Connected vehicles under information-centric architectures

Fast changing topologies, unpredictable network loads, potential broadcast storms, and identity-location conflation problems in IP networks all impose challenges on connected car system design. Prior approaches have proposed smart flooding using location information, carry and forward, and GeoServer assisted algorithms. Due to the complexity and overhead imposed by these approaches, their applications are restricted to local information transfer, dissemination of popular content, or delay-tolerant scenarios. In this work, we propose FastMF to extend both accessibility of the Internet for vehicular nodes and reachability of vehicular nodes from any remote server. Furthermore, by forming clusters of vehicles with similar mobility patterns, leveraging cluster to infrastructure links, and maintaining the mapping between node IDs and network addresses in a logically centralized server, we provide the nodes without a direct Internet connection with the benefit of an indirect association to an Internet gateway. Results from NS3 experiments illustrate the improvements in throughput for downloading large files derived by clustering and multi-hop transfer of data. In addition, experiments with interactive web-browsing scenarios indicate a significant improvement in delay in various mobility scenarios.