Cluster-Based Cooperative Caching With Mobility Prediction in Vehicular Named Data Networking

Abstract

Vehicular named data networking (VNDN) is a networking architecture candidate to support various kinds of content-oriented applications in high dynamic topology environment. Its inherent in-network caching facilitates content delivery and the communication efficiency in the vehicular network. However, during the process of data delivery, fast and varying vehicle mobility changes the position of relay routers, which makes data packet delivered through the reverse path difficult. Furthermore, frequent link disruption and packet retransmission lead to the increase of network load and the decrease of user QoE. To address these problems, we propose a cluster-based cooperative caching approach with mobility prediction (COMP) in VNDN. The main idea of COMP is to establish communication among vehicles with similar mobility pattern to mitigate the impact of vehicle mobility, as the link between nodes with a similar pattern is relatively stable and reliable. Specifically, we design a clustering algorithm to group vehicles with similar mobility pattern via mobility prediction and present a cooperative caching to construct intra-cluster and inter-cluster communication over the vehicle clusters. To increase the cache resource utilization and the diversity of the cached data, we classify the cached data into the most popular data and the less popular data based on request frequency, and furthermore, the corresponding cache placement and transmission schemes are proposed. The evaluation results show that most of the vehicles (>95%) can acquire feasible and efficient data delivery via COMP, and COMP significantly improves network performance and user QoE.