Optimized intrusion detection using particle swarm optimization and neural networks in simulated and physical network...

Area of Science:

• Cybersecurity
• Machine Learning
• Network Security

Background:

• Intrusion detection systems (IDS) are crucial for network security.
• Optimizing machine learning workflows can improve IDS performance.
• Generating diverse datasets is essential for robust IDS evaluation.

Purpose of the Study:

• To present a reproducible framework for building and evaluating an optimized intrusion detection system.
• To investigate the impact of feature selection on machine learning model performance for intrusion detection.
• To compare different machine learning models for network intrusion analysis.

Main Methods:

• Collected network traffic from virtual and physical network environments.
• Utilized Wireshark for packet trace recording and merged with reference datasets.
• Applied Particle Swarm Optimization (PSO) for feature dimensionality reduction.
• Trained and evaluated multiple classification models (tree-based, distance-based, neural networks).

Main Results:

• PSO-selected features led to efficient and refined feature sets.
• The neural network model trained on PSO-selected features demonstrated superior performance.
• The optimized neural model exhibited stable detection behavior with a low false alert rate.

Conclusions:

• The proposed framework provides a reliable method for intrusion detection system development.
• Optimization-driven feature selection significantly enhances classifier reliability.
• The PSO-enhanced neural approach is validated as a robust solution for intrusion analysis.