Project publication · Frontiers in Computer Science
Summary
The article presents TRUST Lab, a CICFlowMeter dataset for evaluating intrusion detection systems in IoT and edge scenarios with unambiguous labels.
Main contribution
The main contribution is a capture methodology based on single-class sessions, with benign traffic or one isolated attack family in each window. The result includes 4.6 million bi-flows and 15 attack families, avoiding temporal contamination between classes.
Resultados y alcance
El conjunto TRUSTLab separa de forma explícita tráfico benigno y malicioso y ofrece flujos CICFlowMeter obtenidos en un entorno operativo reproducible. Las pruebas de clasificación muestran su utilidad para evaluar IDS de IoT y borde, al tiempo que el alcance queda limitado a los servicios, ataques y condiciones de red incluidos en el banco de ensayos.
Link with BoND1
For BoND1, the work strengthens the security and trust dimension in advanced IoT infrastructures. The dataset makes it easier to compare lightweight IDS models deployable at the edge, a useful element for non-cellular and B6G networks with autonomous-operation requirements.
Project publication · Frontiers in Computer Science
Summary
The paper presents TRUST Lab, a CICFlowMeter dataset for evaluating intrusion detection systems in IoT and edge scenarios with unambiguous labels.
Bibliometric indicators
- JCR area
- Computer Science, Interdisciplinary Applications
- JCR year
- 2025
- Scholar query
- 26 October 2026
Main contribution
The main contribution is a single-class session-based capture methodology, with benign traffic or an isolated attack family in each window. The result brings together 4.6 million biflows and 15 attack families, avoiding temporary contamination between classes.
Results and scope
The TRUSTLab suite explicitly separates benign and malicious traffic and provides CICFlowMeter flows obtained in a reproducible operating environment. Classification tests are useful for evaluating IoT and edge IDS, while limiting the scope to the services, attacks, and network conditions included in the testbed.
Link with BoND1
For BoND1, the work reinforces the dimension of security and trust in advanced IoT infrastructures. The dataset makes it easy to compare lightweight and edge-deployable IDS models, a useful piece for non-cellular and B6G networks with autonomous operation requirements.
UPCT Repository: https://hdl.handle.net/10317/24191