부산대학교 도서관

Fog computing allows for energy-efficient and low-latency offloading of computationally intensive tasks from wireless devices to nearby servers. The integration of this technology in drone communications enables the managing of challenging tasks such as the ones found in remote areas with complex civil protection environments, such as disaster areas and emergency zones. In this paper, we propose a joint resource allocation scheme that optimizes both radio and computational resources for fog-assisted drone communication networks. Each drone decides whether to execute its task locally on its edge node or offload it to a fog node deployed on the base station (BS). Our scalable solution effectively minimizes service latency and energy consumption jointly, while taking into account physical- and application-layer constraints. Specifically, we allocate the CPU frequency capacity of both the local edge node and the remote fog node, as well as link bandwidth. Wireless channels to access the BS are limited, so only the most beneficial drones offload their tasks, while others use their local edge nodes. We formulate the power dissipation of various electronic circuits in the network using practical models. To develop the bi-objective minimization for each drone, we apply the Tchebysheff theorem, which derives the Pareto boundary between the two objectives (service latency and energy consumption). The competition among drones is modeled using the non-cooperative game framework, and the existence and uniqueness of the Nash equilibrium (NE) are proven. NE is computed using an algorithm based on subgradient projection. Numerical results concerning both theoretical aspects and a practical case study are presented to corroborate the efficiency of the proposed solution.