A hybrid metaheuristic algorithm with machine learning for detecting denial-of-service attacks in wireless sensor...

Area of Science:

• Computer Science
• Network Security
• Artificial Intelligence

Background:

• Wireless Sensor Networks (WSNs) are vulnerable to denial-of-service (DoS) attacks due to their inherent security limitations in unattended environments.
• Effective intrusion detection systems are crucial for safeguarding WSNs against various DoS attack vectors.
• Existing defense mechanisms often struggle with the dynamic and resource-constrained nature of WSNs.

Purpose of the Study:

• To develop and evaluate a novel hybrid optimization technique for detecting DoS attacks in WSNs.
• To enhance the accuracy and efficiency of intrusion detection systems in WSN environments.
• To address the challenge of feature selection for optimal DoS attack classification.

Main Methods:

• A hybrid Grasshopper Optimization Algorithm-Genetic Algorithm (GOA-GA) was proposed for optimal feature selection from WSN datasets.
• Machine learning classifiers, including K-Nearest Neighbors (KNN), were employed for traffic classification (normal vs. DoS attacks).
• The GOA-GA technique was validated against standard performance metrics, comparing its effectiveness with existing methods.

Main Results:

• The GOA-GA hybrid approach, when combined with the KNN classifier, achieved a high accuracy of 95.51%.
• The model demonstrated a recall of 95.51%, indicating strong performance in identifying DoS attack instances.
• The proposed method effectively reduced feature dimensionality and computational overhead compared to other approaches.

Conclusions:

• The GOA-GA hybrid optimization strategy provides a robust and efficient solution for DoS attack detection in WSNs.
• This research contributes to advancing information security by offering an improved method for WSN intrusion detection.
• The findings highlight the potential of metaheuristic algorithms in optimizing feature selection for enhanced network security.