Table 36 VANET-DDoSNet++ Performance across different DDoS attack types.
From: A hybrid deep learning model for detection and mitigation of DDoS attacks in VANETs
Attack scenario | Accuracy | Precision | Recall | F1-Score | AUC-ROC | Remarks |
|---|---|---|---|---|---|---|
Standard DDoS Attack Types | ||||||
SYN Flood | 99.20% | 98.90% | 99.10% | 99.00% | 0.998 | High detection accuracy across TCP floods |
UDP Flood | 98.70% | 98.30% | 98.50% | 98.40% | 0.996 | Effective against stateless volumetric attacks |
ICMP Flood | 98.90% | 98.60% | 98.70% | 98.60% | 0.997 | Reliable against echo-based attacks |
HTTP GET Flood | 97.80% | 97.50% | 97.60% | 97.50% | 0.994 | Slightly lower due to payload variance |
HTTP POST Flood | 97.60% | 97.20% | 97.40% | 97.30% | 0.993 | Performance affected by irregular session traffic |
DNS Amplification | 98.30% | 98.00% | 98.10% | 98.00% | 0.996 | Detects high-reflection, low-bandwidth abuse |
ACK Flood | 98.50% | 98.20% | 98.30% | 98.20% | 0.996 | Efficient at tracking malicious ACK spikes |
Slowloris | 97.10% | 96.80% | 97.00% | 96.90% | 0.992 | Performance dips due to slow header injection |
NTP Amplification | 98.40% | 98.10% | 98.20% | 98.10% | 0.995 | Handles protocol abuse-based amplification |
Smurf Attack | 98.60% | 98.30% | 98.40% | 98.30% | 0.996 | Robust performance on broadcast address misuse |
Average (Standard DDoS) | 98.51% | 98.21% | 98.34% | 98.27% | 0.9953 | Consistently high across volumetric & protocol attacks |
VANET-Specific & Hybrid Attack Types | ||||||
Blackhole Attack | 98.92% | 98.85% | 98.91% | 98.88% | 0.998 | Common VANET threat, detected robustly |
Sybil Attack | 98.64% | 98.71% | 98.55% | 98.63% | 0.997 | Identity-based spoofing |
Replay Attack | 97.83% | 97.74% | 97.95% | 97.84% | 0.994 | Time-delayed data injection |
DoS + Blackhole (Hybrid) | 98.42% | 98.49% | 98.31% | 98.40% | 0.996 | Multiple-layer disruption |
Sybil + Replay + Timing (Multi-Vector) | 97.61% | 97.52% | 97.71% | 97.61% | 0.993 | Tests model adaptability to complex patterns |
Novel Pattern (Mobility Spoof + DoS) | 97.92% | 97.81% | 97.86% | 97.83% | 0.994 | Previously unseen hybrid variant |
Adversarial Drift Attack (Evolving Patterns) | 96.78% | 96.52% | 96.90% | 96.71% | 0.989 | Adaptive behavior handling |
Average (Hybrid/VANET-specific) | 98.16% | 98.09% | 98.17% | 98.13% | 0.9944 | Confirms model’s high adaptability to complex threats |
