RCoD: Reputation-Based Context-Aware Data Fusion for Mobile IoT

Area of Science:

• Computer Science
• Data Science
• Sensor Networks

Background:

• Mobile sensing technologies and the Internet of Things (IoT) are rapidly expanding.
• Ensuring data reliability and accuracy in mobile IoT systems is challenging due to inherent vulnerabilities.
• People-centric architectures in mobile IoT can lead to inaccurate or corrupted data, especially with malicious participants.

Purpose of the Study:

• To address the challenge of inaccurate data prediction in mobile IoT systems.
• To develop a mechanism resilient against malicious data injection and corruption attacks.
• To recover missing or imprecise data values from correlated data streams.

Main Methods:

• Proposing a Reputation-Based Context-Aware Data-Fusion (RCoD) mechanism.
• Utilizing a Contextual Hidden Markov Model for accurate real-time data prediction.
• Evaluating RCoD's resilience against on-off and data-corruption attacks.

Main Results:

• RCoD accurately identified honest participants even when the majority were malicious.
• The mechanism demonstrated resilience against various malicious data injection rates.
• Empirical evaluation using Beijing's air-quality dataset confirmed RCoD's superiority over state-of-the-art methods.

Conclusions:

• The proposed RCoD mechanism significantly improves data accuracy and reliability in mobile IoT environments.
• RCoD offers a robust solution for data fusion and prediction in the presence of adversarial participants.
• Context-aware data fusion combined with reputation systems is effective for secure and dependable IoT data.