Research Article
Design and Development of an Efficient Network Intrusion Detection System Using Machine Learning Techniques
Table 14
Comparison of the hybrid NID-Shield NIDS with existing approaches in this study.
| | Accuracy | TPR | FPR | TNR | Precision | Recall | -measure | MCC | ROC | PRC | Time (seconds) |
| Proposed approach with NSL-KDD 20% dataset | 99.90 | 0.9990 | 0.007 | 0.993 | 0.999 | 0.999 | 0.999 | 0.992 | 1.000 | 1.000 | 13.785 |
| Proposed approach with UNSW-NB15 dataset | 99.89 | 0.9989 | 0.006 | 0.993 | 0.999 | 0.989 | 0.997 | 0.992 | 1.000 | 1.000 | 318.15 |
| Neha et al. [26] | 99.05% | 0.994 | 0.014 | _ | 0.991 | _ | _ | _ | _ | _ | _ |
| Arif et al. [28] | 96.65% | 0.9271 | 0.136 | _ | 0.9998 | _ | _ | _ | _ | _ | _ |
| Ahmed et al. [29] | _ | 0.9577 | _ | 0.975 | 0.5662 | _ | _ | _ | _ | _ | 3112.87 |
| Tirtharaj [30] | _ | 0.9526 | _ | _ | _ | _ | _ | _ | _ | _ | 103.70 |
| Yao et al. [31] | 99.20% | 0.6699 | _ | _ | 0.9655 | 0.967 | _ | _ | _ | _ | _ |
| Suad et al. [32] | _ | _ | _ | _ | _ | _ | _ | _ | 0.995 | 0.962 | 10.79 |
| Ijaz et al. [33] | 99.8% (DoS) | _ | 0.17 (DoS) | _ | _ | _ | _ | _ | _ | _ | _ |
| Alauthaman et al. [34] | 99.20% | 0.9908 | 0.75 | _ | _ | _ | _ | _ | _ | _ | _ |
| Venkataraman and Selvaraj [35] | 83.83% | _ | _ | _ | _ | _ | _ | _ | _ | _ | 0.23 |
| Kumar and Kumar [36] | 99% | _ | _ | _ | _ | _ | _ | _ | _ | _ | _ |
| Cavusoglu [37] | 99.86% (overall) | 0.9292 (overall) | 0.000035 (overall) | _ | _ | _ | 0.706 (overall) | 0.954 (overall) | _ | _ | 10.62 (overall) |
| Saxena et al. [38] | 98.1% | 0.7 | _ | _ | _ | _ | _ | _ | _ | _ | _ |
| Kambattan and Rajkumar [39] | 99.45% | _ | _ | _ | _ | _ | _ | _ | _ | _ | _ |
| Kar et al. [40] | 93.95% | 0.955 | 0.1034 | _ | _ | _ | _ | _ | _ | _ | _ |
| Mishra et al. [41] | 92.12% | 0.971 | _ | _ | _ | _ | _ | _ | _ | _ | _ |
| Dutta et al. [42] | 91.29% | _ | _ | _ | 92.08% | 90.64% | 0.91 | _ | _ | _ | _ |
| Latah and Toker [43] | 84.29% | _ | 0.063 | _ | _ | 77.18% | 84.83% | _ | _ | _ | _ |
| Sumaiya Thaseen et al. [44] | 98.45%, on NSL-KDD dataset and 96.44% on UNSW-NB15 dataset | 0.9294 on NSL-KDD dataset and 0.504 on UNSW-NB15 dataset | _ | 0.9438 on NSL-KDD dataset and 0.984 on UNSW-NB15 dataset | _ | _ | _ | _ | _ | _ | 500 on NSL-KDD dataset and 1023 on UNSW-NB15 dataset |
| Safaldin et al. [45] | 96% | 0.96 | 0.03 | _ | _ | _ | _ | _ | _ | _ | 69.6 h |
| Vallathan et al. [46] | 98.4% | 0.9602 | _ | 0.998 | _ | _ | _ | _ | _ | _ | _ |
|
|