Hierarchical growing neural gas network (HGNG)-based semicooperative feature classifier for IDS in vehicular ad hoc network (VANET)

In this research, new modeling strategy based hierarchical growing neural gas network (HGNG)-semicooperative for feature classifier of intrusion detection system (IDS) in a vehicular ad hoc network (VANET). The novel IDS mainly presents a new design feature for an extraction mechanism and a HGNG-based classifier. Firstly, the traffic flow features and vehicle location features were extracted in the VANET model. In order to effectively extract location features, a semicooperative feature extraction is used for collecting the current location information for the neighboring vehicles through a cooperative manner and the location features of the historical location information. Secondly, the HGNG-based classifier was designed for evaluating the IDS by using a hierarchy learning process without the limitation of the fix lattice topology. Finally, an additional two-step confirmation mechanism is used to accurately determine the abnormal vehicle messages. In the experiment, the proposed IDS system was evaluated, observed, and compared with the existing IDS. The proposed system performed a remarkable detection accuracy, stability, processing efficiency, and message load.