|
|
|
|
|
|
Beamforming (that we have covered in this article) is based on high mutual correlation between the antenna elements
(distance between consecutive antenna elements is fraction of wavelength, so it gives high mutual correlation).
For multi-layer transmission, beams are chosen in such a way that there is least correlation (sufficient spatial separation) between participating beams.
MIMO Precoding is based on exactly this characteristic of low mutual correlation between the antenna ports
(in other words, different antenna ports experience different channels).
In result, we now have two Precoding metrices, W1 and W2.
W1 to choose Beam or list of Beams and W2 for Beam selection and/or MIMO Precoding.
MIMO Precoding will include co-phasing between participating beams as is expected from MIMO Precoding.
MIMO Precoding (by its design) will require more instantaneous channel knowledge/estimates than Beamforming Precoding.
This implies that we need to have two types of Precoding update cycles.
Slower cycle for Beamforming Precoding updates and Faster cycle for MIMO Precoding updates !!
By same reasoning, W1 wiil be Wideband/frequency independent whereas W2 will be Subband/frequency selective.
5G NR specifies four types of CSI codebooks,
Type 1 Single Panel,
Type 1 Multi-Panel,
Type 2, and
Type 2 Port Selection.
Type 1 are meant for SU-MIMO and Type 2 are meant for MU-MIMO.
Detailed discussions about these codebooks are *not* in the scope of this article.
We shall instead cover antenna representations, which forms the base of these codebooks.
Three representations for antenna are shown in below diagram.
Physical antenna may have multiple antenna elements, but it is logically represented as N1 columns and N2 rows of pairs,
each pair having two cross polarised antenna elements.
This gives us 2*N1*N2 CSI-RS ports.
Further, oversampling (sampling rate higher than Nyquist rate) gives us greater number of beams than possible with the given number of antenna elements.
This brings us O1 and O2 values as shown in the figure below.
Total number of possible beams will then be N1*O1*N2*O2.
Note that distance between antenna elements is half of the wavelength, as required by Beamforming.
However, there could be multiple panels and between the panels, distance could be half of the wavelength or more.
Basic idea is: each beam is identified by (N1,O1,N2,O2) coordinates.
So, in a way, it becomes a protocol between base station and UE.
|
|
|
|
|
|
|
|
|
|
