/DQN-Based-Power-Allocation-For-Multi-Cell-Massive-MIMO

Deep Q network-based power allocation for multi-cell massive MIMO cellular network.

Primary LanguageJupyter NotebookMIT LicenseMIT

DQN-based-Power-Allocation-For-Multi-Cell-Massive-MIMO

We have implemented a DRL-based power allocation system for massive MIMO in cellular networks, utilizing Open-AI Gym and TensorFlow frameworks.

This implementation reinforcement learning algorithm is designed solely for the Deep Q Network (DQN).

(* This code offers the initial draft version of the DRL-based power allocation technique on multi-cell networks.)

Hyperparameters

  • (N) is the number of cells equal to the number of BSs
  • (M) is the number of BS transmission antennas
  • (K) is the number of users in a cell
  • (BW) is bandwidth
  • (NF) is the power of noise figure [dBm]
  • (Ns) is the Number of samples as TDD interval
  • (min_p) is Minimum transmission power [dBm]
  • (max_p) is Maximum transmission power [dBm]
  • (num_p) is the number of action space size