Intelligent reflecting surface-aided computation offloading in UAV-enabled edge networks

The popularity of wireless communication technology and smart devices make emerging tasks tend to be computationally intensive. Unfortunately, mobile devices are often computationally resource-constrained. Mobile edge computing is proposed to offer computing power for these resource-limited devices to solve the computing requirement of their tasks. The unmanned aerial vehicle (UAV) enabled edge networks are flexible and low-cost, so they are considered to provide more flexible computing service for mobile devices. However, UAV-enabled edge networks are limited by the weak wireless propagation environment. To this end, we introduce intelligent reflecting surface (IRS) into the UAV-enabled edge networks in which IRS is used to construct a stronger link between the mobile devices and the UAV for task offloading. We formulate the IRS-aided offloading problem as an optimization problem to optimize the overall delay by jointly optimizing UAV movement, offloading decision, IRS configuration, and UAV’s computation resource. To solve the problem more efficiently, we use the deep reinforcement learning (DRL) model to explore the intelligent action that can minimize the task processing time. Our simulation demonstrates the DRL scheme is more effective compared with the benchmarks.