Dalhousie NIMS Lab IMA Traffic Dataset 2025

This dataset presents encrypted network traffic flows captured from eight Instant Messaging Applications (IMAs), simulating group chat scenarios across multiple devices in a controlled cloud environment. Our methodology utilized a set of automated orchestration scripts to simulate realistic user behavior\u2014such as asynchronous conversations and app switching\u2014using Android Cuttlefish emulators. This approach allowed us to minimize background noise and isolate meaningful IMA communication traffic. The dataset consists of flow-level metadata collected from the following eight IMAs: Discord, Messenger, Rocket.Chat, Slack, Skype, Signal, Teams, and Telegram These flows were extracted from .pcap captures using Tranalyzer2, and include feature-rich session-level metadata useful for application identification, device fingerprinting, and user action classification (e.g., distinguishing group chat activity across applications). Each application\u2019s traffic is stored in a structured format and labeled with device and action metadata. The dataset supports both encrypted and TLS-SNI-disambiguated flows, allowing for flexible downstream analysis. Comprehensive details regarding our traffic generation, emulation architecture, and labeling methodology are provided in our accompanying paper. A structured README further explains the folder hierarchy, features, and filtering logic. This dataset serves as a valuable resource for researchers exploring encrypted traffic analysis, identity-aware network monitoring, and ML-based Zero Trust policy enforcement.