Page 125 - ITU Journal Future and evolving technologies Volume 2 (2021), Issue 4 – AI and machine learning solutions in 5G and future networks

ITU Journal on Future and Evolving Technologies, Volume 2 (2021), Issue 4

Fig. 11 – Accuracy and recall with different features

Fig. 12 – Training time with different features

Table 5 – Comparison of experimental results of different machine learning methods

Method XGBoost LightGBM
Evaluation Criteria Precision Recall F‑measure Precision Recall F‑measure
1: node‑down 1.00 1.00 1.00 1.00 1.00 1.00
3: interface‑down 0.99 1.00 0.99 0.99 0.97 0.98
5, 7: tap‑loss (delay) 0.88 1.00 0.93 0.87 1.00 0.93
9: ixnetwork‑bgp‑injection 1.00 1.00 1.00 1.00 1.00 1.00
11: ixnetwork‑bgp‑hijacking 1.00 0.71 0.83 1.00 0.70 0.82
Method Random Forest Decision Tree
Evaluation Criteria Precision Recall F‑measure Precision Recall F‑measure
1: node‑down 1.00 1.00 1.00 1.00 0.86 0.92
3: interface‑down 0.97 1.00 0.99 0.82 0.89 0.85
5, 7: tap‑loss (delay) 0.88 0.97 0.92 0.85 0.73 0.78
9: ixnetwork‑bgp‑injection 1.00 1.00 1.00 1.00 1.00 1.00
11: ixnetwork‑bgp‑hijacking 0.94 0.72 0.82 0.58 0.76 0.66
Method SVM MLP
Evaluation Criteria Precision Recall F‑measure Precision Recall F‑measure
1: node‑down 0.97 0.97 0.97 1.00 0.83 0.91
3: interface‑down 0.92 0.65 0.76 0.92 0.61 0.73
5, 7: tap‑loss (delay) 0.68 0.99 0.81 0.71 1.00 0.83
9: ixnetwork‑bgp‑injection 0.99 0.54 0.70 1.00 0.65 0.79
11: ixnetwork‑bgp‑hijacking 1.00 0.58 0.73 0.97 0.65 0.78

© International Telecommunication Union, 2021 109