A Lightweight Trust Mechanism with Attack Detection for IoT.

Area of Science:

• Computer Science
• Cybersecurity
• Internet of Things (IoT)

Background:

• Internet of Things (IoT) devices face trust evaluation challenges due to limited resources and susceptibility to trust attacks.
• Existing trust mechanisms may suffer from single-source bias and fail to adapt to dynamic device status changes.

Purpose of the Study:

• To propose a lightweight and adaptable trust mechanism for IoT devices.
• To address challenges of limited resources and sophisticated trust attacks in IoT environments.
• To enhance the accuracy and responsiveness of trust estimations.

Main Methods:

• Utilized Bayesian statistics and Jøsang's belief model for trust quantification, allowing multi-source feedback to mitigate bias.
• Implemented a forgetting algorithm to ensure trust estimations reflect the latest device status.
• Developed a 'tango' algorithm and hypothesis testing for trust attack detection and mitigation.

Main Results:

• The proposed mechanism provides accurate trust estimations even under combined and colluding attacks.
• The system effectively excludes malicious devices and rapidly restores trust for legitimate devices.
• The forgetting algorithm automatically adapts parameters for optimal performance, balancing accuracy and complexity.

Conclusions:

• The developed trust mechanism is robust, lightweight, and adaptable for secure IoT ecosystems.
• It successfully counters various trust attacks, ensuring reliable trust evaluations in dynamic environments.
• The approach offers a significant improvement over conventional methods for IoT trust management.