Description:1. Data Collection & Preprocessing
Healthcare sensors on mobile and wearable devices generate patient data (e.g., body temperature, blood pressure, heart rate, ECG signals). This data is preprocessed, including cleaning, normalization, and feature extraction, and stored for real-time analysis. The system utilizes datasets such as simulated IoT healthcare sensor data and the PTB-XL dataset for ECG records.
2. Edge-Assisted Fault Detection with ResNet
The system initializes a Deep ResNet model for fault detection, trained using the PTB-XL dataset for anomaly detection. This model is deployed on edge nodes for real-time healthcare data analysis. If an anomaly is detected, a potential fault is flagged, triggering adaptive scheduling. The integration of ResNet and Branch-and-Bound Learning (B2L) improves real-time healthcare data processing by preserving critical signal patterns for early fault detection and predictive maintenance.
3. Fault-Resilient Adaptive Scheduling (FRAS)
FRAS ensures continuous system functionality even with faults or failures, crucial for edge computing and real-time systems. Scheduling resources (edge, fog, cloud) are initialized, and reliability functions for each node are computed. Fault probability is predicted using historical data, and tasks are dynamically assigned based on reliability and workload. QoS metrics (latency, availability, reliability) are monitored, and task allocation is adjusted based on real-time feedback to mitigate fault impact.
4. Secure Data Processing with DSTE Encryption
The system employs Dynamic Stepwise Tiny Encryption (DSTE), a lightweight and quantum-secure encryption method. It dynamically updates encryption keys at each step using quantum-inspired randomness, preventing unauthorized access and minimizing overhead. Data from healthcare sensors is immediately encrypted on the device before transmission to fog nodes. DSTE protects data confidentiality and integrity by breaking it into smaller encrypted segments, increasing resilience to attacks without compromising processing efficiency. The encryption process is highly efficient, taking approximately 0.00003528 seconds
5. Quantum Portfolio Optimization (QPO) for Scheduling
QPO is utilized to optimize task-resource allocation, minimizing latency and maximizing efficiency. The task allocation problem is converted into a Quadratic Unconstrained Binary Optimization (QUBO) form, and quantum annealing is applied to find optimal resource allocation. This approach balances encryption strength, computational efficiency, and real-time performance, ensuring that encrypted tasks are optimally allocated to maintain real-time processing
6. Secure Data Transmission & Decryption
Encrypted data is transmitted using a fault-tolerant approach. After task execution, encrypted results are decrypted for final use, ensuring secure end-to-end processing without compromising real-time performance.
7. End-User Decision Making & Alert System
The system supports end-user decision-making by providing real-time, fault-tolerant scheduling and predictive analytics. ML-based predictive analytics are applied to generate alerts for critical patient conditions, notifying healthcare professionals for immediate intervention. The system processes complex data from various IIoHT sources, and after encryption, stores it in the cloud for real-time access to help predict emergencies, assess patient conditions, and tailor treatments effectively.
8. Continuous Monitoring and Adaptive Learning
The model ensures continuous monitoring and adaptive learning to improve system performance and model accuracy over time. System performance, fault rates, and QoS metrics are monitored, and the ResNet model is periodically retrained for improved accuracy. Scheduling policies are updated based on real-time feedback, and encryption parameters are adapted to balance security and efficiency.
, Claims:1. A secure and fault-resilient adaptive scheduling system for Industrial Internet of Healthcare Things (IIoHT) comprising:
a. Data Generation and Encryption Module: utilizing Dynamic Stepwise Tiny Encryption (DSTE) for secure task encryption and transmission of healthcare sensor data.
b. Edge-Assisted Fault Detection Module: integrating an Explainable Deep Residual and Bound Learning (ERBL) framework with Deep ResNet for real-time anomaly detection and optimized resource utilization.
c. Fault-Resilient Adaptive Scheduling (FRAS) Module: dynamically assigning healthcare tasks to edge, fog, or cloud nodes based on reliability, workload, and predicted fault probability.
d. Quantum Portfolio Optimization (QPO) Module: for efficient task-resource mapping and minimizing computational complexity in scheduling through quantum annealing
e. Secure Data Transmission and Decryption Module: ensuring end-to-end data integrity and real-time performance.
f. End-User Decision-Making and Alert System: providing real-time clinical decision support and generating alerts for critical patient conditions based on predictive analytics.
2. The system of claim 1, wherein the DSTE algorithm dynamically updates encryption keys at each step using quantum-inspired randomness and a tiny portfolio-based encryption model to enhance security and reduce computational overhead.
3. The system of claim 1, wherein the ERBL framework combines Explainable AI (XAI) with deep ResNet, trained using a branch and bound method, to provide transparency and interpretability in scheduling and resource management.
4. The system of claim 1, wherein the QPO module converts the task allocation problem into a Quadratic Unconstrained Binary Optimization (QUBO) form and applies quantum annealing to optimize resource allocation.
5. The system of claim 1, further comprising a continuous monitoring and adaptive learning mechanism to periodically retrain models, update scheduling policies, and adapt encryption parameters based on real-time feedback.