Simultaneous Environment Sensing and Channel Knowledge Mapping for Cellular-Connected UAV

Abstract

Cellular-connected unmanned aerial vehicle (UAV), as a promising application of extending cellular service from ground to low-altitude three-dimensional (3D) airspace, has received significant attention recently. However, its practical realization faces some critical challenges, such as the noncontinuous 3D cellular coverage in the sky, as well as the complex physical and radio environment when operating in urban area. In this paper, by exploiting the UAV’s highly controllable mobility, we study the UAV trajectory design problem to minimize the weighted sum of mission completion time and expected communication outage duration, while ensuring obstacle avoidance in complex environment. The formulated problem involves intractable cost function and constraint, which can not be solved by standard optimization techniques. To this end, we first study the performance upper bound based on the Dijkstra’s shortest path algorithm under the ideal assumption that the perfect physical environment information and radio channel knowledge are available. For the practical scenario in the absence of such information, a novel framework with simultaneous environment sensing and channel knowledge mapping is proposed, which aims to construct both the physical environment and radio propagation maps to facilitate the reinforcement learning based path design. Numerical results show that the proposed technique can effectively avoid the coverage holes and physical obstacles, and approaches to the performance upper bound that assumes the perfect physical and radio environment information.