Researchers have proposed a new feature selection algorithm designed to improve the detection of distributed denial-of-service (DDoS) attacks. The method, known as the mathematical curve-based Trochoid Search Optimization (TSO) feature selection algorithm (ASTSO-STF), aims to address the significant challenges posed by the large-scale, sparse, and class-imbalanced nature of data associated with modern network traffic. By enhancing the efficiency of the feature selection process, the algorithm seeks to reduce the computational overhead typically required by complex detection models while simultaneously improving their ability to recognize malicious traffic patterns.
A primary difficulty in developing effective detection systems is the presence of vast quantities of redundant or irrelevant data features. These extraneous elements not only demand greater computational power but also hinder the ability of machine learning models to accurately identify specific attack behaviors. The proposed ASTSO-STF approach addresses this by refining how a model selects the most relevant features from complex datasets.
The technical foundation of ASTSO-STF involves combining three basic types of transfer functions—S-shaped, V-shaped, and U-shaped—with three established random walk models: Lévy flight, Brownian motion, and tangent flight. To further optimize the algorithm’s performance within the feature space, the research team introduced a dynamically changing attenuation coefficient. This coefficient is intended to enhance the exploration and development capabilities of the algorithm as it navigates through potential feature sets, allowing for more robust identification of indicators related to DDoS activity.
A notable innovation in this study is the replacement of the traditional TSO search trajectory with eight distinct mathematical curves, each defined by unique geometric characteristics. By comparing these various trajectories, the researchers aimed to strengthen the algorithm’s capability to search the feature space more effectively. The study ultimately concluded that the implementation of an Archimedean spiral-based trajectory provided the most effective search mechanism for the ASTSO-STF structure.
To validate the effectiveness of the proposed method, the researchers conducted comprehensive experiments using nine large-scale sparse datasets. They also created simulated attack scenarios with varying rates—specifically 2.5%, 4%, and the original attack rates—using the CIC-DDoS2019 and CIC-IDS-2017 datasets. These tests provided a controlled environment to compare the performance of ASTSO-STF against six other intelligent optimization algorithms commonly used for feature selection in network security.
The experimental results indicate that ASTSO-STF demonstrates superior classification performance and greater overall competitiveness in the majority of the tested scenarios. Evaluations performed across several metrics, including classification accuracy, attack detection capability, and the reduction of false alarms and missed detections, suggest that the algorithm can successfully isolate features most relevant to DDoS attack identification.
Subject of Research: Cybersecurity and network attack detection algorithms.
Article Title: Mathematical curves based TSO feature selection algorithm with decaying perturbation dynamic transfer function for DDoS attack detection
Article References: Jin, S.-Y., Sun, W.-Z., Wang, J.-S., Xu, Z.-R., Li, S., Wang, Y.-C., Song, Y.-W., & Qi, Y.-L. (2026). Mathematical curves based TSO feature selection algorithm with decaying perturbation dynamic transfer function for DDoS attack detection. Journal of Cloud Computing. https://doi.org/10.1186/s13677-026-01004-2
Image Credits: AI Generated
DOI: 10.1186/s13677-026-01004-2
Keywords: DDoS detection, feature selection, network security, cybersecurity, optimization algorithms, ASTSO-STF, machine learning, trochoid search optimization, data sparsity, network traffic analysis, classification performance, computational efficiency, intelligent optimization, attack mitigation, dynamic transfer functions
Cite Scienmag News
Reid Dalton. (November 18, 2026). New Algorithm Improves DDoS Attack Detection Using Mathematical Curve-Based Optimization. Scienmag. https://scienmag.com/new-algorithm-improves-ddos-attack-detection-using-mathematical-curve-based-optimization/
Reid Dalton. "New Algorithm Improves DDoS Attack Detection Using Mathematical Curve-Based Optimization." Scienmag, 18 November 2026, https://scienmag.com/new-algorithm-improves-ddos-attack-detection-using-mathematical-curve-based-optimization/. Accessed 18 November 2026.
Reid Dalton. "New Algorithm Improves DDoS Attack Detection Using Mathematical Curve-Based Optimization." Scienmag. November 18, 2026. https://scienmag.com/new-algorithm-improves-ddos-attack-detection-using-mathematical-curve-based-optimization/








