MRNet-1R.py compresses, reports and reconstructs CSI of each receiving antenna separately.
MRNet-4R.py compresses, reports and reconstructs CSI of multiple receiving antennas cooperatively.
The cooperation technique of MRNet-4R can also be applied to other existing (or future) deep learning-based CSI feedback networks to make them have the same ability to process cooperative CSI feedback of multiple receiving antennas.
Relevant research has been submitted to 2022 IEEE International Conference on Communications (ICC 2022).
Relevant research has been accepted by ICC 2022.
environment requirement:
python == 3.6.5
tensorflow-gpu == 1.12.0
keras == 2.2.4