A Dynamic Reinforcement Learning Scheme for UAV-Based Joint Communication and Radar Systems.

In a joint communication and radar (JCR) system that performs radar sensing and communication simultaneously using a single signal, unmanned aerial vehicle (UAV) can efficiently utilize limited frequency resources and provide high-quality services when they can be detected and controlled. In this paper, we discuss the JCR system in a dynamic environment that considers user mobility and air-to-ground channels, and propose a dynamic UAV control solution to maximize system performance. The position of the UAV is a major factor that affects system performance, then, the proposed dynamic reinforcement learning scheme adjusts the trajectory of UAV and subframe length to derive the optimal solution. The simulation results show that the proposed scheme can adapt to the dynamic environment and converge to optimal values, demonstrating the ability to find optimal solutions for UAV trajectory and subframe length control.