Figure 1.
Figure 2.
Figure 3.
Figure 4.
Figure 5.
Figure 6.
Figure 7.
Figure 8.
Average Throughput (%) under different node densities
| Nodes | FLERCF | HSR | EESR | pvalue (FLERCF vs HSR) | p-value (FL-ERCF vs EESR) |
|---|---|---|---|---|---|
| 50 | 96.2 | 84.1 | 79.3 | <0.01 | <0.01 |
| 250 | 91.8 | 82.5 | 76.4 | <0.01 | <0.01 |
| 500 | 87.5 | 83.0 | 78.2 | <0.05 | <0.01 |
Performance under attack conditions (20% malicious nodes)
| Metric | FLERCF | HSR | EESR | pvalue (FLERCF vs HSR) | pvalue (FLERCF vs EESR) |
|---|---|---|---|---|---|
| Throughput drop (%) | 12 | 20 | 28 | <0.01 | <0.01 |
| PDR drop (%) | 9 | 15 | 21 | <0.01 | <0.01 |
| Overhead increase (%) | 14 | 23 | 31 | <0.01 | <0.01 |
Simulation Environment
| Parameter | Value / Range |
|---|---|
| IoT Nodes | 50,100,200,300, 400,500 |
| Simulation Area | 500 m × 500 m |
| Initial Energy per Node | 0.6 Joules |
| Simulation Time | 200 seconds |
| Communication Range | 50 m |
| Node Deployment | Random Uniform |
| Mobility | Static |
| Routing Protocols Evaluated | FL-ERCF, EESR, HSR |
| Attackers (Malicious Nodes) | 10% and20% oftotal nodes |
| FL Round Interval | Every 10 simulation seconds |
| FL Model | Decision Tree Classifier (Scikit-learn) |
| FL Optimizer | FedAvg (FederatedAveraging) |
| Security Technique | ECDH with Digital Certificates (DCESC) |