Energy-optimized 6G communication framework with intelligent resource allocation for massive IoT networks.

Area of Science:

• Wireless Communications
• Network Optimization
• Artificial Intelligence

Background:

• 6G networks demand efficient resource allocation for massive Internet of Things (IoT) deployments.
• Simultaneous Transmitting and Reflecting Reconfigurable Intelligent Surfaces (STAR-RIS) offer novel propagation control capabilities.
• Existing methods often optimize radio resources and STAR-RIS independently, limiting overall system performance.

Purpose of the Study:

• To propose an integrated framework for energy-optimized uplink resource allocation in 6G massive IoT networks.
• To jointly optimize transmit power, subchannel assignment, and STAR-RIS coefficients.
• To leverage deep reinforcement learning for efficient and scalable network management.

Main Methods:

• Developed a novel framework integrating radio resource management and STAR-RIS control.
• Employed a Soft Actor-Critic (SAC) agent with Gumbel-Softmax relaxation for joint optimization.
• Utilized offline centralized training and online edge cloud coordinated execution.
• Conducted simulations using 3GPP Urban Micro channels with up to 200 devices and a 128-element STAR-RIS.

Main Results:

• Achieved a 24.3% increase in energy efficiency compared to baseline methods.
• Demonstrated an 18.7% higher aggregate throughput and a 19.1% reduction in latency.
• Extended network lifetime by 21.6% while maintaining near-optimal fairness.
• Validated the effectiveness of deep reinforcement learning for cross-layer integration.

Conclusions:

• Joint optimization of propagation control and radio resource allocation is crucial for 6G massive IoT.
• Deep reinforcement learning provides a scalable and effective solution for green massive machine-type communications.
• The proposed STAR-RIS assisted framework significantly enhances network efficiency and performance.