Edge-to-Fog Collaborative Computing in a Swarm of Drones

Recently, drones technology started unleashing numerous novel applications and services and became one of the killer applications for the Mobile Edge Computing (MEC) paradigm. The main challenge for real-time applications is, however, transferring data between data processing servers and drones, which yields considerable latency and energy consumption and reduces the autonomy and dynamics of operations. In this paper, we consider a swarm of heterogeneous and autonomous drones deployed to detect objects in run-time video streaming. Each drone can execute data-processing related computation tasks locally, or offload them to other nodes in the swarm. For efficient cooperation at the swarm level, forming a system of systems, drones need to cooperate in using each other’s resources and managing the communication and offloading operations. To overcome the mentioned challenges and also data privacy, we propose a novel edge-to-fog collaborative computing framework employing federated learning (FL)-based offloading strategy with a rating method. The proposed method also aims to offload the computation tasks between drones in a fair and efficient manner.