Securing Intelligent Vehicular Networks Using AI-Driven Federated Learning

Analysis of Detour Index in Graph Joins and Special Graph Structures Vehicular networks are crucial for modern transportation safety and efficiency. However, in adverse weather conditions such as fog, reduced visibility severely disrupts vehicle-to-vehicle (V2V) communication, increasing the risk of chain accidents. Traditional centralized approaches for managing vehicular communication suffer from high latency, limited scalability, and data privacy concerns, making them unsuitable for real-time decision-making in dynamic environments. To address these challenges, this project proposes a Federated Learning Framework for Securing Vehicular Networks Using AI, enabling decentralized, privacy-preserving, and real-time threat detection in vehicular networks. The proposed system leverages Artificial Intelligence (AI) and Federated Learning (FL) to enhance situational awareness and predict collision risks. A deep learning-based object detection model, utilizing the YOLO algorithm, identifies obstacles, vehicles, and hazardous conditions in low-visibility environments. Edge detection techniques further improve visual clarity and detection accuracy. Federated Learning enables vehicles to collaboratively train AI models on local data without sharing raw information, ensuring data privacy and adaptability to real-world scenarios. Secure vehicle-to-vehicle (V2V) communication enhances decision-making speed, reducing latency in critical situations. By integrating AI-driven predictive analysis, federated learning, and decentralized data processing, this system significantly improves vehicular safety in challenging weather conditions. It ensures real-time hazard detection, reduces the likelihood of chain accidents, enhances data privacy, and scales effectively for large vehicular networks. This innovation marks a critical step toward intelligent, autonomous, and resilient transportation systems in smart cities.