Comparative Analysis of Intrusion Detection Models on Internet of Vehicles Using TensorFlow Neural Network Classifiers

Connected vehicles amplify both capability and attack surface. This study trains TensorFlow neural classifiers with multiple optimisers to separate normal traffic from generic attacks and exploits in IoV-style network data.

The comparisons examine normal traffic, generic attacks, and exploits separately. The practical point is the difference between classes: a strong score on normal traffic cannot stand in for a strong result on the attacks that matter.

This earlier study provides context for the later federated IoV work. It focuses on classifier and optimiser comparisons; the later work asks how training can be distributed across devices without pooling all raw records in one place.