Deteksi Serangan DDoS (Distributed Denial of Service) Menggunakan Wavelet Decomposition dan Optimasi Hyperparameter Berbasis Optuna

Ghibran, Andi Khalil
Mustikasari, Mustikasari
Darmatasia, Darmatasia
Antamil, Antamil

TIN: TERAPAN INFORMATIKA NUSANTARA
Vol 6 No 7 (2025): December 2025

This study aims to design and develop a Distributed Denial of Service (DDoS) attack detection system based on Wavelet Decomposition capable of identifying network traffic anomalies in real-time. The main problem addressed is the high false positive rate in conventional detection methods, which often fail to distinguish between legitimate traffic bursts and actual attacks. Two primary data sources were used: the CICIDS2017 dataset and self-generated data representing controlled DDoS attack patterns. The proposed method applies Discrete Wavelet Transform (DWT) to decompose network traffic signals into amplitude and energy components. Detection is then performed using the Median Absolute Deviation (MAD) approach, optimized with three parameter search methods: Grid Search, Random Search, and Optuna. Experimental results indicate that the energy-based method with Optuna optimization achieves the best performance, with an accuracy of 98.6% on the CICIDS2017 dataset and 99.4% on the self-generated data, and error rates of 1.4% and 0.6%, respectively. This research contributes to enhancing the accuracy of DDoS detection systems with low computational overhead, making it suitable for large-scale network environments.