[Solved] Label the channel data for DL
I used the DeepMIMOv1 to generate the data and calculate the achievable rates based on Deep learning coordinated Beamforming for highly-mobile millimeter wave system. But Why do all the elements of DL_output become zeros? It seems like there is something wrong with log2(1+abs(Ch'*BF_codebook)).^2. When I converted the data type into complex double log2(1+abs(double(Ch)'*BF_codebook)).^2, the elements of the DL_output matrix are not zeros anymore, but some of the elements became 1.
Why is the DL_output matrix strange? Or Is there something wrong with my dataset generation?
There was a small problem in running the old code of the “Deep Learning Coordinated Beamforming for Highly-Mobile Millimeter Wave Systems” paper with the new scenario files (which has a single instead of double precision). We have revised the scripts to fix that. Please check the application page to download the updated version, and re-run the simulations.
Note: DeepMIMO versions are also recently updated. Please use the current version to generate exact the same result.
Let me know if you face any other problems.