Deep Reinforcement Learning: Connecting the crowd using MIMO beamforming

Theme:

MIMO beamforming

Application:

Communication

Contact Person:

Nguyen Dao

External collabolator:

n/a

Intorduction:

Summary:

Multiple-input Multiple-output (MIMO) enables communication with multiple users by using multiple antennas. In MIMO networks, beamforming is an efficient method to enhance network performance, enables directional signal transmission and improving energy efficiency, as well as reducing interference.

 
Traditional optimization methods often struggle with complex and dynamic environments. Deep reinforcement learning (DRL), with its ability to adapt to changing scenarios and learn optimal policies from experience, offers a promising alternative solution to deal with the resource allocation and beamforming problems in 6G.

Description:

Method: 

This project will explore the application of DRL to beamforming in wireless communication systems. The primary objective is to design and implement RL-based algorithm that can dynamically optimize beamforming matrices under varying channel conditions, user demands, interference, and power constraints.

The research will involve:
1.         Modelling the problem as a Markov Decision Process (MDP).
2.         Developing DRL algorithms such as Deep Q-Learning, Actor-Critic networks, Proximal policy optimization (PPO), Deep Deterministic Policy Gradient (DDPG)…
3.         Design and analyze DRL networks with feature engineering, hyperparameters tunning.
4.         Validating the proposed solution through simulations in dynamic wireless environments.
5.         Comparing DRL-based approaches with each other or with traditional optimization techniques.

Requirements:

You are familiar with MATLAB. You have a background in Electrical Engineering or Computer Science, have knowledge in wireless communication, optimization, or willing to learn related skills.

Additional Info