Table 5 Ablation study – feature contribution analysis.
From: A hybrid deep learning model for detection and mitigation of DDoS attacks in VANETs
Configuration | Accuracy (%) | F1-Score | FPR (%) | Observation |
|---|---|---|---|---|
All features (Baseline) | 99.18 | 0.992 | 0.78 | Highest performance with full feature set |
Traffic statistics | 96.32 | 0.961 | 3.81 | Decline in early-stage attack differentiation |
Spatiotemporal features | 96.87 | 0.967 | 3.26 | Affects detection of mobility-based anomalies |
Deep traffic embeddings (GCN) | 94.91 | 0.945 | 4.58 | Major drop; GCNs critical for structural attack insights |
Behavioral features (Bi-LSTM) | 95.62 | 0.952 | 4.14 | Weakens detection of time-pattern-based attacks |
