Immune-Inspired Intrusion Detection: An Ensemble Learning Approach using A multiple binary classifications with NSA Validation for WSN Attack Detection)

Abstract

Wireless Sensor Networks (WSNs) have become a vital part of modern applications, but their resource-constrained nature makes them vulnerable to various security threats, such as blackhole, grayhole, and flooding attacks. Conventional security measures are often impractical for WSNs because of their high computational demands and this problem has led to a growing need for lightweight and efficient intrusion detection systems. Many machine learning traditional models struggle with the imbalanced datasets and complex attack patterns found in WSN environments. This research addresses these challenges by introducing a new ensemble classification model uses decision tree detectors combined with the Negative Selection Algorithm (NSA) to convert the multi-class classification problem into multiple binary classifications. This unique approach enhances accuracy while ensuring the detectors are tolerant of "self-class" behavior, thereby reducing false alarms. The proposed model achieved an accuracy of 98.4% which showed that the proposed method significantly outperforms traditional algorithms, demonstrating its potential as a practical and robust intrusion detection solution for wireless sensor networks.