부산대학교 도서관

Unmanned aerial vehicle (UAV) provides a cost-efficient solution for corona detection in high-voltage substations. In this article, we address the integrated sensing, communication, and computing joint optimization problem. The objective is to maximize the average amount of data sensed by UAV through jointly optimizing sensing, i.e., acquisition frequency and image depth selection, communication, i.e., task splitting and power control, and computing, i.e., local computation resource allocation. We propose a dueling deep Q network (DDQN)-based integrated sensing, communication, and computing joint optimization algorithm for self-powered UAV named DESCANT to solve the problem. DESCANT explores DDQN to intelligently learn the optimal strategy under the complex electromagnetic environment of high-voltage substations. The historical information of electromagnetic interference (EMI) is incorporated into the state construction to improve convergence and learning optimality. DESCANT also realizes energy awareness through the dynamic adjustment of resource management based on energy consumption as well as harvested energy. DESCANT is compared with state-of-the-art algorithms and its superiority is verified through simulations.