SDCast: A Software-Defined Networking Based Clustered Routing Protocol for Vehicular Ad-Hoc Networks

The field of vehicular ad hoc networks (VANETs) has made significant advances in recent years by merging software-defined networking (SDN) with reinforcement learning (RL). VANET networking services may be remotely coordinated, and network traffic can be monitored by combining SDN with RL. However, multi-hop communication in VANETs is difficult due to inadequate connectivity, low network utilization, unexpected vehicle movement, and frequent disconnection. This study provides a novel VANET routing protocol known as SDCast, based on ad hoc on-demand distance vector routing that utilizes an SDN-based Q-learning algorithm and considers various constraints (link availability duration, link latency, and bandwidth). The protocol maintains vehicle coordination by employing a clustering architecture that considers the vehicles' relative motions and velocities. Furthermore, SDCast finds the best path by combining a probability distribution function with a global search approach. More stable clusters, reduced cluster transition rates, and a longer cluster head lifetime improve the current routing approach's inadequacies. Since SDCast does not rely on lower-level networking, it is portable and an excellent solution for VANET communication needs.