CICIoMT2024 + IoMT-TrafficData: a unified heterogeneous IoMT dataset for intrusion detection
This deposit is a machine-learning-ready, tabular release of unified Internet of Medical Things (IoMT) network traffic. It combines two public PCAP-derived benchmarks—CICIoMT2024 and IoMT-TrafficData—into one aligned window-level feature table produced with MIOTTA-NPT (W = 100 packets per row). Each...
| Autores: | , , , , |
|---|---|
| Tipo de recurso: | conjunto de datos |
| Fecha de publicación: | 2026 |
| País: | España |
| Institución: | Consorci de Serveis Universitaris de Catalunya (CSUC) |
| Repositorio: | CORA.Repositori de Dades de Recerca |
| OAI Identifier: | oai:dnet:cora.rdr____::4bee55d006679edb771f8d7dba8435e6 |
| Acceso en línea: | https://doi.org/10.34810/DATA3305 |
| Access Level: | acceso abierto |
| Palabra clave: | Computer and Information Science Engineering Medicine, Health and Life Sciences Internet of medical things (IoMT) Intrusion detection systems (IDS) Network traffic Dataset unification MIOTTA-NPT |
| Sumario: | This deposit is a machine-learning-ready, tabular release of unified Internet of Medical Things (IoMT) network traffic. It combines two public PCAP-derived benchmarks—CICIoMT2024 and IoMT-TrafficData—into one aligned window-level feature table produced with MIOTTA-NPT (W = 100 packets per row). Each record contains 58 numeric window statistics plus seven provenance/label fields. Labels are provided at three harmonised granularities: binary (Benign vs. Attack), six classes (Benign, DDoS, DoS, MQTT, Reconnaissance, Spoofing; MQTT-related attack names are collapsed into a single MQTT category), and 26 fine-grained sub-types. Rows retain source provenance (dataset, file, original attack metadata). The table is split into stratified train/validation/test files (70/15/15 at the six-class level; 8,011,534 rows in total). Feature values in the CSVs are z-scores from a StandardScaler fitted on the training split only. For exact downstream preprocessing and external evaluation on new PCAPs processed through MIOTTA-NPT, the publication bundles `scaler.joblib` (fitted StandardScaler) and `metadata.joblib` (feature column order, split row counts, harmonisation scheme tag, near-zero-variance QA list) from the same export run as the CSVs. Without the saved scaler, consumers can still train or analyse models on the provided CSVs as-is, but cannot faithfully re-apply the identical scaling transform to newly extracted raw feature matrices. |
|---|