An Intelligent Failure Detection on a Wireless Sensor Network for Indoor Climate Conditions

Area of Science:

• Computer Science
• Electrical Engineering
• Data Science

Background:

• Wireless sensor networks (WSN) are susceptible to data inaccuracies from environmental factors and sensor degradation.
• Effective functioning of WSNs necessitates intelligent failure detection mechanisms.

Purpose of the Study:

• To propose and validate a supervised learning method, Artificial Hydrocarbon Networks (AHN), for predicting remote temperatures and detecting sensor failures.
• To enhance the reliability and accuracy of data collected by WSNs.

Main Methods:

• Implementation of a supervised learning approach using Artificial Hydrocarbon Networks (AHN).
• Integration with a web service for remote temperature prediction and comparison with field sensors.
• Experimental validation using a small-scale WSN to assess failure detection, identification, and accommodation.

Main Results:

• The proposed AHN-based approach demonstrated accurate detection, identification, and accommodation of sensor failures.
• Experiments showed that 94.18% of testing data were recovered and accommodated, validating the method's effectiveness.

Conclusions:

• Artificial Hydrocarbon Networks (AHN) provide an effective solution for intelligent failure detection in wireless sensor networks (WSN).
• The method significantly improves data integrity and network reliability by accurately managing sensor failures.