Power allocation optimization and location deployment strategy of STAR-RIS based on DDPG

Simultaneously transmitting and reflecting reconfigurable intelligent surface is a new type of intelligent reflector that has both reflection and refraction functions, thereby improving the transmission performance of wireless communication systems. This paper first proposes a wireless communication system assisted by STAR-RIS carried by drones, and then combines the base station beamforming and STAR-RIS phase vector to maximize the user weighted sum rate. Finally, a deep deterministic policy gradient algorithm is proposed to solve the non-convex problem of the joint optimization problem. The simulation results show that the proposed joint beamforming and phase vector algorithm based on DDPG has better performance than the random algorithm and DQN algorithm, and is closer to the performance of the WMMSE algorithm.