Table 10 Performance comparison of proposed work vs. published works.
Ref. No | Methods | Datasets | Performance Analysis | Limitations | |||
|---|---|---|---|---|---|---|---|
Accuracy (%) | Precision (%) | Recall (%) | F-score (%) | ||||
ESOML | UNSW-NB15 | 83.09 | 82.48 | 82.50 | 83.08 | Low accuracy in detecting intrusions | |
XGBoost-TCN | AWID | 93.96 | 64.36 | 66.22 | 65.27 | High computational efficiency | |
HBFL | IoT dataset | 97.89 | - | - | 94.80 | Does not detect sophisticated adversaries | |
BiGRU-DNN | NSLKDD, UNSWNB15, CICIDS2017 | 84.86 | 84.73 | 85.21 | 84.88 | High computational efficiency | |
FL | IoTID20, IoT-23, N-BaIoT | 98 | - | - | - | Low processing power | |
Meta-AdaboostM1 algorithm | UNSW-NB 15 | 90.25 | 86.14 | 94.59 | 86.95 | Low efficiency and scalability | |
PCCNN | NSL-KDD | 98.13 | - | - | - | High computational time | |
BFLIDS, CNN, BiLSTM | NSL-KDD | 96.02 | 96 | 95 | 96 | Difficult to detect complex patterns | |
Ours | Proposed ECapsNet with Blockchain-based Merkel Damgard Cryptographic algorithm | KDD-CUP 99 | 98.90 | 98.78 | 98.65 | 98.45 | Â |
UNSW-NB 15 | 98.78 | 98.74 | 98.54 | 98.32 | |||
