SDN-DDoS Traffic Dataset

In cybersecurity, understanding and mitigating Distributed Denial of Service (DDoS) attacks are paramount. Although public datasets offer valuable insights, they often lack the specific characteristics or scale necessary for comprehensive research. Hence, the generation of tailored datasets is imperative for addressing the limitations of public resources. This dataset serves as a crucial resource for evaluating and refining AI algorithms, particularly in the domains of machine and deep learning. By providing a diverse array of DDoS attack traffic scenarios, researchers can develop and validate models capable of detecting and mitigating such threats in real-time. Furthermore, the dataset encompasses two distinct network topologies, each comprising 12 switches and 24 hosts, orchestrated by a Ryu controller in a Software-Defined Networking (SDN) environment. This setup enables the simulation of complex network behaviors and generation of realistic traffic patterns that are reflective of actual deployment scenarios. The dataset includes a comprehensive set of features that are essential for characterizing network traffic dynamics during DDoS attacks. These features encompass critical parameters, such as source and destination IP addresses, packet and byte counts, duration metrics, flow characteristics, protocol details, port numbers, transmission rates, delay, jitter, packet loss rates, and labeled annotations. In total, our dataset comprised an extensive collection of 1,048,575 rows, ensuring a robust and diverse sample size for rigorous analysis and experimentation. This scale facilitates the exploration of nuanced patterns and behaviors across various attack scenarios and network configurations. Furthermore, our dataset adheres to ethical guidelines and privacy standards, ensuring the anonymization of sensitive information and protection of individual privacy rights. Researchers can leverage this dataset with confidence, knowing that it upholds ethical principles while advancing the state-of-the-art DDoS detection and mitigation strategies.

在网络安全领域，理解与缓解分布式拒绝服务（Distributed Denial of Service，DDoS）攻击是重中之重。尽管公开数据集能够提供宝贵的研究洞见，但此类数据集往往缺乏开展全面研究所需的特定特征与规模。因此，构建定制化数据集对于弥补公开资源的局限性而言至关重要。本数据集可作为评估与优化人工智能（Artificial Intelligence，AI）算法的关键资源，尤其适用于机器学习与深度学习领域。通过提供多样化的DDoS攻击流量场景，研究人员能够开发并验证可实时检测与缓解此类威胁的模型。此外，该数据集涵盖两种截然不同的网络拓扑结构，每种拓扑均包含12台交换机与24台主机，并在软件定义网络（Software-Defined Networking，SDN）环境中由Ryu控制器进行编排。该部署架构可模拟复杂的网络行为，并生成贴合实际部署场景的真实流量模式。数据集包含一套全面的特征集，可用于表征DDoS攻击发生期间的网络流量动态特征。这些特征涵盖多项关键参数，包括源与目的IP地址、数据包与字节计数、时长指标、流特征、协议详情、端口号、传输速率、延迟、抖动、数据包丢包率，以及标注信息。总体而言，本数据集包含总计1048575条数据记录，可为严谨的分析与实验提供充足且多样化的样本规模。该数据集的规模支持对各类攻击场景与网络配置下的细微模式与行为展开探索。此外，本数据集遵循伦理准则与隐私标准，实现了敏感信息的匿名化处理，并保障了个人隐私权利。研究人员可放心使用本数据集，因其既恪守伦理原则，又能推动DDoS检测与缓解技术的前沿发展。