Aerial Multi-Functional RIS in Fluid Antennas-Aided Full-Duplex Networks: A Self-Optimized Hybrid Deep Reinforcement Learning Approach

2026-04-21 · Source: cs.AI updates on arXiv.org · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems, Emerging Technologies & Innovation · Depth: Expert, extended

Summary

This paper introduces a novel architecture, AM-RIS, integrating autonomous aerial vehicles (AAVs) and multi-functional reconfigurable intelligent surfaces (MF-RISs) into fluid antenna (FA)-assisted full-duplex (FD) 6G wireless networks. The AM-RIS provides hybrid functionalities including signal reflection, amplification, and energy harvesting (EH), enhancing signal coverage and sustainability. Fluid antennas at the FD-enabled base station (BS) offer fine-grained spatial adaptability, complementing residual self-interference (SI) suppression. The research aims to maximize overall energy efficiency (EE) by jointly optimizing transmit downlink (DL) beamforming, uplink (UL) user power, AM-RIS configuration (amplification, phase-shifts, EH coefficients), and the positions of FA and AM-RIS. To address the problem's hybrid continuous-discrete parameters and high dimensionality, a self-optimized multi-agent hybrid deep reinforcement learning (DRL) framework (SOHRL) is proposed, combining multi-agent deep Q-networks (DQN) for discrete actions and multi-agent proximal policy optimization (PPO) for continuous actions. SOHRL incorporates attention-driven state representation and meta-level hyperparameter optimization for enhanced self-adaptability.

Key takeaway

For AI Scientists and Research Scientists designing future 6G wireless networks, adopting the proposed AM-RIS architecture with the SOHRL algorithm is crucial for maximizing energy efficiency. Your designs should incorporate multi-functional RIS on AAVs and fluid antennas at the base station, managed by a hybrid DRL approach that includes attention mechanisms and self-optimizing hyperparameters. This integrated strategy significantly outperforms conventional methods, offering superior EE, especially with optimal AM-RIS deployment heights and element configurations.

Key insights

Integrating aerial MF-RIS with fluid antennas and a hybrid DRL framework optimizes 6G network energy efficiency.

Principles

• Hybrid DRL excels in complex continuous-discrete optimization.
• Attention mechanisms improve DRL performance by prioritizing features.
• Self-optimizing hyperparameters enhance DRL adaptability and stability.

Method

SOHRL integrates MADQN for discrete actions and MAPPO for continuous actions, using attention-driven states and a meta-agent for adaptive hyperparameter tuning to maximize energy efficiency.

In practice

• Deploy AM-RIS at ~150m height for optimal EE and rate.
• Utilize 4 multi-head attention units with 0.3 dropout for DRL stability.
• Prioritize full hyperparameter optimization for maximum EE gains.

Topics

• Aerial Multi-Functional RIS
• Fluid Antenna Systems
• Full-Duplex Communications
• Hybrid Deep Reinforcement Learning
• Energy Efficiency Optimization

Best for: AI Scientist, Research Scientist

Related on AIssential

• Heterogeneous Tasks Offloading in Vehicular Edge Computing: A Federated Meta Deep Reinforcement Learning Approach — FedMAGS optimizes VEC task offloading using federated meta-learning, GATs, and Seq2Seq for privacy-preserving, scalable performance.
• Digital Twin-Assisted Adaptive Multi-Agent DRL for Intelligent Spectrum and Resource Management in Open-RAN UAV-Enabled 6G Networks — A digital twin-assisted adaptive multi-agent DRL framework optimizes spectrum and resource management in dynamic Open-RAN UAV-enabled 6G networks.
• The End of Dead Zones : How RIS “Smart Wallpaper” is Engineering the 6G Air Interface — Reconfigurable Intelligent Surfaces (RIS) program the wireless environment to enhance 6G signal control and efficiency.
• Leveraging Large Vision Model for Multi-UAV Co-perception in Low-Altitude Wireless Networks — A novel framework optimizes multi-UAV perception by sparsifying visual data and using LVMs with DRL for efficient communication.
• A Reinforcement Learning Based Universal Sequence Design for Polar Codes — A reinforcement learning framework designs universal Polar code sequences, scaling to 2048 code lengths with performance gains.
• Learning from the Self-future: On-policy Self-distillation for dLLMs — d-OPSD adapts self-distillation for dLLMs by using suffix conditioning and step-level supervision, significantly boosting sample efficiency.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.AI updates on arXiv.org.