Dempster-Shafer Theory for Modeling and Treating Uncertainty in IoT Applications Based on Complex Event Processing

Area of Science:

• Computer Science
• Artificial Intelligence
• Data Science

Background:

• The Internet of Things (IoT) generates vast event flows from connected devices and sensors.
• Sensor data in IoT applications can be unreliable due to failures, poor calibration, or inherent inaccuracies, leading to uncertainty.
• Complex Event Processing (CEP) is crucial for analyzing these event flows, but uncertainty in primitive events and derived rules poses a significant challenge.

Purpose of the Study:

• To investigate the identification and treatment of uncertainty within CEP-based IoT applications.
• To propose a novel approach, DST-CEP, for effectively handling data uncertainty in IoT environments.
• To demonstrate the capability of DST-CEP in combining unreliable and conflicting sensor data for accurate complex event detection.

Main Methods:

• Development of DST-CEP, an architectural model integrating Dempster-Shafer Theory (DST) to manage uncertainty in IoT events and their propagation.
• Implementation of DST-CEP within a multi-sensor fire outbreak detection system case study.
• Experimental evaluation using a real-world sensor dataset and standard performance metrics (Accuracy, Precision, Recall, F-measure, ROC Curve).

Main Results:

• DST-CEP successfully combines unreliable and conflicting sensor data, yielding accurate results despite data uncertainties.
• The approach demonstrated promising performance across key metrics, including Accuracy, Precision, Recall, F-measure, and ROC Curve analysis.
• The experimental results validate the suitability and flexibility of DST-CEP in addressing uncertainty in CEP-based IoT systems.

Conclusions:

• DST-CEP provides a robust framework for managing uncertainty in IoT data streams within CEP systems.
• The Dempster-Shafer Theory integration enables effective handling of conflicting sensor information, enhancing the reliability of complex event detection.
• DST-CEP is a flexible and effective solution for improving the accuracy and trustworthiness of IoT applications facing data uncertainty.