Network-side ML for QoE-aware video resource allocation

Technology
In development
University
Scouted & summarized by Halo AI

A machine learning-based solution for 5G O-RAN architectures that enables mobile network operators to estimate and optimize video streaming Quality of Experience (QoE) using only network-side data. The system supports high-resolution video up to 4K and improves resource allocation by predicting user experience without needing application-level data.

Overview

This research presents a machine learning-based solution designed to enable mobile network operators to estimate the Quality of Experience (QoE) of video streaming sessions directly from network-side data. By integrating this model into the 5G Open Radio Access Network (O-RAN) architecture, the system allows for QoE-aware resource allocation without requiring direct access to application-level data from service providers. This approach facilitates proactive, human-centric network management, improving user experience for both fixed and mobile devices.

Technical specifications
  • Architecture: Designed for integration into the 5G O-RAN framework, utilizing the Non-Real Time RAN Intelligent Controller (Non-RT RIC) for offline training and the Near-RT RIC for real-time inference.
  • Machine Learning: Employs supervised regression models, with Gradient Boosting demonstrating the highest performance (R2 = 0.986, RMSE = 0.091).
  • Inputs: Relies exclusively on network and mobility-related parameters, such as user channel quality variations, device position, and network load.
  • Capabilities: Supports video streaming with resolutions up to 4K and frame rates up to 60 fps.
  • Workflow: The system uses a closed-loop control mechanism where the model estimates QoE to inform dynamic Radio Resource Management (RRM) or handover processes.
Technology readiness level

The technology is at level 4 (Component and/or breadboard validation in laboratory environment). The solution has been developed and evaluated through simulation-based performance testing using supervised machine learning models, as documented in academic research published in 2025. It is currently a research-based result rather than a commercially deployed product.

Generated Sep 27, 2026. Please verify key information.

About Universita degli Studi di Cagliari

The University of Cagliari is a comprehensive public research university serving Sardinia, combining broad academic depth with applied research and clinical training. Industry collaborates through shared laboratories on campus and in affiliated hospitals, enabling translation from discovery to real-world validation. The university works closely with the region’s science and technology park and innovation agencies, linking companies to prototyping spaces, contract research, internships, and executive education tailored to workforce needs. Research is supported by competitive European programs such as Horizon Europe alongside Italian national and regional funding mechanisms. A dedicated technology transfer office supports IP strategy, licensing, collaboration agreements, and spinout creation.

Halo home
Partner smarter. Move faster.
Get new partnering requests
delivered to your inbox.