Joint Caching and Recommendation in Vehicular Networks with Federated Graph Learning

To address the high dynamics and diverse user preferences in edge caching systems, this paper proposes a personalized recommendation and cache optimization strategy based on location-aware federated graph learning. Specifically, a Bi-directional Long Short-Term Memory(Bi-LSTM) model is first employed to predict the real time user locations. Subsequently, federated graph learning is utilized to locally construct and train the user preference model, effectively capturing the diversity of individual preferences. By integrating location predictions with user preference models, personalized recommendation lists for pre-cached contents are generated, improving recommendation accuracy while preserving user privacy. To optimize both the recommendation and caching strategies, a one-to-one matching algorithm and a greedy algorithm are applied in an alternating optimization procedure. Simulation results demonstrate that the proposed approach significantly improves the recommendation accuracy, reduces the average content transmission latency, and enhances the overall caching performance.