Adaptive and stabilized real-time super-resolution control for UAV-assisted smart harbor surveillance platforms

Nowadays, there are active research for deep learning applications to smart cities, e.g., smart factory, smart and micro grids, and smart logistics. Among them, for industrial smart harbor and logistics platforms, this paper proposes a novel two-stage algorithm for large-scale surveillance. For the purpose, this paper utilizes drones for flexible localization, and thus, the algorithm for scheduling between multiple drones and multiple multi-access edge computing (MEC) systems is proposed under the consideration of stability in this first-stage. After the scheduling, each drone transmits its own data to its associated MEC for enhancing the quality and then eventually the data will be used for surveillance. For improving the quality, super-resolution is used. In the second-stage algorithm, the self-adaptive super-resolution control is proposed for time-average performance maximization subject to stability, inspired by Lyapunov optimization. Based on data-intensive simulation results, it has been verified that the proposed algorithm achieves desired performance.