CsiNet-LSTM

Repository for reproduction of CsiNet-LSTM used in MarkovNet paper (currently preprint).

• [1] Liu, Z., del Rosario, M., & Ding, Z. (2020). A Markovian Model-Driven Deep Learning Framework for Massive MIMO CSI Feedback. arXiv preprint arXiv:2009.09468.
• [2] T. Wang, C. Wen, S. Jin, and G. Y. Li, “Deep learning-based csi feedback approach for time-varying massive mimo channels,” IEEE Wireless Communications Letters, vol. 8, no. 2, pp. 416–419, April 2019.

Data

TODO: Add a link to 20 timeslot COST2100 data.

Dependencies

TODO: Add a) exhaustive list of dependencies for this repo and/or b) .yml/.dockerfile for setting up working environment.

This repository relies heavily on the brat repository. In particular, the brat/utils directory handles parsing .json config files, loading data, and evaluating network performance.

The typical hierarchy which this

• home (your home directory)
  • git
    • brat (repo available here)
    • csinet-lstm (this repo)
      • csinet-lstm
        • csinet_train.py (training script for csinet at single timeslot)
        • csinet_lstm_train.py (training script for csinet_lstm; typically ten timeslots)