Modelling mmWave MIMO Channels

[@MThe3]

Hello Dr. Iain, I am a huge fan of your short videos. They are extremely helpful and right-to-the-point. For that I thank you so much and extend my appreciation to you. I would love to see a series of videos on OFDM, comb/staggering patterns and multiplexing of different users/reference signals in the time-frequency grid.

[@iain_explains]

Thanks. Glad you like the videos. I've added your suggestion to my "to do" list.

[@bk_space]

Dr. Iain, thank you so much! Your explanation is perfectly understandable even for students from Russia like me)

[@iain_explains]

Glad it was helpful!

[@nithinbabu4962]

Hello Professor, Many thanks for the exciting video! I have the following questions: a) Is the model still valid if the channel between the base station and a stationary user is a line-of-sight channel? How to model \gamma_l of ho_{s,l}? Is it dependent on the distance between the BS and the user? Finally, how to obtain the value of ho_{s,l} for s=1 inorder to do a simulation?

[@xinliu5261]

Thanks, Dr. Very helpful video. I have some questions about the model described above, which are as follows.1)why was this mmWave MIMO channel, which contains the AoA and AoD vectors, not modelled like that in MIMO or Massive MIMO channel in previous talks. 2) Is the multipath here referred to that in the frequency selective fading channel or just some multipath used to construct the channel, which is still a flat fading? 3)I am confused about the channel model description of the ray-based MIMO model, statistical model, and multipath model, which is the sum of H. I would appreciate it if you can give explanations for that.

[@iain_explains]

Answers: 1) Different models have different accuracies and are easier/harder to analyse. It tends to be that some are good for accuracy, and others are easier to analyse. In the case of mmWave, there need to be lots of antennas, and this leads to narrow beams, which means they can be directed along the AoA and AoD. With fewer antennas the AoA and AoD cannot be resolved. 2) The assumption is that all the paths are fairly similar in delay, since they are short distances in mmWave comms, so it is a flat fading assumption. But some mmWave channels may be frequency selective 3) I'll think about making a video to try to describe the relationships.

[@cesesr8825]

Thanks Dr. Iain. Could you please do a dedicated presentation on Jake's and Clarke's models giving theoretical basis of using these models.

[@iain_explains]

Thanks for the suggestion. I've put that topic on my "to do" list.

[@zizi7324]

Thank you so much for the clear explanation. I would appreciate it if you can have a video about mmwave channel model vs. Terahertz channel model. How are they different?

[@iain_explains]

Thanks for the topic suggestion. I've got it on my "to do" list.

[@mastermmop]

For evaluating the performance of beam tracking algorithms in mmWave, the AoD and AoA are not assumed to be constant. Is there a model how the AoD and AoA change? That change, just like the change of the rho, should also depend on the speed of the user right? Higher speed -> higher change in AoD and AoA. Is there a model that considers the change of rho, AoD, and AoA in dependence of the speed of the user?

[@ali_abdi]

Thank you for the great explanation. I wonder what is the difference with the Saleh-Valenzuela channel model? can we use the SV model for mmWave? Is it possible for you to also add a description video for the SV channel model? Thank you and have a nice day :)

[@iain_explains]

Thanks for the topic suggestion. I've added it to my "to do" list.

[@dimitrisv.1729]

Very helpful video!! Expected result since coherence time is proportional to λ and one of the drawbacks of mmWave is the need for more frequent channel estimation. For smaller values of τ_s, the value of α would have smaller fluctuations. Under which assumptions is "Gauss-Markov model" valid?

[@iain_explains]

It depends on what you mean by "valid". It's a model, so it approximates reality. It's really more of a question of whether it is an accurate approximation to reality, or not. I guess you could say that if the approximation is accurate, then the model is "valid". From measurements that have been reported in the literature, the Gauss-Markov model for the channel path gain is accurate for sub-2 GHz channels, and appears to be accurate at higher frequencies too, but there are limited measurement sets at those frequencies.

[@bobbaberson3654]

Can you please add a video on Hybrid mmWave precoding to this series?

[@iain_explains]

Thanks for the suggestion. I'll put it on my "to do" list.

[@aquinocbe]

Hello Dr lain, I am a huge fan of your short videos. Can you please help me with the coding part of channel modelling and direction of arrival estimate in mMIMO mmwave in MATLAB.

[@iain_explains]

Thanks for the suggestion of adding Matlab code. It's something I've been thinking about for a while. I just need to find time to do it.

[@pangpangliu4461]

Sir. I have a doubt. In the video, why do you write ${\bf a}^H$? Does the ${\bf a}^H$ refer to the steering vector of the transmitter/base station? or does ${\bf a}$ refer to the steering vector of the transmitter/base station? Thx.

[@iain_explains]

${\bf a}^H(theta_ell)$ is the channel gain in the direction of the "ell"-th scatterer, that is in the direction theta with respect to the transmitter array.

[@nicolasperez4292]

could you explain the vectors in more detail? Or give a link to a source that does?

[@iain_explains]

The vectors contain the phase shifts that relate to the angle of departure and arrival. Hopefully this video will help: "What is Beamforming?" kzbin.info/www/bejne/d2LRZnuerNmtbZo

[@鍾明憲-l6j]

good

[@abdulrahmanalayidh5752]

Thank you Dr for this amazing video. I have one question about carrier frequency and bandwidth. Some papers assumed that the carrier frequency is 28GHz and the bandwidth is 100 MHz. Is 100 MHz enough when we consider mmWave with madsive MIMO?

[@iain_explains]

Lots of factors go into the choice of channel bandwidth allocation, so your question does not have a simple answer. Certainly 100 MHz is much wider than current mobile and WiFi standards at sub-mm-wave carrier frequencies, which are typically 10 or 20 MHz.

[@uriartaramadhani7747]

Hi Dr. Iain, I am very thankful for your videos. They are very helpful! I have a question. I am working in a more simple environment where there's only single antenna source and destination. Using non millimeter wave, can I still use the channel model in this video? But only considering one ray/one beam

[@iain_explains]

If you only have one antenna, then you don't have beams (unless it is a dish antenna, in which case there is a single beam in the direction of the dish). For single antennas, with mobile users, you'll tend to have Rayleigh fading. See: "What is Rayleigh Fading?" kzbin.info/www/bejne/Y3eyn4x4j5yNnLM

[@uriartaramadhani7747]

@@iain_explains Thank you for your response, Dr.!

[@wangyehui4878]

Hello professor, I want to know how can I show large-scale effects in modeling, for example, path loss and shadowing. Can I simple add another variable represent gain in the sum equation?

[@iain_explains]

Yes, that's right. The important thing is the signal to noise ratio. The actual receive power is important for designing the front-end filters and amplifiers, but in terms of performance, it's only the SNR that counts.

[@wangyehui4878]

Thank you very much for your reply! Regarding the SNR you mentioned, can I understand it this way: fine channel modeling is not the most important thing, analyzing the SNR is the most important thing. But I don’t quite understand the connection between channel modeling and analyzing the SNR, for example, how should I analyze the SNR based on the formula Hs​(τ)=∑l=1L​ρs,l​b(ϕl​)aH(θl​)δ(τ−τl​) in your video? My research direction is not wireless communication, but I need to establish a wireless communication model. I am not clear about many common sense in the field of wireless communication. Your video and reply are very helpful to me, thank you very much.@@iain_explains

[@abdulrafay7154]

Thanks, Dr. It's a very useful and tremendous video. I just want to ask one question if one train moves in a one-dimensional way with a speed of 300kmph and communicating to mmWave based remote antenna units (RAUs) like 60-GHz means. In this scenario, RAUs are installed in fixed positions, and the direction (position) of the antenna installed on the top of the train is also fixed. And the distance between the train's antenna and RAU is almost 5 meters. So now how I will calculate channel gain by using your approach as you discussed in your lecture related to the alpha (Velocity/Speed, frequency). Please let me know about this. I shall be very thankful to you.

[@iain_explains]

In the case you've described, the channel is likely to be line-of-sight (LoS), with not many reflectors, so there won't be significant multipath. The LoS component will also be largely predictable since the train is moving on fixed tracks in a predictable way. Also it depends on the types of antennas being used. If they are directional dish-type fixed antennas, then the channel model will be quite different to what's described in this video.

[@abdulrafay7154]

@@iain_explains Thanks Dr. for your reply. Dear Dr. In my case which channel models will be best suitable in the LOS scenario. Thanks.

[@abdulrafay7154]

If I use directional antenna like horn antenna.

[@iain_explains]

There's no easy answer for this. For one thing, it depends on how accurately the horn antennas can be rotated to face each other as the train moves along.

[@comedyzonetube]

Hi all, Is there anyone who help me with the performance analysis of the adaptive beamforming-based algorithm for interference mitigation in the mm-wave massive Mimo system.

[@mopol4400]

hello Mr what is spatial-channel-correlation?? THX

[@iain_explains]

I think this video will help: "Statistical Modelling of MIMO Communication Channels" kzbin.info/www/bejne/h2Sbk3uorsyDkMk