Quantify the joint effect of mobility and urban environment on computation offloading to multi-UAV MEC network: Sojourn time

Abstract

The integration of Multi-Access or Mobile Edge Computing (MEC) with Unmanned Aerial Vehicles (UAVs) offers transformative prospects for 5G/6G networks, especially with compute offloading in urban settings. The combined effect of the mobility of airborne MECs (airMECs) and urban environmental dynamics has not been explored in current research, which has shown that the optimization strategy may be impractical in densely populated areas due to the restricted computational resources available on airMECs. In this paper, we analyze the combined mobility of users and airMECs while considering urban dynamics constraints to quantify sojourn time, which serves as a valuable input for mobility characterization in planning. We introduce novel analytical and statistical approaches to quantify sojourn time for both omnidirectional and directional antenna scenarios. We develop the statistical approach by simulating three-dimensional mobility in numerous urban configurations and gather ray-tracing line-of-sight data for two scenarios. The first scenario excludes buildings to verify the statistical approach against the analytical one, whereas the second scenario includes buildings to assess their influence. Following the quantification of the suggested environment-dependent sojourn time, we present a stochastic task size quantification as an application example. Additionally, we present many evaluations to ensure the accuracy and the practicality of the proposed models. For example, we analyze the circular directionality of sojourn time using the Von Mises probability distribution, examining task sizes with offloading time, and assessing handover times for a designated task size. The findings indicate that environmental dynamics significantly influence sojourn time and computational offloading, necessitating explicit consideration in airMEC network planning and MEC application design. This study offers critical insights for developing resilient, adaptive networks that support computation-intensive applications in dynamic urban environments.