About this article
Published Online: Feb 05, 2025
Received: Sep 20, 2024
Accepted: Dec 30, 2024
DOI: https://doi.org/10.2478/amns-2025-0066
Keywords
© 2025 Jian Li, published by Sciendo
This work is licensed under the Creative Commons Attribution 4.0 International License.
Figure 1.
Figure 2.
Figure 3.
Figure 4.
Figure 5.
Bot-IoT multiple evaluation index
| Categories | Precision ratio | Recall rate | F1-score | Total accuracy |
|---|---|---|---|---|
| Normal | 99.00% | 99.81% | 99.81% | 99.02% |
| DDoS | 99.20% | 99.43% | 99.97% | |
| DoS | 99.02% | 99.71% | 99.13% | |
| Reconnaissance | 99.99% | 99.29% | 99.88% | |
| Theft | 96.77% | 95.19% | 95.52% |
Network intrusion detection results
| Known type | |||
|---|---|---|---|
| Attack name | Recognition rate | False rate | Leakage rate |
| OOB | 94.41% | 3.25% | 2.34% |
| Land | 97.92% | 2.07% | 0.01% |
| Ping of death | 96.11% | 3.71% | 0.18% |
| Smurf | 96.20% | 1.93% | 1.87% |
| CGI | 95.04% | 3.70% | 1.26% |
| Finger Print | 96.96% | 1.02% | 2.02% |
| Unknown type | |||
| Attack name | Alarm rate | Leakage rate | |
| BACKDOOR | 91.39% | 8.61% | |
| DOS | 90.00% | 10.00% | |
| FTP | 89.34% | 10.66% | |
Typical network attack sample library
| Sample number | 1 | 2 | 3 | 4 | 5 | …… |
|---|---|---|---|---|---|---|
| Feature 1 | Tcp | Tcp | icmp | udp | icmp | …… |
| Feature 2 | 35 | 29 | 28 | 54 | 26 | …… |
| Feature 3 | 1 | 21 | 35 | 0 | 49 | …… |
| Feature 4 | 2164 | 529 | null | 126 | null | …… |
| Feature 5 | 92 | 1264 | null | 152 | null | …… |
| Feature 6 | null | null | 9 | null | 6 | …… |
| Feature 7 | null | null | 0 | null | 5 | …… |
| Feature 8 | A | F | null | null | null | …… |
| Feature 9 | 419 | 52 | 76 | 164 | 72 | …… |
| Feature 10 | Get-cgi | $ i/n | 1a | 3c | 3f | …… |
| Categories | 0 | 2 | 1 | 3 | 1 | …… |
The number of TON-IoT data set categories
| Numbering | Categories | Sample number |
|---|---|---|
| 0 | Bengin | 1126548 |
| 1 | Scanning | 718192 |
| 2 | Xss | 703991 |
| 3 | DDoS | 504635 |
| 4 | Password | 470633 |
| 5 | DoS | 411088 |
| 6 | Injection | 370432 |
| 7 | Backdoor | 4526 |
| 8 | Mitm | 1897 |
| 9 | Ransomware | 692 |
The number of Bot-IoT data set categories
| Numbering | Categories | Sample number |
|---|---|---|
| 0 | Normal | 516 |
| 1 | DDoS | 1847593 |
| 2 | DoS | 1956425 |
| 3 | Reconnaissance | 89452 |
| 4 | Theft | 95 |
Comparison of the IoT multiple subclass results
| Model | Precision ratio | Recall rate | F1-score | Accuracy rate |
|---|---|---|---|---|
| ViT | 83.30% | 79.59% | 85.92% | 87.19% |
| MobileNetV2 | 81.35% | 77.03% | 80.71% | 92.55% |
| Xception | 79.30% | 81.28% | 79.85% | 95.39% |
| EfficientNetB0 | 80.89% | 80.64% | 81.09% | 89.02% |
| DenseNet121 | 84.93% | 81.88% | 80.60% | 87.52% |
| This article | 97.85% | 95.62% | 89.64% | 99.06% |
Comparison of the results of Bot-IoT multiple categories
| Model | Precision ratio | Recall rate | F1-score | Accuracy rate |
|---|---|---|---|---|
| Xception | 85.24% | 81.66% | 86.63% | 95.84% |
| EfficientNetB0 | 88.00% | 86.68% | 89.51% | 96.40% |
| DenseNet121 | 94.23% | 95.24% | 93.65% | 96.88% |
| FNN | 90.29% | 91.20% | 92.61% | 96.57% |
| TSODE | 99.38% | 97.33% | 95.44% | 97.28% |
| This article | 98.80% | 98.69% | 98.86% | 99.02% |