Popularity Prediction Caching Based On Logistic Regression In Vehicular Content Centric Networks

To improve the network performance caused by mobility and sporadic connectivity in the vehicular network, vehicular content centric network (VCCN) is proposed by applying CCN into the vehicular network. The open in-network caching of CCN makes nodes cache contents cooperatively to facilitate information access. To improve the network performance such as access delay and hit ratio, road side units (RSUs) should try to cache more popular contents and provide better service for mobile users. This paper aims to propose a novel cache replacement policy - popularity prediction content caching (PPCC) for VCCN. In PPCC, we incorporate the future popularity of contents into our decision making. By learning the popularity of contents, we propose a cache replacement method based on logistic regression for RSUs in order to store those frequently access contents. The input data are related to the inherent characters of the received interests and the output is the predicted content popularity which guarantees that only popular contents are cached in the network infrastructures (i.e., RSUs). Simulation evaluations demonstrate that our scheme is very effective with higher cache hit, lower access latency and higher caching efficiency compared to other state-of-the-art schemes.