Question about generating training, validating and testing datasets about camerapredictsbeams HOT 4 CLOSED

Thank you very much! In fact, I can now run the code normally. I just want to confirm the parameter setting of "UPA_codebook_generator_DFT.m".

from camerapredictsbeams.

Like you, I also want to ask how to generate the codebook and how to set the parameters for the UFA_codebook_generator.

from camerapredictsbeams.

Thanks for your interest, both @nasirsinani and @Haotian-Zhang-PKU.

For the dataset, I really recommend you, guys, to read the notes in the beginning of the function you're using, for they are provided to help you "understand" what the function is doing. I know they are not detailed notes, but in this case they provide an answer to your question about preparing the datasets.

From the "getMATLAB" function notes:
":param matlab_file: path to the matlab data structure to read. The structure is expected
to have two fields: wireless channels and user locations. The channel field
should be a 4D array: #of antennas, # of subcarriers, # of user positions, # of BS
user location field should be a 2D array: 3, # of user positions"
The .mat file must be a "MATLAB" structure with two fields, whatever you want to name them. The important thing is that one is for channel data and the other is for the location data. Both fields must contain MATLAB arrays with different sizes, 4D for channel and 2D for location.

Please make sure you process the generated data to get the described structure!

from camerapredictsbeams.

Now for the codebook function, i.e., "UPA_codebook_generator_DFT.m", I have added some notes in the beginning of the function to explain the arguments. Please, use the parameters in Table 1 in the paper "Millimeter Wave Base Stations with Cameras: Vision Aided Beam and Blockage Prediction" to configure your codebook.

In the end, I would really like to emphasize two points:

1. Please read the notes provided in any script.
2. When you get stuck with a script, consider playing around with it. This usually resolves 70% of the problems you may face.

from camerapredictsbeams.