Table 12 SIoT integration types with purpose, strengths, and limitations.

¹⁴⁰

Blockchain + AI/ML + Cloud or Edge/Fog

Enhance trust and decision-making through verifiable data and intelligent analytics

Trust score prediction, intrusion detection, anomaly detection

Ensures immutable trust history + intelligent behavior learning

High complexity and model drift; blockchain latency; requires frequent model updates

¹⁴⁶

Cloud + Edge/Fog + AI/ML

Enable real-time anomaly detection with local edge processing (on Raspberry Pi) and cloud-based visualization

Smart home elderly monitoring system

Real-time ML on Raspberry Pi- Non-wearable passive sensing, Privacy-aware design with dashboard

Used simulated sensor data (not live), Caregiver mobile app not yet developed

¹⁴⁶

Digital Twin + Edge/Fog/Cloud + Blockchain

Enable cyber-physical mirroring, predictive analytics, and secure control

Smart manufacturing, predictive maintenance, secure twin control

Real-time digital mirroring, decentralized analytics and traceability

High synchronization overhead; model mismatch; digital twin setup cost

¹⁴²

Blockchain + Federated Learning + Edge-Fog-Cloud Computing

Privacy-preserving ECG anomaly detection and real-time decision making

Healthcare IoT; Remote cardiac monitoring with mobile/wearable ECG sensors

Low latency, enhanced privacy, decentralized model training, tamper-proof storage via smart contracts

Increased cost, execution time, energy use due to blockchain overhead; partial energy modeling only

¹⁴⁸

AI/ML + IoT

Improve waste classification accuracy using deep learning and optimized ensemble learning in IoT-enabled environments

Smart waste management in smart cities with automated waste sorting from images

Low-complexity model with 85% accuracy; CSO optimization improves performance; outperforms traditional models (SVM, XGBoost)

No explicit use of edge/cloud; real-time deployment and scalability not discussed; limited to image-only input