Reflective Distributed Denial of Service Detection: A Novel Model Utilizing Binary Particle Swarm...

Area of Science:

• Computer Science
• Network Security
• Artificial Intelligence

Background:

• The proliferation of the internet has amplified network security concerns, necessitating robust Intrusion Detection Systems (IDSs).
• Traditional IDSs struggle with massive, high-dimensional data, leading to low efficiency and prolonged detection times.
• Effective feature selection (FS) is critical for enhancing IDS performance by reducing complexity and improving accuracy.

Purpose of the Study:

• To propose a novel feature selection (FS) approach, BPSO-SA, for optimizing intrusion detection systems (IDSs).
• To develop an advanced Distributed Denial of Service (DDoS) attack detection model using optimized machine learning algorithms.
• To improve the accuracy, precision, recall, F1-score, and reduce prediction time in network security.

Main Methods:

• A new feature selection (FS) approach, BPSO-SA, combining Binary Particle Swarm Optimization (BPSO) and Simulated Annealing (SA).
• Integration of the Gray Wolf Optimization (GWO) algorithm to optimize the LightGBM model's hyperparameters.
• Development of a reflective DDoS attack detection model leveraging the BPSO-SA and GWO-optimized LightGBM.

Main Results:

• The BPSO-SA algorithm effectively identifies optimal feature subsets, enhancing global search capabilities.
• The GWO algorithm optimizes LightGBM, significantly boosting detection performance.
• Experimental results demonstrate superior performance of the proposed model over conventional methods in key metrics.

Conclusions:

• The proposed BPSO-SA feature selection method combined with GWO-optimized LightGBM offers a highly effective solution for network intrusion detection.
• The developed reflective DDoS attack detection model exhibits strong resilience and generalization capabilities.
• This approach addresses the limitations of traditional IDSs in handling large-scale, high-dimensional network traffic data.