Minimization Age of Information for Aerial STAR-RIS-assisted MEC Networks: A Reinforcement Learning Approach

To effectively improve the freshness of information, this study investigates an aerial simultaneous transmitting and reflecting reconfigurable intelligent surface (STAR-RIS)-assisted mobile edge computing network, with a STAR-RIS mounted on the unmanned aerial vehicle (UAV) to strengthen the communication link between the base station and ground devices. In particular, an average age of information (AoI) minimization problem is formulated for jointly optimizing task offloading strategy, the UAV’s trajectory, amplitude and phase shift coefficients of the STAR-RIS, and the transmission power of devices. However, the optimization problem has non-convex characteristics and operates in a dynamically changing environment, traditional optimization approaches face challenges in handling high-dimensional variables and ensuring convergence to an optimal solution within complex operational environments. To tackle this difficulty, a deep reinforcement learning framework employing the proximal policy optimization algorithm is adopted to address this intractable problem effectively. Finally, numerical results reveal that the proposed algorithm has better convergence performance than other algorithms, and the presented approach surpasses baseline strategies in reducing the average AoI.