Learning uplinks and downlinks transmissions in RF-charging IoT networks

Area of Science:

• Wireless Communication Networks
• Internet of Things (IoT)
• Machine Learning Applications

Background:

• Radio frequency-powered IoT devices face energy constraints impacting data transmission.
• Existing scheduling methods for IoT networks do not fully leverage causal information for energy-aware transmissions.

Purpose of the Study:

• To develop a learning-based approach for optimizing uplink and downlink transmissions in RF-powered IoT networks.
• To enable a hybrid access point and IoT devices to learn optimal power allocation, frame size, transmission probability, and power split ratios.

Main Methods:

• A novel learning approach is proposed for a hybrid access point and IoT devices.
• The approach optimizes downlink power allocation and uplink frame size using causal information.
• Devices learn to optimize transmission probability, data slot, and power split for harvested energy and data rate.

Main Results:

• The proposed learning approach significantly outperforms non-learning methods.
• Achieved higher average sum rates compared to Aloha, time division multiple access, and round-robin scheduling.
• Demonstrates effective energy coupling management across time slots.

Conclusions:

• The learning-based strategy offers a superior method for managing resources in RF-powered IoT networks.
• This approach enhances overall network efficiency and data throughput.
• Highlights the potential of intelligent, adaptive algorithms in future IoT deployments.