Energy optimization for wireless sensor network using minimum redundancy maximum relevance feature selection and...

Area of Science:

• Wireless Sensor Networks (WSN)
• Energy Conservation
• Machine Learning

Background:

• Energy conservation is a critical challenge in wireless sensor networks (WSN).
• Optimizing energy consumption is essential for extending network lifetime and operational efficiency.

Purpose of the Study:

• To apply feature selection techniques for minimizing sensor count and conserving energy in WSN.
• To evaluate the performance of different classifiers in conjunction with feature selection for WSN energy optimization.

Main Methods:

• Utilized Minimum Redundancy Maximum Relevance (MRMR) for sensor feature selection.
• Compared Support Vector Machine (SVM), Naïve Bayes, and k-nearest neighbors (KNN) classifiers.
• Assessed classifier accuracy and impact on sensor lifetime extension factor.

Main Results:

• MRMR effectively ranked sensors by significance, enabling reduction in sensor numbers.
• The k-nearest neighbors (KNN) classifier demonstrated superior performance compared to SVM and Naïve Bayes.
• Simulation results indicated an improvement in the lifetime extension factor of sensors.

Conclusions:

• Feature selection using MRMR is a viable strategy for energy conservation in WSN.
• KNN classifier is recommended for WSN applications requiring high accuracy and extended sensor lifetime.