Table 4 Summary of FL experimental setup. The table details models configuration across single-dataset and combined multi-dataset scenarios. This setup ensures fair and consistent comparison of federated algorithms under varying data heterogeneity conditions.
From: Dataset-centric evaluation of federated intrusion detection models in IoT networks
Dataset | Number of clients | Partition method | Data distribution | Local epochs | Batch size | Learning rate (LSTM / transformer) |
|---|---|---|---|---|---|---|
Edge-IIoTset | 6 | By device/ application type | Moderate non-i.i.d. | 5 | 128 | 0.001 / 0.0005 |
CIC-IoT2023 | 10 | Device groups | Moderate non-i.i.d. | 5 | 128 | 0.001 / 0.0005 |
TII-SSRC-23 | 5 | Random stratified | Near i.i.d. | 5 | 128 | 0.001 / 0.0005 |
Combined (All) | 3 | Each dataset as client | Highly heterogeneous | 5 | 128 | 0.001 / 0.0005 |
