1) In @2:34 , U and V are assumed to be determined using SVD of G. What is the reason of such assumption? 2) What do the elements of the corresponding diagonal matrix represent? 3) How to determine the 'water level' q_k?
1) One can prove mathematically that this is the optimal design. It is the only way that one can divide the MIMO channel into parallel subchannels. 2) They represent the channel quality of each of the parallel subchannels. In the example given in the video, they represent three different paths. 3) You use the formula for q_k that is provided in the video and increase mu (the water level) until q_1 + ... + q_S = q, where q is the total available power.
Yes, there is an illustration of that at 3:40 where there signals are transmitted through spatial multiplexing. Each signal is sent with one beam and received with one beam. The main thing that makes this “MIMO” is that all these signals are transmitted simultaneously.
We don’t need to remove it. With the precoding, the channel will change from G to G*V. We don’t need to undo this operation, just apply the receiver filter U to handle this channel instead. We pick U and V to make U’*G*V a diagonal matrix.
According to equations, V is required at transmitter side, but is obtained from G. So how is this achieved? I found that there is concept of Pre-coding to provide CSI at Transmitter end. So it means, without precoding SVD for MIMO cannot be achieved?
V is also known as the precoding matrix. As you have noticed, the transmitter needs to have CSI (know the matrix G) to compute V in the way that is described. The common way to learn G is to send known signals called “pilots” and use the received signal to estimate G.
@@WirelessFuture Thank you very much for the updates. The videos are of great help. Looking forwards to more topics on wireless comm, akin, SCMA, NB-IoT, Massive MIMO, NOMA, CFO etc. Thanks Indeed for the timely replies on comments.
![Line-of-sight Channels [Video 7]](http://i.ytimg.com/vi/wuDlZKtRWro/mqdefault.jpg)
![Line-of-sight Channels [Video 7]](/img/tr.png)
![Capacity of Point-to-point SIMO and MISO Channels [Video 5]](http://i.ytimg.com/vi/WJ1e_0sIJRk/mqdefault.jpg)
The conclusion in the end is awesome, thank you professor
1) In @2:34 , U and V are assumed to be determined using SVD of G. What is the reason of such assumption?
2) What do the elements of the corresponding diagonal matrix represent?
3) How to determine the 'water level' q_k?
1) One can prove mathematically that this is the optimal design. It is the only way that one can divide the MIMO channel into parallel subchannels.
2) They represent the channel quality of each of the parallel subchannels. In the example given in the video, they represent three different paths.
3) You use the formula for q_k that is provided in the video and increase mu (the water level) until q_1 + ... + q_S = q, where q is the total available power.
@@WirelessFuture thank you :)
Are transmitter and receiver both using beamforming in this scenario?
Yes, there is an illustration of that at 3:40 where there signals are transmitted through spatial multiplexing. Each signal is sent with one beam and received with one beam. The main thing that makes this “MIMO” is that all these signals are transmitted simultaneously.
Thanks you very much, ihave question about how we remove the effect of the precoding vector at the receiver i mean V*V?
We don’t need to remove it. With the precoding, the channel will change from G to G*V. We don’t need to undo this operation, just apply the receiver filter U to handle this channel instead. We pick U and V to make U’*G*V a diagonal matrix.
@@WirelessFuture thanks you very much
According to equations, V is required at transmitter side, but is obtained from G. So how is this achieved?
I found that there is concept of Pre-coding to provide CSI at Transmitter end.
So it means, without precoding SVD for MIMO cannot be achieved?
V is also known as the precoding matrix. As you have noticed, the transmitter needs to have CSI (know the matrix G) to compute V in the way that is described. The common way to learn G is to send known signals called “pilots” and use the received signal to estimate G.
@@WirelessFuture This would estimate G at the Receiver end. But we need G (to use V at Tx.) How will Tx. know the estimated value of G ?
Either the transmitter estimates G from pilot signals sent in the opposite direction, or the receiver feeds back its estimate.
@@WirelessFuture Thank you very much for the updates. The videos are of great help. Looking forwards to more topics on wireless comm, akin,
SCMA, NB-IoT, Massive MIMO, NOMA, CFO etc. Thanks Indeed for the timely replies on comments.
Your videos are nice but I sometimes face problem in understanding as you speak very fast.