Dynamic quality of service prediction with synergistic effect in IoV based on spatiotemporal graph convolutional simple recurrent unit network

The comprehensive popularization of 5G technology has facilitated the evolution of service-centric Internet of Vehicles, establishing it as an important means to meet the diverse demands of future intelligent transportation systems (ITS). With the proliferation of in-vehicle devices, accurately selecting the most appropriate services has emerged as a critical challenge in service-oriented ITS. Quality of Service (QoS) prediction in cellular vehicle communication, as an effective solution, can dynamically adjust service quality based on spatiotemporal changes in the IoV environment to ensure users consistently receive optimal service. However, existing QoS prediction schemes fail to fully capture the impact of dynamics and synergy among heterogeneous devices. To address these issues, we propose a D ynamic Q oS P rediction model with S ynergistic E ffect (DQPSE) that integrates the spatiotemporal graph convolutional network and bidirectional enhanced simple recurrent unit (Bi-ESRU) network. The spatiotemporal graph convolutional network is equipped with a weighted synergistic adjacency matrix to enhance the expressiveness of the graph structure and fully mine the effective feature information. Bi-ESRU effectively integrates spatiotemporal feature information to accurately capture complex spatiotemporal dynamic changes and their dependencies. In addition, DQPSE employs the optimization algorithm with enhanced Borges derivative to strengthen the model’s generalization ability. Numerous elaborate experiments conducted on real-world QoS datasets evaluate the superiority of our DQPSE model.