Optimization-Driven DRL for Resource Allocation Under Licensed and Unlicensed UAV Spectrum Sharing Networks Against Uncertain Jamming

作者：Rui Ding, Fuhui Zhou, Quan Wu, Kai-Kit Wong, Naofal Al-Dhahir · 发表于：IEEE Transactions on Mobile Computing · 年份：2026 · DOI：10.1109/tmc.2026.3673261 · 被引用次数：1 · 研究领域：UAV Applications and Optimization、Security in Wireless Sensor Networks、Advanced Wireless Communication Technologies

Unmanned aerial vehicle (UAV) communication is of crucial importance for heterogeneous practical wireless communications. However, it is susceptible to the severe spectrum scarcity with the rapidly expanding market of wireless broadband, multimedia users, and high data-rate applications. Exploring the underutilized unlicensed spectrum through spectrum sharing is promising to tackle this issue, but the openness of the unlicensed spectrum makes UAVs susceptible to security threats from potential jammers. Therefore, a licensed and unlicensed UAV spectrum sharing network against uncertain jamming attack is studied. Moreover, to overcome the high complexity of the pure model-based optimization resource allocation schemes, the low learning efficiency and strong data dependency of data-driven deep reinforcement learning (DRL) methods, a novel optimization-driven DRL framework is proposed for the resource allocation. In particular, a model-based optimization module is exploited to derive the worst-case lower bound and a better informed target value of the formulated complex non-convex optimization problem. Furthermore, the model-based informed target value is integrated into the DRL to guide the agents for better strategies. Simulation results demonstrate that our proposed scheme can significantly improve the convergence speed and achieve a better reward performance than the pure DRL based scheme. It is also shown that the exploitation of the unlicensed spectrum can achieve approxima...