QoE Prediction Using QoS Proxy Modeling in 5G Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining quality-of-experience in 5G networks are limited by the inability to access QoE feedback or reduced feedback datasets, which hampers long-term optimization of application behavior and session setup in certain deployment conditions.
Innovation Solution
A method that involves obtaining a grouped frequency distribution of quality-of-service measurements over time, generating a model describing QoS-QoE relationships, and extrapolating statistical evolution of quality-of-experience to predict future satisfaction, using techniques like hidden Markov modeling and expectation maximization algorithms to optimize resource allocation and maximize application satisfaction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If direct QoE feedback from application clients is used for optimization, then long-term optimization of application behavior and session setup is achieved, but the system becomes inapplicable in deployment conditions where the network cannot access QoE feedback or has access to reduced feedback datasets
Solution Approach 1:
The patent introduces QoS measurements as an intermediary variable that the network can observe directly. Instead of relying on QoE feedback from application clients, the system uses QoS metrics (throughput, latency, packet loss) as proxy indicators to infer QoE states. This mediator enables the optimization framework to function in deployment scenarios where direct QoE feedback is unavailable or limited.
Solution Approach 2:
The patent replaces the mechanical feedback collection system (direct QoE feedback from clients) with a computational modeling system. Hidden Markov Models and expectation maximization algorithms substitute for direct client feedback, allowing the network to predict QoE states and optimize resource allocation based on QoS measurements and statistical inference rather than direct user feedback.
2Measurement precision
If QoS measurements are collected and modeled using hidden Markov models to predict QoE, then accurate prediction of quality-of-experience is achieved, but the system complexity increases due to model generation and parameter optimization requirements
Solution Approach 1:
The patent performs preliminary actions by pre-collecting QoS measurements and pre-training Hidden Markov Models during periods when optimization is not critical. The expectation maximization algorithm pre-optimizes model parameters offline, so that during actual operation, the system can quickly apply pre-computed models for QoE prediction without real-time computational burden, thus reducing operational complexity while maintaining prediction accuracy.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
The invention relates to a method for determining and/or maximizing a quality-of-experience related to an application deployed over a mobile network, the method being implemented by at least one network function of the mobile network, the method comprising: - obtaining information indicating a distribution of possible levels of quality-of-service provided by the mobile network for the application over a period of time, - obtaining or generating a model describing, for the application, relationships between the possible levels of quality-of-service and possible states of quality-of-experience, and - extrapolating a statistical evolution of the quality-of-experience related to the application based on the obtained information and on the model. The invention further relates to a corresponding device and to a corresponding computer program.