmitter) to -130 dB (1 kilometer away or more). You can have a look at the following video: ruclips.net/video/OA4viERrlzA/видео.html
@@WirelessFutureSir my professor asked me that...
For highly dense area, what value of variance should we take for modelling the channel parameter relative to less dense area. He gave me two values 1 and 3
@@mdraqibkhan8785 One can never receive more power than was transmitted so the attenuation will always be much smaller than 1. However, it is possible to normalize variables to get different scales. The important thing is the SNR makes sense for the setup at hand.
@@WirelessFuture Ok I got it, thanks sir
thank you prof , in order to work in WISP who provide 4G LTE and in not distant future 5G what is the best books to read to fully understand wireless world (wave , frequency , modulation , MIMO , P2P microwave links) if you already have a good background about optic network
thanks
This is a hard question to answer. It depends a lot on what kind of book you are looking for. Should it be theoretic and scientifically rigor? Should it be explaining the features of the 4G and 5G standards on a practical level? Should it provide basic insights, without deep theory?
One suggestion is "Wireless Connectivity: An Intuitive and Fundamental Guide" by Petar Popovski. It covers a large number of topics and provide the core properties in a fairly simple manner. So I would recommend reading it first.
@@WirelessFuture thank you , i mean practical level i appreciate your suggestion and i will read it soon
Dr Emil you explain the book"massive mimo "all the topic by RUclips or missing any topic please respond to me this can help me in my research
The videos of the course TSKS14 Multiple Antenna Communications are covering most of the book "Fundamentals of Massive MIMO" but not everything.
The book "Massive MIMO Networks" is going much deeper into the theory and only a few parts of it are covered by videos.
What should be the pathloss in case of massive mimo channel
The pathloss between each transmit antenna and each receive antenna can be modeled in the same way as in single-antenna systems. Typical numbers in wireless communications are -70 dB to -150 dB. A classical pathloss model is the Hata Model: en.wikipedia.org/wiki/Hata_model
@@WirelessFuture would it be same for sparse channel?
@@WirelessFuture if I want to simulate channel estimation so in channel model I would use H=Ar*(capital lambda)*At^H
Where (capital lambda) depends on pathloss so for sparse channel what value should I choose
You can compute the pathloss for each diagonal element as in a SISO system, as long as you make sure that each column of At and Ar have a squared norm equal to the number of antennas that exist at that side.
which is the most accurate fading model in context to 6G?
The more advanced/refined systems we want to build, the more detailed and accurate the fading models must become. The problem is that one can either build a model that is accurate for one particular setup but not for other setups, or statistical models that are not entirely accurate but gives you the possibility to quickly simulate many different setups. So the answer is that no model will be accurate for all purposes. There are general models such as spatially correlated Rayleigh fading that are fairly good, as well as 3GPP models. But we always need to make measurements in different setups to find suitable model parameters. The more advanced models are used, the more parameters need to be selected, and the more measurements are needed...
@@WirelessFuture Thank you. What about the fading model in industrial IOT network?