Project publication · IEEE Internet of Things Journal
Summary
The article proposes model-based reinforcement learning for random access control in NB-IoT, aimed at improving NPRACH resource allocation.
Main contribution
The technique adjusts access parameters and coverage thresholds to balance success rate, congestion and available resources. The use of a model improves early-stage learning efficiency compared with purely empirical approaches.
Resultados y alcance
La evaluación muestra que el aprendizaje por refuerzo basado en modelo puede adaptar el acceso aleatorio de NB-IoT a cambios de carga y reducir colisiones y retrasos. La ventaja depende de la precisión del modelo, de la observabilidad del sistema y del coste de explorar políticas en condiciones no vistas.
Link with BoND1
Although NB-IoT is a cellular technology, massive access is a common problem across many IoT systems. For BoND1, the article contributes knowledge on adaptive resource control, a relevant element for designing scalable connectivity in networks with many devices.
Project publication IEEE Internet of Things Journal
Summary
The paper proposes a random access control for NB-IoT based on model-based reinforcement learning, aimed at improving NPRACH resource allocation.
Bibliometric indicators
- JCR area
- Computer Science, Information Systems
- JCR year
- 2025
- Scholar query
- 26 October 2026
Main contribution
The technique adjusts access parameters and coverage thresholds to balance success rate, congestion, and available resources. The use of a model improves learning efficiency in early phases compared to purely empirical approaches.
Results and scope
The evaluation shows that model-based reinforcement learning can adapt NB-IoT random access to load changes and reduce collisions and delays. The advantage depends on the accuracy of the model, the observability of the system, and the cost of exploring policies under unseen conditions.
Link with BoND1
Although NB-IoT is a cellular technology, the mass access problem is common to many IoT systems. For BoND1, the article provides knowledge about adaptive resource control, a relevant piece for designing scalable connectivity in networks of many devices.