Scholarlink

<p>Just like traditional network environments, cloud environments are also subjected to volumetric Distributed Denial of Service (DDoS) attacks. These attacks can overwhelm cloud services and cause service outages and disruptions by excessively increasing traffic volume. On the other hand, these attacks can occur in a sophisticated manner and at exceptionally high machine speeds, which makes it difficult for cloud defense systems to detect and prevent them in a timely manner. To address this challenge, this study aims to combine the sensitivity of the signature-based volumetric DDoS attack detection model with the agility of a multi-agent system. In this study, a multi-agent-driven defense system (CDA-CLOUD) using a Long Short-Term Memory (LSTM) model to detect and mitigate volumetric DDoS attacks in the public cloud environment is designed. To further enhance this approach, the system integrates federated learning (FL), enabling agents to learn from their local environments while maintaining data privacy and security. AES encryption is used to secure agent communications, while blockchain technology is integrated to ensure transparency and integrity by recording agent actions in a decentralized, tamper-proof ledger. In the context of this research, an LSTM-based model (LSTM-CLOUD) with a high accuracy rate of 99.89% on the CICDDoS2019 dataset is developed, particularly by using volumetric data obtained from the dataset. However, its ability to be used effectively in real-time events is crucial. The CDA-CLOUD system ensures this with minimal human intervention through autonomous agent collaboration.</p>

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

A multi-agent-based DDoS detection and defense system design with federated learning and blockchain in public cloud network environment

  • 期刊论文

  • 发表日期：

    2025年10月17日

查看全文
  • Hakan Aydin,
  • Muhammed Ali Aydin

摘要

Just like traditional network environments, cloud environments are also subjected to volumetric Distributed Denial of Service (DDoS) attacks. These attacks can overwhelm cloud services and cause service outages and disruptions by excessively increasing traffic volume. On the other hand, these attacks can occur in a sophisticated manner and at exceptionally high machine speeds, which makes it difficult for cloud defense systems to detect and prevent them in a timely manner. To address this challenge, this study aims to combine the sensitivity of the signature-based volumetric DDoS attack detection model with the agility of a multi-agent system. In this study, a multi-agent-driven defense system (CDA-CLOUD) using a Long Short-Term Memory (LSTM) model to detect and mitigate volumetric DDoS attacks in the public cloud environment is designed. To further enhance this approach, the system integrates federated learning (FL), enabling agents to learn from their local environments while maintaining data privacy and security. AES encryption is used to secure agent communications, while blockchain technology is integrated to ensure transparency and integrity by recording agent actions in a decentralized, tamper-proof ledger. In the context of this research, an LSTM-based model (LSTM-CLOUD) with a high accuracy rate of 99.89% on the CICDDoS2019 dataset is developed, particularly by using volumetric data obtained from the dataset. However, its ability to be used effectively in real-time events is crucial. The CDA-CLOUD system ensures this with minimal human intervention through autonomous agent collaboration.