Thanks a lot. To my mind, the sound in this video and in the podcast is not as good as on former shows. But, perhaps that is just my perception. Short question, please, in all these modern base stations, we have crossed antenna elements. In the equation we multiply the element factor with the array factor. How do we have to assume the element factor in this case? Are these crosses 2 crossed dipol antennas which are fed at the centre? Thanks
The sound quality is indeed worse since it is a live recording in a lecture hall. Yes, it is common to use cross-polarized antennas. The element factor for a specific polarization is the same as for a uni-polarized antenna of the same kind. In a base station you typically have a backplane so the gain will be larger than for a normal lambda/2-dipole.
Many Thanks, a short generell question. To deploy dig. beamforming we just multiply with a matrix (ZF- or MMSE-matrix). With known H, I can compute it. When I now just want to minimize the interference to a special direction (alpha) and the second criterion is to max. the SNR in the transmit direction. How can I achieve that? Do I need a special multiplication matrix other than ZF and MMSE? I think we need alpha for the computation? How to put into the equation? Can I do that with more directions alpha 1,....,n? Thank you very much.
Hi! One way to achieve your goal is to treat the channel vector to the special direction as an additional user channel, and then generate the beamforming matrix using conventional theory. Then you just disregard the beamforming vector created for that special direction; the other beamforming vectors will be designed to limit the interference to that "special user". Another way to achieve zero interference in a particular direction with the vector a is to take your precoding matrix P that has been generated arbitrarily and replace it with (I-aa^H / ||a||^2) P. The first factor is a project matrix that ensures that the new precoding matrix is not putting any components in the a-direction.
@@WirelessFuture thanks, that is conceivable. When we have just the direction in degree (transmission to 0 deg. and interferer at 39 deg.), is there a possibility to do the same (no channel matrices)? Intuitively, I would say it must be possible. Thanks
@@manutauer Yes, if the channel vector a goes to the desired direction and the channel vector b goes to the interferer, then you can use the precoder p = (I-b*b^H/||b||^2)*a, where I is an identity matrix. You can verify that b^H*p = 0 because the expression within parentheses is a projection matrix that removes everything from the b-dimension. At the same time a^H p ≠ 0 (except if a and b are parallel vectors, so interference cannot be removed). For the considered scenario, you can pick a and b as the array response vectors (a.k.a. steering vectors) for the considered angles.
@@WirelessFuture Thank you very much, ok instead of calculating with the 3d vectors in space (x,y,z), we use the array response vectors for the calculation. Very interesting, I did not know that.
![Ep 38. Things We Learned at the 6G Symposium [Wireless Future Podcast]](http://i.ytimg.com/vi/j0HOLZjGRUQ/mqdefault.jpg)
![Ep 38. Things We Learned at the 6G Symposium [Wireless Future Podcast]](/img/tr.png)
![Ep 36. 6G from an Operator Perspective [Wireless Future Podcast]](http://i.ytimg.com/vi/odqBsaSBo-4/mqdefault.jpg)
Thanks a lot. To my mind, the sound in this video and in the podcast is not as good as on former shows. But, perhaps that is just my perception.
Short question, please, in all these modern base stations, we have crossed antenna elements. In the equation we multiply the element factor with the array factor. How do we have to assume the element factor in this case? Are these crosses 2 crossed dipol antennas which are fed at the centre? Thanks
The sound quality is indeed worse since it is a live recording in a lecture hall.
Yes, it is common to use cross-polarized antennas. The element factor for a specific polarization is the same as for a uni-polarized antenna of the same kind. In a base station you typically have a backplane so the gain will be larger than for a normal lambda/2-dipole.
So good to see so many bright minds in one podcast!
Many Thanks, a short generell question. To deploy dig. beamforming we just multiply with a matrix (ZF- or MMSE-matrix). With known H, I can compute it.
When I now just want to minimize the interference to a special direction (alpha) and the second criterion is to max. the SNR in the transmit direction. How can I achieve that? Do I need a special multiplication matrix other than ZF and MMSE? I think we need alpha for the computation? How to put into the equation?
Can I do that with more directions alpha 1,....,n?
Thank you very much.
Hi! One way to achieve your goal is to treat the channel vector to the special direction as an additional user channel, and then generate the beamforming matrix using conventional theory. Then you just disregard the beamforming vector created for that special direction; the other beamforming vectors will be designed to limit the interference to that "special user".
Another way to achieve zero interference in a particular direction with the vector a is to take your precoding matrix P that has been generated arbitrarily and replace it with (I-aa^H / ||a||^2) P. The first factor is a project matrix that ensures that the new precoding matrix is not putting any components in the a-direction.
@@WirelessFuture thanks, that is conceivable. When we have just the direction in degree (transmission to 0 deg. and interferer at 39 deg.), is there a possibility to do the same (no channel matrices)? Intuitively, I would say it must be possible. Thanks
@@manutauer Yes, if the channel vector a goes to the desired direction and the channel vector b goes to the interferer, then you can use the precoder p = (I-b*b^H/||b||^2)*a, where I is an identity matrix. You can verify that b^H*p = 0 because the expression within parentheses is a projection matrix that removes everything from the b-dimension. At the same time a^H p ≠ 0 (except if a and b are parallel vectors, so interference cannot be removed).
For the considered scenario, you can pick a and b as the array response vectors (a.k.a. steering vectors) for the considered angles.
@@WirelessFuture Thank you very much, ok instead of calculating with the 3d vectors in space (x,y,z), we use the array response vectors for the calculation. Very interesting, I did not know that.
Thanks, to some extend hard to understand due to acustic
nice talk I always follow the videos