A Comprehensive Survey of Deep Learning for Authentication in Vehicular Communication

In the rapidly evolving landscape of intelligent transportation systems, the security and authenticity of vehicular communication have emerged as critical challenges. As vehicles become increasingly interconnected, the need for robust authentication mechanisms to safeguard against cyber threats and ensure trust in an autonomous ecosystem becomes essential. On the other hand, using intelligence in the authentication system is a significant attraction. While existing surveys broadly address vehicular security, a critical gap remains in the systematic exploration of Deep Learning (DL)-based authentication methods tailored to these communication paradigms. This survey fills that gap by offering a comprehensive analysis of DL techniques—including supervised, unsupervised, reinforcement, and hybrid learning—for vehicular authentication. This survey highlights novel contributions, such as a taxonomy of DL-driven authentication protocols, real-world case studies, and a critical evaluation of scalability and privacy-preserving techniques. Additionally, this paper identifies unresolved challenges, such as adversarial resilience and real-time processing constraints, and proposes actionable future directions, including lightweight model optimization and blockchain integration. By grounding the discussion in concrete applications, such as biometric authentication for driver safety and adaptive key management for infrastructure security, this survey bridges theoretical advancements with practical deployment needs, offering a roadmap for next-generation secure intelligent vehicular ecosystems for the modern world.