Timely data collection for UAV-based IoT networks: a deep reinforcement learning approach

With the increasing development of the Internet of Things (IoT), the number of sensor nodes is growing explosively. The future application systems have stricter requirements on the timely delivery of the data collected from the sensor nodes. For such applications, unmanned aerial vehicles (UAVs) can help to collect data from the sensor nodes (SNs) and then fly to the data center (DC) to deliver the data. UAVs have the advantages of rapid deployment, strong maneuverability and low cost. Compared with the method of multi-hop data transmission, the UAV can flexibly adjust its position to improve communication environment. This helps to save energy and extend the battery lifetimes of the nodes. In addition, by constructing the communication systems between UAVs and ground terminals, and between UAVs, this helps to satisfy the needs of network services in various scenarios in the future. These include reliable and safe communication in public areas, network enhancement in hotspots, data collection in smart cities and improving network coverage in remote areas, etc.