Health, Security and Fire Safety Process Optimisation Using Intelligence at the Edge.

Area of Science:

• Computer Science
• Electrical Engineering
• Data Science

Background:

• Exponential growth in sensor networks generates vast data, straining constrained infrastructure and causing delays.
• Real-time data loss and temporal hierarchy issues hinder critical applications in health, security, and safety.
• Existing solutions struggle to manage large-scale sensor data efficiently on limited networks.

Purpose of the Study:

• To develop a multisensor fusion framework for temporally synchronous, high-resolution sensor data acquisition.
• To enhance energy conservation in sensor networks through intelligent microcontroller activation.
• To enable real-time data analysis and sensor customization at the edge for critical domains.

Main Methods:

• Developed a multisensor fusion framework incorporating domain-specific "wake up" triggers.
• Utilized low-power, model-driven microcontrollers with machine learning (TinyML) models.
• Investigated anomaly detection for real-time insights and energy-efficient edge data analysis.

Main Results:

• Achieved temporally synchronous, high-resolution data from multisensor systems.
• Demonstrated significant energy conservation through optimized microcontroller activation.
• Successfully applied edge computing for real-time sensor data analysis and customization.
• Validated the framework in a healthcare scenario with generalized benefits for other domains.

Conclusions:

• The proposed multisensor fusion framework effectively addresses challenges in real-time data management for critical applications.
• Energy-efficient edge analysis and TinyML integration are key to scalable and responsive sensor networks.
• The framework offers a robust solution for improving data quality and timeliness in health, security, safety, and fire domains.