Task Offloading Optimization Algorithm for UAV Assisted Mobile Edge Computing Network

As the widespread growth of Internet of Things (IoT), the number of applications on mobile devices (MDs) has gradually increased, resulting in a significant volume of compute-intensive and delay-sensitive tasks. However, the traditional Mobile Edge Computing Network (MEC) deploys servers at the edge of the network and lacks flexibility. In addition, due to the flexible mobility of MDs, the communicate on link between MDs and edge servers is unstable. Taking into account the previously mentioned problems, we propose a UAV-assisted mobile edge network system, which allows MDs to offload tasks to UAV. Secondly, we model the combined challenge of computation offloading and resource allocation as a Markov Decision Process (MDP). Then, we propose Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm to find the most efficient tactics. Finally, considering that the variables of the UAV are in different ranges in the state set, we further propose a state normalization algorithm. The simulation findings indicate that our algorithm demonstrates superior performance compared to the other algorithms across various parameters.