Table 3 Comparative analysis of recent IoT intrusion detection frameworks.
From: A lightweight framework to secure IoT devices with limited resources in cloud environments
Framework | Approach | Acc. (%) | Mem. (MB) | Time (ms) | Energy (W) | FPR (%) | Key Innovation |
|---|---|---|---|---|---|---|---|
Proposed | Leaf-Cut DT + Cloud-Edge | 98.2 | 12.5 | \(<1\) | 0.45 | 0.8 | Adaptive resource-aware optimization |
Altunay et al.57 | CNN-LSTM Hybrid | 99.1 | 180 | 2500 | 3.2 | 0.5 | Deep learning fusion |
Sinha et al.58 | Advanced LSTM-CNN | 98.9 | 220 | 3200 | 4.1 | 0.6 | Attention mechanism |
Karunamurthy et al.59 | Federated Learning IDS | 97.8 | 45 | 150 | 1.8 | 1.2 | Distributed learning |
Olanrewaju et al.60 | FL-based Deep Learning | 96.5 | 35 | 200 | 1.5 | 1.8 | Privacy-preserving |
Almotairi et al.61 | Ensemble ML Models | 97.2 | 25 | 50 | 0.9 | 1.5 | Feature selection |
Ghaffari et al.62 | Lightweight Security | 96.8 | 22 | 80 | 0.7 | 2.1 | Model optimization |
Asaithambi et al.63 | Blockchain-Edge Computing | 97.5 | 40 | 120 | 1.2 | 1.4 | Distributed processing |
Oliveira et al.64 | IoIT Edge ML | 96.9 | 30 | 100 | 0.8 | 1.8 | Multi-level detection |
Traditional SVM | Support Vector Machine | 94.2 | 60 | 300 | 2.1 | 3.2 | Statistical learning |
Neural Network | Multi-layer perceptron | 95.8 | 80 | 250 | 2.8 | 2.8 | Nonlinear mapping |
