[Paper Review] Mobile Reconfigurable Intelligent Surfaces for NOMA Networks: Federated Learning Approaches

Ruikang Zhong, Xiao Liu+3 more|arXiv (Cornell University)|Mar 20, 2021

Advanced Wireless Communication Technologies7 citations

TL;DR

This paper proposes a mobile reconfigurable intelligent surface (RIS) system integrated with non-orthogonal multiple access (NOMA) to enhance spectral efficiency and user capacity in indoor wireless networks. By employing a federated learning-enhanced deep deterministic policy gradient (FL-DDPG) algorithm, the system jointly optimizes RIS mobility, phase shifts, and power allocation, achieving up to 3× higher data rates than fixed RIS and 42% sum rate gain over OMA.

ABSTRACT

A novel framework of reconfigurable intelligent surfaces (RISs)-enhanced indoor wireless networks is proposed, where an RIS mounted on the robot is invoked to enable mobility of the RIS and enhance the service quality for mobile users. Meanwhile, non-orthogonal multiple access (NOMA) techniques are adopted to further increase the spectrum efficiency since RISs are capable to provide NOMA with artificial controlled channel conditions, which can be seen as a beneficial operation condition to obtain NOMA gains. To optimize the sum rate of all users, a deep deterministic policy gradient (DDPG) algorithm is invoked to optimize the deployment and phase shifts of the mobile RIS as well as the power allocation policy. In order to improve the efficiency and effectiveness of agent training for the DDPG agents, a federated learning (FL) concept is adopted to enable multiple agents to simultaneously explore similar environments and exchange experiences. We also proved that with the same random exploring policy, the FL armed deep reinforcement learning (DRL) agents can theoretically obtain a reward gain compare to the independent agents. Our simulation results indicate that the mobile RIS scheme can significantly outperform the fixed RIS paradigm, which provides about three times data rate gain compare to the fixed RIS paradigm. Moreover, the NOMA scheme is capable to achieve a gain of 42% in contrast with the OMA scheme in terms of sum rate. Finally, the multi-cell simulation proved that the FL enhanced DDPG algorithm has a superior convergence rate and optimization performance than the independent training framework.

Motivation & Objective

• Address the limited line-of-sight (LoS) and poor channel quality for mobile users in obstructed indoor environments.
• Overcome the limitations of fixed RIS deployment by enabling dynamic, robot-mounted RIS mobility to adaptively improve signal coverage.
• Leverage NOMA to enhance spectral efficiency and user capacity in RIS-empowered networks by enabling superposition coding and successive interference cancellation.
• Develop a scalable, privacy-preserving training framework for multi-agent reinforcement learning in distributed RIS networks using federated learning (FL).

Proposed method

• Deploy a mobile RIS mounted on a robot to dynamically reposition and adjust phase shifts for optimal signal reflection to mobile users.
• Integrate NOMA with RIS to exploit artificial channel control for improved user fairness and spectral efficiency through superposition precoding.
• Formulate a joint optimization problem for RIS deployment, phase shifts, and power allocation to maximize sum rate under quality-of-service constraints.
• Apply deep deterministic policy gradient (DDPG) for continuous action space control of RIS mobility and beamforming.
• Enhance DDPG training via federated learning (FL), enabling multiple agents in different cells to collaboratively train using shared global model updates while preserving data privacy.
• Use a global neural network model to aggregate local gradients from agents, improving exploration diversity and convergence speed in dynamic environments.

Experimental results

Research questions

• RQ1 Can mobile RIS deployment significantly improve spectral efficiency and user data rates compared to fixed RIS in obstructed indoor environments?
• RQ2 To what extent does NOMA outperform orthogonal multiple access (OMA) in RIS-empowered networks when combined with dynamic RIS beamforming?
• RQ3 How does federated learning improve the training efficiency and convergence of deep reinforcement learning agents in multi-cell mobile RIS networks?
• RQ4 What is the impact of environmental heterogeneity (e.g., varying fading characteristics) on the performance of FL-enhanced DRL in distributed RIS systems?

Key findings

• The mobile RIS scheme achieves approximately 30.1% data rate gain over the no-RIS scenario, significantly outperforming the fixed RIS paradigm, which yields only 12.4% gain.
• Compared to OMA, the NOMA-based system with dynamic decoding order achieves a 42% sum rate gain, demonstrating the effectiveness of power-domain multiplexing in RIS-aided networks.
• The FL-enhanced DDPG algorithm reduces training time by 40% compared to independent DRL training, achieving equivalent performance in 150 episodes versus 250 episodes for the single-cell baseline.
• Even under high environmental diversity (DF = 1, indicating independent channel characteristics), the FL-DDPG framework maintains stable convergence and achieves sum rates comparable to the single-cell case.
• The mobile RIS with dynamic deployment provides a 15.1% data rate gain over the fixed RIS, exceeding the performance gap between fixed RIS and no-RIS networks.
• Phase shift optimization via DRL yields a 10.2% gain over static decoding order, and random phase shifts provide only marginal improvement over no-RIS, highlighting the necessity of intelligent control.