Skip to content

Apply Deep Reinforcement Learning aided by Federated Learning to Wireless Comunication

Notifications You must be signed in to change notification settings

akramIOT/FLDRL-in-Wireless-Communication

 
 

Repository files navigation

FLDRL-in-Wireless-Communication

  • Simulation code for Paper:
    Lyutianyang Zhang1 , Hao Yin1, Zhanke Zhou, Sumit Roy, Yaping Sun, Enhancing WiFi Multiple Access Performance with Federated Deep Reinforcement Learning, VTC2020-Fall.
    1 Both authors contribute equally to this work.
  • Cite our work:
@INPROCEEDINGS{FrmaVTC2020,
  author={L. {Zhang} and H. {Yin} and Z. {Zhou} and S. {Roy} and Y. {Sun}},
  booktitle={IEEE 92nd Vehicular Technology Conference (VTC2020-Fall)}, 
  title={Enhancing {WiFi} Multiple Access Performance with Federated Deep Reinforcement Learning}
  }

Contributors: Hao Yin, Zhanke Zhou

The paper can be found https://ieeexplore.ieee.org/document/9348485

Simulations

Author Notes:

  • Please check config.py for model loading and saving setups.

    • self.saveModel = False
      self.loadModel = True
      
  • Run python3 test_CSMA_DQN_withModelAllocation.py to proceed training.

  • Throughput is about 5.2-5.4

Training log

Number of Station Max Avg Throughput Total training epoch
5 5.45 10w
10 5.46 13w
20 5.28 22w

About

Apply Deep Reinforcement Learning aided by Federated Learning to Wireless Comunication

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 70.7%
  • Jupyter Notebook 29.3%