This professor is absolutely outstanding. He explains this very well. He utilizes the math in his presentations, but doesn't lose me, because he has a good pace and logical explanations. Thank you!!!
I saw a video that if you want to increase your SINR (dB), you should check the antenna of your neighbors. If everybody is tilting up, you should tilt your antenna a bit lower. Thanks for the video.
Does formula of the effective antenna area refers to an antenna length of lambda? The antenna area is dependend on the size of the antenna. Thanks. Or does it refers to a lambda/2 dipole?
The effective area of an isotropic antenna is a hypothetical number, which is explained here: en.wikipedia.org/wiki/Isotropic_radiator For aperture antennas, the effective area is roughly the same as physical area of the antenna. But for a dipole, it is not as easy. If the length is lambda/2, the directivity gain is 2.15 dB, which means that is effective area is 2.15 dB ≈ 1.6 times larger than that of an Isotropic antenna.
Hi Professor. I have a question on the calculation of SINR. Is SINR the power of the useful signal over the power of the sum of the interfering signals plus the noise power, or is it the power of the useful signal over the sum of the power of the interfering signals plus the noise power? The difference is that when we consider multi-user downlink communication, some articles directly state the sum of the interfering signal power, which really confuses me. Looking forward to your answer.
Thanks for the question. Whenever the interfering signals are independent, they appear in the SINR as a summation of their powers: P_useful / ( P1+…PN+noise power) where P1,…,PN are the powers of N interfering signals.
At 2:34, what is called "diameter (d)" is typically referred to as "radius (r)" -- the distance from the center of the sphere to its surface. The formula works based on the diagram drawn, but the term "diameter" is used in a way I've not seen before.
The variable "d" in the video is the distance to the receiver and, yes, it is the radius of the sphere. Was the word "diameter" used anywhere in the video? I couldn't find that.
1)What is meant by transmitted power? At @9:11, transmitted power will be multiplied by antenna gain? Or it will be considered in the definition of channel gain? Although the resultant SNR will be the same in both cases. 2)What will happen for the channel gain in a directional antenna case where the denominator will be the same as the area of the entire sphere? Is such a definition of channel gain based on the assumption of spherical mode expansion only? 3) What will be the SNR when the receiving location does not lie in the far-field region? For example in an indoor scenario.
1) The transmit power is the power of the transmitted signal. The received power is the transmit power multiplied with the channel gain. We consider unit antenna gain in this video, but otherwise I recommend including it in the channel gain. 2) The definition is based on the assumption of an isotropic antenna. Otherwise the numerator will change. 3) There is no strong far-field assumption in this video. As long as we are beyond the reactive near-field, it will be fine. That will be the case also indoors.
It works very well. You preferably estimate the multipath channels in the uplink and utilize the estimate in both uplink and downlink. The transmitted signal will not be look like a beam but exploit all the paths.
@@WirelessFuture Works very well in theory - in practice, not so much. Ask Verizon about their experience rolling out 28 GHz 5G roll. So far, not so good . . .
@@WirelessFuture Can I ask again please? Since Massive MIMO results in channel hardening , i.e. a channel with a condition number ~1 , so orthogonal, why is there still a need for zero-forcing?
I think you are referring to "favorable propagation" (channel hardening = the small-scale fading average out). The thing is that the orthogonality between the channels of different users only happens asymptotically, as we let the number of antennas go to infinity. In practice, when we have ~100 antenna, there will be some interference left and it can be suppressed using zero forcing. In the beginning of the Massive MIMO research, there was a hope that the difference would be small, but that hypothesis didn't hold. In many cases, one can double the data rate by using (regularized) zero-forcing.
@@WirelessFuture If I may again: So I simulated in Matlab a random wireless channel with 8 users and a large number of antennas (up to 1000) and I saw after doing a SVD that the singular values converge to a certain mean value as I was increasing the number of antennas, is this channel hardening?
Hi, Professor. I have a question on the channel bandwidth. I notice the noise power is set as -80 dBm in many wireless communication papers. According to the equation -80dBm = -174 + 10*log10( B), B equals to 10^(9.4) Hz. Is it correct? cause the bandwidth is 2.5GHz, it's so large actually. So is B the channel bandwidth or signal bandwidth?
The thermal noise variance can indeed be calculated as -174 + 10*log10(B) dBm, but there will also be noise amplification (called noise factor) in the receiver hardware. It might be around 8-10 dB that should be added to that number. If the noise factor is 10 dB, then 250 MHz will be bandwidth instead. It is large but not unreasonable. It is the signal that has a bandwidth, not the channel.
Excellent thankyou so much. Can you please recommend me where to get a MATLAB code fore linear precoding techniques for beginners which is easy to understand.
Hello Dr. Emil Thank you for your efforts to explain this important subject. Sorry Sir, but I did not understand how you calculate the noise power in dbm if the noise power spectal density is assumed 10 power (-14.4) Watt/Hertz then you divide by 1 milliwatt and you get the constant (-174)?
Yes, it is the power of the noise in the receiver circuits. (There is also noise in the transmitter circuits, but it is usually neglected in communications because it is much weaker than the transmitted signals.)
another great video , thank you prof , when you calculate B (bandwidth) you calculate 10Mhz but if you calculate this for LTE which one do you calculate 15Khz or 180Khz (12*15 OFDM) or whole bandwidth ? which could be 20Mhz or 40Mhz depending on bands and countries . does this calculation works for microwave point to point link too ?
If you want to compute data rate over the entire bandwidth, then you use the whole bandwidth. If you want to compute the data rate over one LTE resource block then you use 180 kHz. If you want to compute the data rate on one subcarrier then you use 15 kHz. The formulas applies for any type of point-to-point link.
When working with noise and SNR, you should probably use proper SI units as KHz is literally decoded as kelvin hecto seconds, not kilo hertz as you likely intended. This is because K is the SI unit symbol for temperature in kelvin, and the SI unit for frequency is hertz (Hz). The SI unit prefix h is for the 10 to the second power and k is the 10 to the third power multiplier. Temperature is a component of the noise power equation kTB where k is Boltzmann's constant, T is temperature in kelvin, and B is the bandwidth in hertz or some multiple. With the SI unit errors above you would end up with a temperature factor left over in your kTB calculation. For example, at room temperature you would have 1.38 x 10^-23 J/K * 290 K * 15 K * Hz = 6 x 10^-20 K * Hz^-1. But in answer to your question, the concept of SNR works for all kinds of communications links including microwave and satellite communications, you just have to apply the appropriate necessary bandwidth (see ITU Radio Regulations Vol-1 for definition).
When you say isotropic antenna, do you mean lossless isotropic antenna, i.e. the antenna gain is 1 = 0 dB in all directions? When you talk about antenna size you are explaining antenna effective area. The antenna size is usually not the same as the antenna effective area.
pulsatorius Yes, 0 dBi. Since I only consider isotropic antennas in this video, the effective and “physical” antenna area are the same. But you are right that this is not the case in practice.
Thank you again, while I do not want to take too much advantage of time, kindness and this opportunity to ask you questions , I still have a general MIMO question that is bothering me, please allow me to ask it: In MIMO (spatial multiplexing) or even in receive diversity antennas need to be uncorrelated as much as possible Now suppose we have 2 antennas and we have two streams of data that need to be received independently by these antennas and suppose the isolation is say 15 dB between the two antennas and suppose that each of these two data streams require an SNR of say 20 dB in order to be detected (64 QAM for example) Won't they contaminate each other by 15 dB?, i.e. a level of contamination that will make the detection difficult, since they need an SNR of 20 dB? Would in that case spatial multiplexing not be possible?...wouldn't make sense to me I would really appreciate an answer
I'm not sure if you are referring to isolation in the hardware or over the wireless channel, but precoding is the solution to the issue that you describe. If you know the propagation channel, you can use zero-forcing precoding to transmit two signals from the two antennas in such a way that each receive antenna observes one of the signals without any interference. You essentially make sure that the two antennas transmit copies of the same signal so that these copies add destructively at the undesired receiver. In this way, the interference disappear, at the price of an SNR reduction since the copies will not add fully constructively at the desired receiver. To mitigate this SNR loss, Massive MIMO is all about having more antennas than receivers to have sufficient degrees-of-freedom to suppress interference without affecting the SNR too much.
The description in the video is for omnidirectional antennas. If you have a directional antenna, then you need to multiply the SNR with the antenna gain in the direction of the signal.
Man I got turd brains. I got a wifi signal of -90 to -95 DB bro. I have the same set up at my moms house but she pays more and gets 107 to 100 SNR. The internet at my moms is amazing but at my trailer is garbage. Can someone tell how to improve Dbn. I moved all my fucking furniture and put it on a fucking stupid stool in the middle of the room and not a fucking thing changed please someone help before I have a annerism
You are right that there is noise also in the transmitter. However, since the propagation loss is normally -60 dB to -120 dB in wireless communications, the transmitter noise will be attenuated by such a huge factor before reaching the receiver. Hence, when computing the communication performance (which is measured at the receiver), it is normally the noise at the receiver that matters while the transmitter noise is negligible.
@@WirelessFuture As much as noise will be attenuated, so will the signal so Transmit SNR will not change because of path loss. Usually transmit SNR is much higher than receive SNR , that is why its impact can usually be ignored, but at mm wave for example, when PLL/VCO phase noise is not good enough, then transmit SNR could start having an impact in the overall SNR
@@dreamerhavingfun2289 Yes, my point is that the transmit SNR typically is 60-120 dB higher than the receive SNR, so that is why it can be ignored. This argumentation is based on the noise being thermal noise. If you bring in phase noise or other types of hardware distortion into the game, then the situation could be different, but I wouldn't talk about SNR in that case but rather signal-to-distortion-and-noise ratio (SNDR).
![Three Benefits of Using Multiple Antennas in Communications [Video 2]](http://i.ytimg.com/vi/tw0yD65viVI/mqdefault.jpg)
![Three Benefits of Using Multiple Antennas in Communications [Video 2]](/img/tr.png)
Where have you guys been all this time, I've been looking for such practical training ALL MY LIFE. Thank you for the detailed explanation for SNR.
This professor is absolutely outstanding. He explains this very well. He utilizes the math in his presentations, but doesn't lose me, because he has a good pace and logical explanations. Thank you!!!
Thank you professor Emil Björnson, Impressive I learn a lot watching your tutorials.
Thank you Professor Björnson. I am a telecommunications engineer and this is probably the best Video on SNR.
very well explained and illustrated 🎉🎉🎉 for such a complex field. As a telecommunication technician found it to be a gem 🎉🎉🎉. thumbs up and subs.
Great teaching skills, easy to understand difficult topics with this style.
Excellent teaching, clear explanation and easy to understand.
Superb!! . Very explicative and easy to understand .
Thank you sir for providing this Inst. SNR concept.
Crystal clear concepts ! Superb teaching
how to regenerate graph 2 at 8:07? specially how did you get 60dB SNR at 8:18
Great Explanation , explicit and clear
I saw a video that if you want to increase your SINR (dB), you should check the antenna of your neighbors. If everybody is tilting up, you should tilt your antenna a bit lower. Thanks for the video.
Does formula of the effective antenna area refers to an antenna length of lambda?
The antenna area is dependend on the size of the antenna. Thanks.
Or does it refers to a lambda/2 dipole?
The effective area of an isotropic antenna is a hypothetical number, which is explained here: en.wikipedia.org/wiki/Isotropic_radiator
For aperture antennas, the effective area is roughly the same as physical area of the antenna.
But for a dipole, it is not as easy. If the length is lambda/2, the directivity gain is 2.15 dB, which means that is effective area is 2.15 dB ≈ 1.6 times larger than that of an Isotropic antenna.
Excellent explanation! Thank you.
Very good explanation 😊
Hi Professor. I have a question on the calculation of SINR. Is SINR the power of the useful signal over the power of the sum of the interfering signals plus the noise power, or is it the power of the useful signal over the sum of the power of the interfering signals plus the noise power? The difference is that when we consider multi-user downlink communication, some articles directly state the sum of the interfering signal power, which really confuses me. Looking forward to your answer.
Thanks for the question. Whenever the interfering signals are independent, they appear in the SINR as a summation of their powers:
P_useful / ( P1+…PN+noise power)
where P1,…,PN are the powers of N interfering signals.
@@WirelessFuture Many thanks!
At 2:34, what is called "diameter (d)" is typically referred to as "radius (r)" -- the distance from the center of the sphere to its surface. The formula works based on the diagram drawn, but the term "diameter" is used in a way I've not seen before.
The variable "d" in the video is the distance to the receiver and, yes, it is the radius of the sphere. Was the word "diameter" used anywhere in the video? I couldn't find that.
he was just referring a distance from the antenna, not diameter nor radius
1)What is meant by transmitted power? At @9:11, transmitted power will be multiplied by antenna gain? Or it will be considered in the definition of channel gain? Although the resultant SNR will be the same in both cases.
2)What will happen for the channel gain in a directional antenna case where the denominator will be the same as the area of the entire sphere? Is such a definition of channel gain based on the assumption of spherical mode expansion only?
3) What will be the SNR when the receiving location does not lie in the far-field region? For example in an indoor scenario.
1) The transmit power is the power of the transmitted signal. The received power is the transmit power multiplied with the channel gain. We consider unit antenna gain in this video, but otherwise I recommend including it in the channel gain.
2) The definition is based on the assumption of an isotropic antenna. Otherwise the numerator will change.
3) There is no strong far-field assumption in this video. As long as we are beyond the reactive near-field, it will be fine. That will be the case also indoors.
@@WirelessFuture thank you.
I agree with the notion of SNDR . I have another question please: How does massive MIMO work in a non line of sight environment?
It works very well. You preferably estimate the multipath channels in the uplink and utilize the estimate in both uplink and downlink. The transmitted signal will not be look like a beam but exploit all the paths.
@@WirelessFuture Works very well in theory - in practice, not so much. Ask Verizon about their experience rolling out 28 GHz 5G roll. So far, not so good . . .
So is Eigen beamforming used on the downlink after using the uplink for channel estimation (TDD reciprocity)?
Yes, that is one option. If you serve multiple users, then regularized zero-forcing is a better choice.
@@WirelessFuture Thank you
@@WirelessFuture Can I ask again please? Since Massive MIMO results in channel hardening , i.e. a channel with a condition number ~1 , so orthogonal, why is there still a need for zero-forcing?
I think you are referring to "favorable propagation" (channel hardening = the small-scale fading average out). The thing is that the orthogonality between the channels of different users only happens asymptotically, as we let the number of antennas go to infinity. In practice, when we have ~100 antenna, there will be some interference left and it can be suppressed using zero forcing. In the beginning of the Massive MIMO research, there was a hope that the difference would be small, but that hypothesis didn't hold. In many cases, one can double the data rate by using (regularized) zero-forcing.
@@WirelessFuture If I may again: So I simulated in Matlab a random wireless channel with 8 users and a large number of antennas (up to 1000) and I saw after doing a SVD that the singular values converge to a certain mean value as I was increasing the number of antennas, is this channel hardening?
Hi, Professor. I have a question on the channel bandwidth. I notice the noise power is set as -80 dBm in many wireless communication papers. According to the equation -80dBm = -174 + 10*log10( B), B equals to 10^(9.4) Hz. Is it correct? cause the bandwidth is 2.5GHz, it's so large actually. So is B the channel bandwidth or signal bandwidth?
The thermal noise variance can indeed be calculated as -174 + 10*log10(B) dBm, but there will also be noise amplification (called noise factor) in the receiver hardware. It might be around 8-10 dB that should be added to that number.
If the noise factor is 10 dB, then 250 MHz will be bandwidth instead. It is large but not unreasonable.
It is the signal that has a bandwidth, not the channel.
@@WirelessFuture So clear explanation and thx!
Excellent thankyou so much. Can you please recommend me where to get a MATLAB code fore linear precoding techniques for beginners which is easy to understand.
The code associated with Chapter 1 in “Massive MIMO networks” might be what you are looking for: massivemimobook.com
@@WirelessFuture Thankyou so much again and I will refer on it.
@@WirelessFuture I tried to check but the link doesn't open is there another way to open the link? Thank you with regards
thewodros ayele The link works for us, try massivemimobook.com/wp/
The code is available on Github: github.com/emilbjornson/massivemimobook
@@WirelessFuture yes it works thankyou
Plsease, Can we use this parameter to design g5 network
The SNR is used in 5G and other networks to determine what data rate the system supports.
Nice Video, Clear Concept..... :-)
Hello Dr. Emil
Thank you for your efforts to explain this important subject.
Sorry Sir, but I did not understand how you calculate the noise power in dbm if the noise power spectal density is assumed 10 power (-14.4) Watt/Hertz then you divide by 1 milliwatt and you get the constant (-174)?
It is a typo in the video and, unfortunately, I haven't found a way to correct it. -174 dBm/Hz is the correct number it becomes 10^(-20.4) Watt/Hz.
@@WirelessFuture No=KTo=4x10^-21 W/Hz. This is also -174 dBm/Hz
Great explanation! Could you suggest some excellent videos for Analog and Digital communication course.
Someone recently recommend the channel “Ian Explains Signals and Systems”. Maybe it can be of interest?
@@WirelessFuture Thanks a lot sir!!
One quick question professor. so, the noise power will be always the same regardless of the distance?
Yes, it is the power of the noise in the receiver circuits. (There is also noise in the transmitter circuits, but it is usually neglected in communications because it is much weaker than the transmitted signals.)
Great quality videos. Success.
another great video , thank you prof , when you calculate B (bandwidth) you calculate 10Mhz but if you calculate this for LTE which one do you calculate 15Khz or 180Khz (12*15 OFDM) or whole bandwidth ? which could be 20Mhz or 40Mhz depending on bands and countries .
does this calculation works for microwave point to point link too ?
If you want to compute data rate over the entire bandwidth, then you use the whole bandwidth. If you want to compute the data rate over one LTE resource block then you use 180 kHz. If you want to compute the data rate on one subcarrier then you use 15 kHz. The formulas applies for any type of point-to-point link.
@@WirelessFuture thank you
When working with noise and SNR, you should probably use proper SI units as KHz is literally decoded as kelvin hecto seconds, not kilo hertz as you likely intended. This is because K is the SI unit symbol for temperature in kelvin, and the SI unit for frequency is hertz (Hz). The SI unit prefix h is for the 10 to the second power and k is the 10 to the third power multiplier.
Temperature is a component of the noise power equation kTB where k is Boltzmann's constant, T is temperature in kelvin, and B is the bandwidth in hertz or some multiple.
With the SI unit errors above you would end up with a temperature factor left over in your kTB calculation. For example, at room temperature you would have 1.38 x 10^-23 J/K * 290 K * 15 K * Hz = 6 x 10^-20 K * Hz^-1.
But in answer to your question, the concept of SNR works for all kinds of communications links including microwave and satellite communications, you just have to apply the appropriate necessary bandwidth (see ITU Radio Regulations Vol-1 for definition).
Wow. Just awesome !
Thank you Prof...
Why -174 and not -144 due to N0 being 10^(-14.4)?
It should be -174 dBm/Hz and 10^(-20.4) W/Hz. The 1000 times difference has to with mW versus W.
@@WirelessFuture Makes sense to me now. Thank you so much for the quick reply!
AWESOME BROTHER
When you say isotropic antenna, do you mean lossless isotropic antenna, i.e. the antenna gain is 1 = 0 dB in all directions?
When you talk about antenna size you are explaining antenna effective area. The antenna size is usually not the same as the antenna effective area.
pulsatorius Yes, 0 dBi. Since I only consider isotropic antennas in this video, the effective and “physical” antenna area are the same. But you are right that this is not the case in practice.
Thanks a lot!
I like it Sir...
Thank you again, while I do not want to take too much advantage of time, kindness and this opportunity to ask you questions , I still have a general MIMO question that is bothering me, please allow me to ask it:
In MIMO (spatial multiplexing) or even in receive diversity antennas need to be uncorrelated as much as possible
Now suppose we have 2 antennas and we have two streams of data that need to be received independently by these antennas and suppose the isolation is say 15 dB between the two antennas and suppose that each of these two data streams require an SNR of say 20 dB in order to be detected (64 QAM for example)
Won't they contaminate each other by 15 dB?, i.e. a level of contamination that will make the detection difficult, since they need an SNR of 20 dB?
Would in that case spatial multiplexing not be possible?...wouldn't make sense to me
I would really appreciate an answer
I'm not sure if you are referring to isolation in the hardware or over the wireless channel, but precoding is the solution to the issue that you describe. If you know the propagation channel, you can use zero-forcing precoding to transmit two signals from the two antennas in such a way that each receive antenna observes one of the signals without any interference. You essentially make sure that the two antennas transmit copies of the same signal so that these copies add destructively at the undesired receiver. In this way, the interference disappear, at the price of an SNR reduction since the copies will not add fully constructively at the desired receiver. To mitigate this SNR loss, Massive MIMO is all about having more antennas than receivers to have sufficient degrees-of-freedom to suppress interference without affecting the SNR too much.
Nice, perfect!
thank you very much
thank you for the worthy knowledge and clear explanation. may know theSNRr in the omnidirectional and directional antenna?
The description in the video is for omnidirectional antennas. If you have a directional antenna, then you need to multiply the SNR with the antenna gain in the direction of the signal.
Man I got turd brains. I got a wifi signal of -90 to -95 DB bro. I have the same set up at my moms house but she pays more and gets 107 to 100 SNR. The internet at my moms is amazing but at my trailer is garbage. Can someone tell how to improve Dbn. I moved all my fucking furniture and put it on a fucking stupid stool in the middle of the room and not a fucking thing changed please someone help before I have a annerism
With all my respect SNR does not depend only on receive SNR, if transmit SNR is not too much higher than receive SNR, it will depend on both!
You are right that there is noise also in the transmitter. However, since the propagation loss is normally -60 dB to -120 dB in wireless communications, the transmitter noise will be attenuated by such a huge factor before reaching the receiver. Hence, when computing the communication performance (which is measured at the receiver), it is normally the noise at the receiver that matters while the transmitter noise is negligible.
@@WirelessFuture As much as noise will be attenuated, so will the signal so Transmit SNR will not change because of path loss. Usually transmit SNR is much higher than receive SNR , that is why its impact can usually be ignored, but at mm wave for example, when PLL/VCO phase noise is not good enough, then transmit SNR could start having an impact in the overall SNR
@@dreamerhavingfun2289 Yes, my point is that the transmit SNR typically is 60-120 dB higher than the receive SNR, so that is why it can be ignored. This argumentation is based on the noise being thermal noise. If you bring in phase noise or other types of hardware distortion into the game, then the situation could be different, but I wouldn't talk about SNR in that case but rather signal-to-distortion-and-noise ratio (SNDR).
🙏🙏🌹🌹
thank you very much