LSTM-Based QoS-Aware Proactive Flow-Rule Placement in Software-Defined Internet of Vehicles

With the rapid growth of mobile devices and bandwidth-intensive applications in the Internet of Vehicles (IoV), traffic and flow management among vehicular nodes poses a challenge to routing efficiency. In Software Defined Networks (SDN), the control plane is abstracted from the data plane, providing a centralized, programmable network control. The default flow-rule placement in SDN, being reactive, leads to suboptimal network performance, such as latency issues and packet loss. Therefore, there is a need for intelligent, proactive, and adaptive network traffic management systems to maximize network performance. This study introduces a Long Short-Term Memory (LSTM)-based Quality of Service (QoS)-aware proactive flow-rule placement in Software Defined Internet of Vehicles (SDIoV). First, by exploiting trends in vehicle movements, an LSTM model is trained to forecast the next-step Access Point (AP) linked to vehicles. Second, the SDN controller proactively installs QoS-aware flow rules in the predicted AP for classified traffic to meet its QoS requirements. These flow rules dynamically reroute traffic using OpenFlow rules and allow execution of incoming request actions without the controller's intervention. The proposed scheme demonstrates superiority over existing methods by combining SDN with machine learning to enable proactive, adaptive, and efficient traffic management in dynamic IoV networks.