AoI-aware DAV Data Collection Based on the IRS's Phase Estimation in Wireless Sensor Networks

In Wireless Sensor Networks (WSNs), ensuring the real-time and reliable collection of state information is crucial for maintaining the efficient operation of the network. To ensure freshness of information in delay-sensitive applications such as autonomous driving, this paper combines Unmanned Aerial Vehicle (UAV) and Intelligent Reflecting Surface (IRS) technologies to deal with the IRS-aided UAV online data collection problem. In the formulated problem, the average Age of Information (AoI) of sensor nodes (SNs) is minimized by jointly optimizing the IRS's phase shifts, the UAV's trajectory, and SN association. The problem is divided into two coupled subproblems: the IRS's phase control and UAV's data collection. We propose the framework of Proximal Policy Optimization (PPO) to deal with the two subproblems individually, referred to as PPO-Dual. In particular, given the UAV's position and SN association, the optimal phase shifts of the IRS is estimated using PPO based on the partial channel state information (CSI) obtained by sending pilot signals through some active elements mounted on the IRS. Based on the well trained IRS's phase estimation strategy, the UAV's online data collection scheme is also determined using PPO. Simulation results show that the proposed PPO-Dual algorithm can effectively estimate the IRS's phase shifts and achieve the average AoI performance close to that under perfect CSI conditions.