Distributed Real-Time Multimodal Data Forwarding in Unmanned Aerial Systems

UAVs can support different applications, such as forest fire surveillance and precise agriculture. UAVs' service-on-demand preference drives the need for multiple UAVs to enhance surveillance coverage and data stability. Due to the limited capacity of the UAV, UAVs desire network coordination based on the importances of collected data. For example, UAVs at different locations may have different priorities of data forwarding tasks while the required data sizes of different tasks are varying. In this paper, we propose a system framework for UAV array to structure and prioritize the data forwarding based on i) forward-looking channel quality; ii) priorities of tasks of multimodal data on demand. Our proposed distributed real-time framework aims to optimize effective data throughput given a channel quality and effective scheduling of the channel usage among multiple UAVs. We conducted extensive evaluations using multiple UAVs and results show that our modeling of forward-looking channel quality prediction achieves 90% accuracy. Moreover, our scheduling algorithms can effectively optimize the overall data quality of forwarding tasks between UAVs and the base station.