Joint Optimization Risk Factor and Energy Consumption in IoT Networks With TinyML-Enabled Internet of UAVs

The high mobility of Internet unmanned aerial vehicles (IUAVs) has attracted attention in the field of data collection. With the rapid development of the Internet of Things (IoT), more and more data are generated by IoT networks. IUAV-aided IoT networks can efficiently collect data in specific areas, which is of great significance in disaster relief. In the data collection task, it is necessary to plan the flight trajectory for the data collector— IUAV, so that the IUAV can collect data efficiently. However, existing research basically only considers the efficiency of data collection by IUAVs, but rarely considers the safety of IUAVs during flight. Therefore, this article proposes an IUAV trajectory planning algorithm that integrates energy efficiency and safety using local search to address the issues mentioned above. At the same time, a tiny machine learning (TinyML) algorithm is designed to assist the IUAV in making real-time decisions during flight. First, we build a general mathematical model that describes the risk in a particular region. Then, consider guiding the IUAV to a safer trajectory by introducing virtual nodes in the flight trajectory. Furthermore, we designed a local search algorithm for the three tasks of IUAV access sequence, IoT networks cluster heads selection, and virtual nodes selection and solved them through iterative optimization. We also consider the unreachable situation of the virtual nodes and use TinyML technology to help the IUAV adjust the position of the virtual nodes in real time in case of an emergency. In the end, an IUAV trajectory is obtained that can efficiently collect IoT networks’ data and fly safely. We have conducted a large number of simulation experiments to demonstrate the efficiency of the proposed algorithm compared to the baseline algorithm.