An Enhanced VANET SDN Control Plane: Leveraging Hybrid Gaussian SVM-Dijkstra Approach

Fog networking has become a significant area of research, both theoretically and practically. With the emergence of technologies such as the Vehicular Ad hoc Network (VANET), vast amounts of data are being generated, and exchanging this data between Vehicle Edge Cloud (VEC) and vehicles can provide valuable information for collision warnings, traffic congestion, and route planning. To provide these services, a powerful VEC cloud architecture is required. Despite the high-performance storage and network resources offered by fog computing, it remains susceptible to internal network path congestion and link over-utilization. Additionally, the complexity of VEC architecture arises from operating protocols on multiple sub-networks, each with various configurations and performance levels. To address these challenges, this study proposes a route optimization approach that integrates the Software-Defined Network (SDN) architecture into the fog cloud network. SDN is a relatively new paradigm that allows centralized management of the network system by separating the networking functionality from the forwarding functionality. This separation provides complete network visibility with centralized control over all underlying fog nodes. Furthermore, the proposed SDN is optimized using machine learning capabilities that classify link labels using the Gaussian SVM algorithm. The zones with the least utilization are then optimized using the Dijkstra deterministic algorithm. Experimental results demonstrate that the Gaussian SVM algorithm achieves an 88.888 % classification accuracy rate, while the Dijkstra algorithm identifies optimal path. Combining these two algorithms enhances the intelligence of the SDN controller by classifying and optimizing the best path in the VANET cloud network. In summary, the proposed optimized architecture improves performance and reduces congestion and over-utilization in the VANET cloud network.