QoS-Aware UAV Coverage path planning in 5G mmWave network

Conventional UAV coverage path planning problems assume that UAVs offload sensor data only after the flight is completed. This type of "record-offload-process" approach incurs long delays and has serious risks of data loss. Emerging 5G networks enable high-throughput and low-latency network connectivity, allowing UAVs to transmit sensor data to the receiver during the flight. However, it is challenging to meet the network QoS requirements for such "online" data transfer during a flight because the performance of a 5G network may vary significantly due to channel conditions, structural blockage and movement of UAVs. In this paper, we first present a model of this new UAV path planning problem with network QoS constraint, which turns out to be a nonlinear optimization problem. We then transform the problem into two subproblems and derive solutions. We leverage an integrated full-stack simulation framework to study the end-to-end performance of UAV data transfer over a 5G mmWave network and evaluate our proposed solution. The experimental results show that our solution achieves significant reduction of data losses as well as improvement on flight time.