MU-Beamforming MIMO

[Phy_comm]

To re-frame the question, can someone help me understand the output of weight calculation matrix block used for downlink precoding in MIMO Application framework.

I could not decipher if the out put is 12*1 and 12 correspond to subcarrier of one antenna. 

I actually need 128*12 matrix of channel estimates corresponding to multiuser beamforming. How can I locate these vectors in this output?

[ks30]

Phy_comm,

I believe you had this question answered via the support channels, but I am posting the answer here for community documentation purposes:

"Part 1

• Please refer to the Section 4.2.2.2.3.2 for the mathmatical description of the computation in the shown FPGA blocks.
• The calculation is done once per resource block (RB).
• Each MS uses one subcarrier out of each RB to transmit its pilot.
• Zero order hold in frequency is assumed on the other subcarriers of the RB.
• Because of the zero order hold the dimension of MS can be exchanged by layers at the end of the chain (1-1 mapping in this case). So the weight matrix would have the dimension 12 layers x 128 antennas = W
• In the equalizer W would be multiplied with 128 antennas x 12 subcarrier -> 12 layers x 12 subcarriers
• The precoder using the same matrix sizes but the other way around. W^T * (12 layer x12 subcarrier) = (128 antennas x 12 subcarrier)

Part 2

The 128 values for each layer/user are never available in one clock cycle through the entire design. There are two possible ways:

1. Wait for the 128 values to pass by and use a shift register to capture them so that in the last clock cycle there is the full vector available.
   For the application of the noise signal a serialization would be needed since the precoder also provides data for one antenna at a time.
2. Match the noise generation algorithm to the current data flow."

Regards,

Kyle S.