CRII: CNS: IoT-aware Dynamic Spectrum Sharing in 5G C-RAN with Reinforcement Learning

The Internet of Things (IoT) is expected to revolutionize many key sectors of everyday life, such as transportation, health, power grid, and smart cities operation. To facilitate its deployment, 3GPP has proposed novel technologies, such as NarrowBand-IoT (NB-IoT) and LTE cat-M, characterized by i) enhanced coverage; ii) low spectrum requirements, iii) simple and cheap radio chipsets; iv) reduced energy consumption; v) increased redundancy.

A key advantage to their adoption is the fact that NB-IoT and LTE-M can be deployed on the same spectrum and can use the same base stations as cellular technologies (e.g., LTE and 5G). However, massive deployment of IoT networks can have a potentially disruptive effect on cellular infrastructures, thus degrading the quality of service by human users.

The objective of this project is to further research on novel techniques to improve the ability to share the available radio resources between traditionally human-centric technologies and IoT devices. The project involves undergraduate and graduate students in research, and organizes a special session within the Buffalo Day for 5G and IoT, an event that brings together critical stakeholders from academia, industry and municipalities to discuss the latest advancements in this domain.

The objective of this project will be attained by i) providing a comprehensive IoT-aware framework to characterize traffic in mobile networks, ii) formulating a dynamic spectrum sharing model for the 5G C-RAN based on Reinforcement Learning. The two thrusts will be developed following the guidelines by 3GPP in terms of IoT traffic use cases and C-RAN specifications.

Moreover, real data on the position of IoT devices and of network base stations will be employed to increase the fitness to reality of the proposed methodology. Real data, retrieved from publicly available sources, will be used in this research in combination with random distributions, such as Poisson Point Processes. A web-based framework for the collection and sharing of spatial and temporal data on IoT and human traffic will be deployed.

The framework will be made publicly available and will provide researchers in wireless networks and IoT with (i) realistic scenarios for their analytical models, (ii) input data for their network simulations/emulations research, and iii) training data for their machine-learning algorithms.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.