Unified Multimodal Network Intrusion Detection Systems Dataset

The Unified Multimodal Network Intrusion Detection System (UM-NIDS) dataset is a comprehensive, standardized dataset that integrates network flow data, packet payload information, and contextual features, making it highly suitable for machine learning-based intrusion detection models. This dataset addresses key limitations in existing NIDS datasets, such as inconsistent feature sets and the lack of payload or time-window-based contextual features. UM-NIDS was created by processing raw PCAP files from four well-established datasets: CIC-IDS 2017, CIC-IoT 2023, UNSW-NB15, and a CIC-DDoS 2019 dataset. By offering both flow-level and payload-based data, along with contextual features that capture historical and temporal patterns in network traffic, UM-NIDS enables comprehensive analysis and robust model development. It is ideal for cross-dataset validation and supports flexible customization, allowing users to add new datasets or adjust configurations for their specific research needs.统一多模态网络入侵检测系统（Unified Multimodal Network Intrusion Detection System, UM-NIDS）数据集是一套全面且标准化的数据集，整合了网络流数据、数据包载荷信息与上下文特征，使其极适用于基于机器学习的入侵检测模型。该数据集解决了现有网络入侵检测系统（Network Intrusion Detection System, NIDS）数据集存在的核心局限，例如特征集不统一，以及缺乏载荷或基于时间窗口的上下文特征。UM-NIDS通过处理四个广受认可的成熟数据集的原始PCAP文件构建而成，这四个数据集分别为CIC-IDS 2017、CIC-IoT 2023、UNSW-NB15与CIC-DDoS 2019数据集。该数据集同时提供流层级与基于载荷的数据，并辅以可捕获网络流量历史与时间模式的上下文特征，能够支持全面的分析与稳健的模型开发。它非常适用于跨数据集验证场景，且支持灵活的自定义配置，允许用户根据自身研究需求添加新数据集或调整参数设置。

搜集汇总

背景与挑战

背景概述

该数据集是一个综合性的标准化网络入侵检测系统数据集，集成了网络流数据、数据包负载和上下文特征，解决了现有数据集中特征不一致和缺乏负载信息的问题。它来源于四个知名数据集，支持跨数据集验证和机器学习模型开发，适用于网络安全和异常检测研究。