Intelligent Framework Aims To Optimize Data Transfer in 5G Networks

A North Carolina State University researcher has developed technology designed to allow cellular communication nodes in 5G systems to partition bandwidth more efficiently in order to improve end-to-end data transmission rates. In simulations, the tech is capable of meeting the international goal of 10 gigabits per second in peak performance areas.

“End-to-end transfer means that the technology accounts for all of the connections between a data source and the end user,” says Shih-Chun Lin, an assistant professor of electrical and computer engineering at NC State and author of a paper on the work.

“My technology, incorporating both hardware and software, is a framework that takes into account data transfer rates, wired and wireless bandwidth availability, and the power of base stations – or eNodeBs – in a 5G network,” Lin says. “It then uses stochastic optimization modeling to determine the most efficient means of transferring and retrieving data – and it does this very quickly, without using a lot of computing power.”

Lin says that simulation testing of the framework is promising, and he and his research team are in the process of building a fully functional prototype.

“The prototype will allow us to conduct tests on a 5G testbed platform, since full-scale 5G networks are not yet online,” Lin says. “But simulation results suggest that we’ll be able to meet the 3GPP goal of 10 gigabits per second data transfer in peak coverage areas.

“We are currently seeking industry partners to work with us on developing, testing and deploying the framework to better characterize its performance prior to widespread adoption of 5G networks,” Lin says.

The paper, “End-to-End Network Slicing for 5G&B Wireless Software-Defined Systems,” will be presented Dec. 11 at IEEE GLOBECOM’18, being held in Abu Dhabi, UAE.

-shipman-

Note to Editors: The study abstract follows.

“End-to-End Network Slicing for 5G&B Wireless Software-Defined Systems”

Authors: Shih-Chun Lin, North Carolina State University

Presented: Dec. 11, IEEE GLOBECOM’18, Abu Dhabi, UAE

Abstract: As a key enabling technology for 5G&B systems, network virtualization allows multiple service providers to simultaneously and independently serve their users via virtualized network slices. However, this innovative technology cannot slice wireless resources without a paradigm shift in existing hardware-based architectures. In this paper, end-to-end network slicing is treated from a perspective of wireless software-defined networking architectures. It jointly optimizes all communication functionalities in both radio access and core networks to ensure optimal data throughput and congestion-free systems. First, based on software-defined cellular architectures, the idea of end-to-end (across access and core network domains) virtualization is introduced with dedicated control units, including high-level controllers and local baseband servers. Next, a stochastic utility-optimal virtualization problem is formulated, which jointly optimizes congestion control, flow routing, and power slicing to maximize the total incoming rates of wireless/wired flows, while satisfying the flow-queue stability and system-level constraints. After transforming the virtualization problem into a tractable form, an iterative network slicing algorithm is proposed that employs a primal-dual Newton method with quadratic convergence and achieves resource-efficient virtualization via control-unit coordination. Numerical results validate the efficacy of our solution, facilitating the 5G&B infrastructure-as-a-service.

This post was originally published in NC State News.