A Learning-Based Flexible Beamforming Method for Movable Antenna-Enabled Integrated Sensing, Communication, and Power Transmission System

The Integrated Sensing, Communication, and Power Transmission (ISCPT) system is crucial for next-generation intelligent communication, enabling efficient spectrum management. By utilizing movable antennas (MAs), ISCPT enhances reliability and flexibility, supporting various intelligent Internet of Things (IoT) scenarios. While joint optimization of parameters like beamforming and antenna positioning improves performance, it also introduces computational complexity that may affect efficiency. To enable flexible beamforming, we propose a novel deep reinforcement learning (DRL) architecture where heterogeneous agents independently adjust antenna configurations. This approach improves sensing accuracy, communication reliability, and power transfer efficiency, enhancing system capabilities and adaptability to dynamic environments. This work lays the foundation for wireless systems that integrate intelligent communication, sensing, and power transfer, offering improved performance in real-world applications.