<![CDATA[The threat of Distributed Denial of Service (DDoS) is increasing develop along with increasing use of the Internet of Things (IoT) and Software-Defined Networking (SDN) architecture . Although SDN provides convenience in management network , properties its centralized control make it prone to to flooding attacks that can paralyze controller performance . Detection method conventional , such as approach statistics and machine learning, still own limitations in matter accuracy , high false positive rate , and dependence on extracted features manually . To overcome problem said , research This propose a hybrid deep learning based DDoS detection and mitigation model that combines Convolutional Neural Network (CNN) to extraction feature spatial from RGB and Gated Recurrent Unit (GRU) images for understand temporal correlation between traffic data network . System tested through network test-bed Mininet based with Ryu/Floodlight controller, using simulation DDoS attacks (Hping3, LOIC) and normal traffic (video streaming, HTTP server). Traffic data cross recorded in PCAP format, processed become RGB image measuring 200×200 pixels, and labeled based on type traffic . Evaluation results with metric accuracy , precision, recall, F1-score, and MCC show that the CNN–GRU model has performance more superior compared to baseline approaches such as CNN-only, GRU-only, as well as classical ML methods such as SVM and Random Forest. In addition , the system capable apply mitigation adaptive through automatic flow rule creation on edge switches. Findings This confirm that effective deep learning- based spatial -temporal hybrid approach in increase detection early and response DDoS attacks on SDN networks adaptive and real-time. ]]>
Copyrights © 2025
