Adaptive QoE-Aware SFC Orchestration in UAV Networks: A Deep Reinforcement Learning Approach
Yao Wu, Ziye Jia, Qihui Wu, Zhuo Lu
Abstract
In the low altitude intelligent network (LAIN), unmanned aerial vehicles (UAVs) are extensively utilized to provide flexible communication and data transmission services. Besides, based on the network function virtualization technology, the service function chain (SFC) orchestration is an effective solution for optimizing communication performance and network adaptability in UAV networks. Hence, this paper investigates an adaptive SFC orchestration scheme for UAV networks in LAIN by defining and managing service pathways. Firstly, to quantify the quality of experience (QoE) of users, we employ the fuzzy analytic hierarchy process to construct a mathematical model to elucidate the relationship between the quality of service and QoE. Subsequently, we introduce the markov decision process model to capture the dynamic network state transitions, and then devise an algorithm of dueling double deep Q-network with regularization for adaptive online SFC deployment. Finally, we investigate the adaptability of deep reinforcement learning algorithms to resource constraints within the UAV network scenarios. Numerical results indicate that compared with the baseline algorithms, the proposed algorithm can enhance training stability, ensure the QoE of users, and optimize key indicators such as energy consumption and task completion.