README.md

Python code and related materials for DualNet and U2D network.

Introduction

This repository contains the code and related materials for DualNet and its extension U2D network. DualNet is described in Zhenyu Liu, Lin Zhang, and Zhi Ding, “Exploiting Bi-Directional Channel Reciprocity in Deep Learning for Low Rate Massive MIMO CSI Feedback,” IEEE Wireless Communications Letters, 2019. [Online]. Available: https://ieeexplore.ieee.org/document/8638509/. U2D network has been submitted.

Requirements

• Python 3.5 (or 3.6)
• Keras (>=2.1.1)
• Tensorflow (>=1.4)
• Numpy

Data Set

The CSI data is generated using COST 2100 channel model. You can refer the paper below and the corresponding implementations: L. Liu, J. Poutanen, F. Quitin, K. Haneda, F. Tufvesson, P. De Doncker, P. Vainikainen and C. Oestges, “The COST 2100 MIMO channel model,” IEEE Wireless Commun., vol 19, issue 6, pp 92-99, Dec. 2012. [Online]. Available: https://ieeexplore.ieee.org/document/6393523/

The original downlink and uplink CSI in delay domain: indoor channel with 5.1 GHz uplink and 5.3 GHz downlink bands. Normalization is required using training_testing_data_generation.m to generate the training set and testing set.

https://www.dropbox.com/s/wmi2wuq4betzryu/mat_indoor5351_bw20MHz_up.mat?dl=0

https://www.dropbox.com/s/av0u0m9kfr95vtf/mat_indoor5351_bw20MHz_down.mat?dl=0

CsiNet

The implementation of CsiNet can be found in https://github.com/sydney222/Python_CsiNet. Thank authors for sharing their code.