Autonomous driving technology offers a promising solution to reduce road accidents, traffic congestion, and fuel consumption. The management of vehicular networks is challenging as it demands mobility, location awareness, high reliability and low latency of data traffic. In this paper, we propose a novel communication architecture for vehicular network with 5G Mobile Networks and SDN technologies to support multiple core networks for autonomous vehicles and to tackle the potential challenges raised by the autonomous driving vehicles. Data requirements are evaluated for vehicular networks with respect to number of lanes and cluster size, to efficiently use the frequency and bandwidth. Also, the network latency requirements are analysed, which are mandatory constraints for all the applications where real time end-to-end communication is necessary. A test environment is also formulated to evaluate improvement in vehicular network using SDN-based approach over traditional core networks.