Wireless Sensor Networks Intrusion Detection Based on SMOTE and the Random Forest Algorithm.

Area of Science:

• Computer Science
• Network Security
• Data Science

Background:

• Wireless sensor networks (WSNs) are crucial for military and environmental monitoring but face significant security challenges.
• Data transmission in WSNs is vulnerable to malicious attacks due to the absence of physical security measures.
• Effective intrusion detection systems are vital for safeguarding WSNs.

Purpose of the Study:

• To propose an intrusion detection method for WSNs that addresses the challenge of imbalanced datasets.
• To improve the performance of intrusion detection systems in WSNs.

Main Methods:

• Utilized the Synthetic Minority Oversampling Technique (SMOTE) to balance the intrusion detection dataset.
• Employed the Random Forest algorithm for training the intrusion detection classifier.
• Conducted simulations on a benchmark intrusion dataset to evaluate the proposed method.

Main Results:

• The Random Forest algorithm achieved an accuracy of 92.39% on the benchmark dataset.
• Applying SMOTE to oversample minority class samples improved the accuracy to 92.57%.
• The proposed SMOTE-Random Forest approach demonstrated superior performance compared to other algorithms.

Conclusions:

• The combination of SMOTE and Random Forest offers an effective solution for handling class imbalance in WSN intrusion detection.
• The proposed method significantly enhances the performance and accuracy of intrusion detection systems in wireless sensor networks.