DRQN-based Task Offloading in UAV-assisted Mobile Edge Computing Environments with Hidden Channel Conditions

Mobile edge computing (MEC) with unmanned aerial vehicle (UAV) assistance offers promising solutions for reducing latency in <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$5\mathbf{G}$</tex> networks. However, uncertainty in task processing requirements poses challenges for efficient task of-floading. This paper introduces a deep recurrent Q-network (DRQN) based algorithm for task offloading in UAV-assisted MEC environments with hidden channel conditions. We for-mulate the problem as a partially observable Markov decision process (POMDP), where the UAV decides whether to process tasks locally or offload them to a cloud server without complete information about the channel state. Our approach leverages temporal dependencies in the environment to make informed decisions under uncertainty. Simulations demonstrate that our DRQN-based algorithm outperforms previous methods, including DQN-based, offload-only, and local-only strategies. This work contributes to the development of robust task offloading strategies for dynamic edge computing environments in 5G and beyond networks.