Table 1 The comparison of existing methods.
SI No | Methods | Advantages | Limitations |
|---|---|---|---|
1 | Ensemble LSTM-based IDS for ICTS12 | Effective in detecting cyber threats in V2V, IVN, and V2I networks Maximises data flow and enhances security Increases smart network efficiency | Requires large datasets for training Might not handle highly dynamic or unpredictable data well |
2 | AI-Based Anomaly Detection in CAN Systems13 | Improves vehicle data security using AI Identifies complex anomalies in CAN systems Focus on privacy | Limited by the scalability of the techniques Performance can vary depending on data quality and availability |
4 | AI-Based VANET Solutions14 | Enhances traffic safety and efficiency Optimises routing and driver awareness Improves passenger comfort and road experience | Potential difficulties in implementing AI in real-time traffic conditions High computational resources may be required for some AI techniques |
5 | AI-Based Security Solutions for Vehicular Networks15 | Identifies and addresses security issues in vehicular networks Proposes a new taxonomy for comparing AI-based solutions | May face challenges in handling highly dynamic and large-scale vehicular network data Complexity of integration with existing infrastructure |
6 | AI-IoT-5G/6G for VANETs16 | Combines AI, IoT, and 5G/6G to enhance connectivity and security Improves routing, mobility prediction, and driver awareness | 5G/6G infrastructure is still developing Potential privacy and security concerns when handling sensitive data across various systems |
7 | Hybrid CNN-GAN Model for Anomaly Detection17 | High detection rates and minimal false positives Improves network security by generating normal traffic patterns to detect anomalies | Require a large labeled dataset Complex training process and resource-intensive |
8 | ML-Based Security for EnFVs20 | Enhances safety with predictive maintenance and cyberattack detection Aims for privacy-preserving and real-time solutions | Real-world implementation challenges Ethical considerations and the complexity of applying ML techniques in resource-constrained environments |
9 | Transfer learning BILSTM21 | High detection accuracy for IoT botnet attacks Transfer learning improves generalisation | Potential dataset bias Computational overhead and resource constraints |
