AoI Minimization in UAV-Assisted IoT Network: A Reinforcement Learning Approach

Unmanned aerial vehicles (UAVs) are often used to assist internet of things (IoT) networks to collect data from IoT devices more securely and efficiently by virtue of their flexibility and low deployment cost. In this paper, we consider a UAV-assisted IoT network in which a large number of IoT devices are deployed along with eavesdroppers. The communication UAV and the jamming UAV cooperating to collect data from IoT devices. In order to minimize the age of information (AoI) of data from IoT devices at the communication UAV, this paper introduces reinforcement learning (RL) solutions. Considering the number of IoT devices is not fixed in the coverage area of communication UAV, which incurs the dimensional variations for the observation and action space, so we propose an effective twin delayed deep deterministic policy gradient (TD3) algorithm based on attention mechanisms to design the policy and critic networks. Simulation results show that the proposed algorithm is able to significantly reduce the average AoI of data from IoT devices at the communication UAV compared to baseline algorithms.