Learning-based Second Price Auction for Distributed Delivery Scheduling in Smart and Autonomous Urban Air Mobility

This paper proposes a novel service-oriented and learning-based autonomous control in urban air mobility (UAM). In this paper, we consider drone-taxi services which is one of major applications of UAM networks. The proposed algorithm designs joint truthful and distributed aerial drone-taxi scheduling which is based on second price auction (SPA). Furthermore, the SPA procedure is improved via deep learning framework for revenue-optimality. After the truthful and distributed scheduling, the passengers in the drone-taxi can utilize video streaming services where the services are under delay constraints. Therefore, in order to maximize video streaming quality under the delay constraints, the proposed algorithm is designed based on Lyapunov optimization framework. Simulation results verifies that the proposed algorithm achieves desired performance improvements.