RIS-Aided Proactive Mobile Network Downlink Interference Suppression: A Deep Reinforcement Learning Approach.

Area of Science:

• Wireless communication networks
• Signal processing
• Artificial intelligence

Background:

• Proactive mobile networks (PMN) offer low-latency communication but suffer from interference and reliability issues due to open-loop transmission and virtual cell technology.
• Managing reconfigurable intelligent surfaces (RIS) in complex, time-varying PMN environments with limited channel state information is challenging.

Purpose of the Study:

• To enhance the interference suppression capabilities and overall capacity of proactive mobile networks (PMN).
• To address the challenges of managing multi-reconfigurable intelligent surfaces (RIS) in dynamic PMN environments.

Main Methods:

• Formulated an optimization problem for RIS phase shifts and reflection coefficients.
• Developed a deep reinforcement learning (DRL) approach using an asynchronous advantage actor-critic (A3C) algorithm.
• Designed specific action spaces, state spaces, and reward functions tailored for the PMN-RIS system.

Main Results:

• Deployment of RIS in PMN regions significantly suppresses user interference.
• The proposed A3C-based RIS management scheme outperforms baseline methods in terms of network capacity.
• Capacity improvements approach theoretical limits as the number of RIS deployed increases.

Conclusions:

• Multi-reconfigurable intelligent surfaces (RIS) are effective in mitigating interference within proactive mobile networks (PMN).
• Deep reinforcement learning, specifically the A3C algorithm, provides an efficient solution for dynamic RIS management in PMNs.
• Integrating RIS technology holds significant promise for improving the performance and capacity of future low-latency communication systems.