A hybrid trust computing approach for IoT using social similarity and machine learning

Area of Science:

• Computer Science
• Cybersecurity
• Artificial Intelligence

Background:

• The proliferation of Internet of Things (IoT) devices introduces significant data privacy and security risks.
• Autonomous decision-making in future IoT necessitates robust trust computing and prediction mechanisms.
• Current IoT security models require enhancement to address user data privacy concerns effectively.

Purpose of the Study:

• To develop and evaluate a hybrid approach for computing trust in social IoT scenarios.
• To enhance the security and privacy of user data within interconnected IoT ecosystems.
• To propose a dynamic aggregation method for computed trust values.

Main Methods:

• A hybrid approach combining distributed computation and global machine learning is proposed.
• The method incorporates social similarity and user ratings within a cloud-based architecture.
• Dynamic aggregation techniques are employed to combine different trust computations.

Main Results:

• The proposed hybrid approach demonstrates superior performance compared to existing related work.
• Machine learning integration offers a slight performance advantage over purely computational models.
• Both methods effectively mitigate malicious ratings without necessitating complex algorithms.

Conclusions:

• The developed hybrid trust computation model enhances security and privacy in social IoT.
• The approach successfully addresses the challenge of malicious ratings in distributed environments.
• Future IoT systems can benefit from trust computing and machine learning for autonomous operations.