Scholarlink

When it comes to networking, software-defined networking (SDN) is evolving to be versatile as the industry moves away from conventional networking and toward automation. When it comes to protecting networks, data is always king. Disruptive attacks like these are more common in single-controller SDNs. Recent studies have shown that Distributed Denial-of-Service (DDoS) attacks are among the most disruptive kinds of network attacks. The goal of this study is to show how well the deep neural network model can analyze the dataset’s properties to identify packets that contain DDoS attacks. Here, we used Mininet and RYU Controller to mimic an SDN in order to create a real-time DDoS dataset. After that, both regular packets and packets representing a User Datagram Protocol (UDP) SYN Flood DDoS assault were simulated and sent to the virtual SDN. We used Rapid Miner to model a Deep Neural Network (DNN). An overall accuracy of 96.31% was found in the examination of the DNN model’s performance measures. Doing simulations of SDNs with numerous controllers might be part of future development. Further, by fine-tuning the machine learning technique, the research can be extended.

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

Deep Neural Network Model for UDP SYN Flood Distributed DoS Attack Detection in SDN

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • V. Mohan,
  • B. K. Madhavi,
  • S. B. Kishor

摘要

When it comes to networking, software-defined networking (SDN) is evolving to be versatile as the industry moves away from conventional networking and toward automation. When it comes to protecting networks, data is always king. Disruptive attacks like these are more common in single-controller SDNs. Recent studies have shown that Distributed Denial-of-Service (DDoS) attacks are among the most disruptive kinds of network attacks. The goal of this study is to show how well the deep neural network model can analyze the dataset’s properties to identify packets that contain DDoS attacks. Here, we used Mininet and RYU Controller to mimic an SDN in order to create a real-time DDoS dataset. After that, both regular packets and packets representing a User Datagram Protocol (UDP) SYN Flood DDoS assault were simulated and sent to the virtual SDN. We used Rapid Miner to model a Deep Neural Network (DNN). An overall accuracy of 96.31% was found in the examination of the DNN model’s performance measures. Doing simulations of SDNs with numerous controllers might be part of future development. Further, by fine-tuning the machine learning technique, the research can be extended.