Computation Offloading and Trajectory Definition in UAV-MEC-based 5G Networks

UAV-based MEC (MEC-UAV) is a solution that provides computing resources close to users in 5G networks when ground-based infrastructure is not available, enabling computation offloading services. However, since the UAV is a battery-powered device, both the application (e.g., latency) and the UAV's energy consumption are impacted by computational and flight activities. Therefore, a careful analysis of the energy and performance tradeoff is required during the dimensioning and operation of UAV-MEC systems. This paper presents a MEC-UAV model that incorporates the data return and fusion at the user device, as well as considers errors that may occur during data transmission and processing at MEC-UAV, which have been overlooked in previous works. Utilizing the model, we conducted an analysis of energy consumption, latency, and the percentage of users served by MEC-UAV from the perspectives of computation offloading and trajectory optimization. The results emphasize the importance of considering these features when configuring the MEC-UAV system and defining supported applications.