An Intelligent Caching Scheme Considering the Spatio-Temporal Characteristics of Data in Internet of Vehicles

The high mobility of vehicles in vehicular networks presents difficulties in timely and low-delay content delivery. Edge caching is a promising approach for overcoming these challenges. By properly caching data, we aim to improve service performance and reduce the load on the backhaul network. We propose two unique methods for caching data in the Internet of Vehicles (IoV), based on the division of IoV data into two groups: safety and infotainment contents. Due to the different nature of each type, two unique methods are proposed for caching each type of content. The Federated Learning-based Mobility-aware Collaborative Content Caching (FM3C) method is designed for infotainment content, while the Spatio-Temporal Characteristics Aware Emergency Content Caching (STAECC) method is designed for emergency content. To predict the popularity of infotainment content, we implement a Long Short-Term Memory (LSTM) model, trained through federated learning for user privacy protection. The predicted popularity is then combined with other content characteristics through a multi-criteria decision-making method to determine the most suitable content for caching in each Road-Side Unit (RSU). Our proposed cache-aware intelligent routing method, enabled by software-defined networks (SDN), allows for cooperation among RSUs to respond to requests and deliver content. Experimental results demonstrate the effectiveness of our proposed FM3C and STAECC methods in improving cache hit rate, reducing delay, and enhancing the quality of experience (QoE) for users. In conclusion, our proposed methods offer a promising solution for the efficient provision of content in vehicular networks.