Distributed trust-driven intelligence for edge-level prediction in mobile industrial internet of things.

Area of Science:

• Computer Science
• Electrical Engineering
• Network Security

Background:

• The Industrial Internet of Things (IIoT) and edge computing are crucial for real-time distributed systems, enabling advanced industrial applications.
• Existing Tiny Machine Learning (TinyML) approaches for IIoT face challenges with computational complexity, limited resources, and security vulnerabilities in trustless environments.
• A significant research gap exists in achieving reliable communication and robust security against malicious devices within growing IIoT networks.

Purpose of the Study:

• To propose a predictive regression model for real-time data processing and autonomous decision-making in IIoT networks.
• To enhance energy efficiency and address security concerns in resource-constrained IIoT environments.
• To design secure edge-level communication ensuring a chain of trust and network integrity under unpredictable conditions.

Main Methods:

• Development of a predictive model utilizing regression for real-time data processing and autonomous decision-making.
• Implementation of secure edge-level communication protocols designed for unpredictable industrial environments.
• Validation through simulations comparing the proposed model against existing solutions using key performance metrics.

Main Results:

• Significant improvements in packet reception rate: 45% across IoT devices and 52% across attack levels.
• Substantial reduction in data latency: 37% across IoT devices and 44% across attack levels, even with malicious devices.
• Demonstrated enhancements in energy consumption, reduced complexity overhead, and a lower false-positive rate compared to baseline schemes.

Conclusions:

• The proposed predictive model effectively improves energy efficiency and real-time data processing in IIoT networks.
• Secure edge-level communication enhances network integrity and reliability, successfully mitigating security threats from malicious devices.
• The model offers a robust and efficient solution for advanced industrial applications, outperforming existing methods in critical performance metrics.