Scholarlink

The software-defined networking (SDN) architecture is now a life of all the networking frameworks operating on modern IoT devices that eases out the manual manipulations being done on the conventional networking structures comforting the humans as well as machines life too. This SDN framework always remained a target of suspicious activities because of its centralized control. The major security threat to this paradigm is Distributed Denial of Service (DDoS) attack, which ceases the entire networks access if not being anticipated, causing loses of money, time, reputation, information, etc. In a report, FS-ISAC confirmed a 22% increase of DDoS attacks towards financial organizations in the year 2022–2023. This work aims to classify the traffic as benign or DDoS maligned by utilizing two deep learning supervised algorithms, namely convolutional neural networks (CNN) along with long short-term memory (LSTM), ensemble together to fulfil the purpose. The study included the refined version of the CICDDOS2019 dataset that contains DDoS affected traffic along with normal traffic as well, comprising of 84 attributes, from which 21 SDN relevant features are selected with a statistical method and put into the model to classify the attack. Cross-validation is performed also, to improve the efficacy of the model. The results were achieved with an accuracy of about 99.81% and with a negligible loss of 0.36% when tested on the DDoS maligned SDN dataset, ensuring promising results for detecting the DDoS attack on SDN-based environment.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

CNN-LSTM Synergized Framework for Enhanced DDoS Attack Detection on SDN

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Abhishek,
  • Sanjeev Kumar,
  • Rajiv Singh,
  • Ayush Verma,
  • Manju Khari

摘要

The software-defined networking (SDN) architecture is now a life of all the networking frameworks operating on modern IoT devices that eases out the manual manipulations being done on the conventional networking structures comforting the humans as well as machines life too. This SDN framework always remained a target of suspicious activities because of its centralized control. The major security threat to this paradigm is Distributed Denial of Service (DDoS) attack, which ceases the entire networks access if not being anticipated, causing loses of money, time, reputation, information, etc. In a report, FS-ISAC confirmed a 22% increase of DDoS attacks towards financial organizations in the year 2022–2023. This work aims to classify the traffic as benign or DDoS maligned by utilizing two deep learning supervised algorithms, namely convolutional neural networks (CNN) along with long short-term memory (LSTM), ensemble together to fulfil the purpose. The study included the refined version of the CICDDOS2019 dataset that contains DDoS affected traffic along with normal traffic as well, comprising of 84 attributes, from which 21 SDN relevant features are selected with a statistical method and put into the model to classify the attack. Cross-validation is performed also, to improve the efficacy of the model. The results were achieved with an accuracy of about 99.81% and with a negligible loss of 0.36% when tested on the DDoS maligned SDN dataset, ensuring promising results for detecting the DDoS attack on SDN-based environment.