QoE Adjustment Factor for CDN-OTT Measurement Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing network monitoring systems struggle to accurately assess Quality of Experience (QoE) for Over-The-Top (OTT) video streaming services in real-time, particularly in identifying the root cause of issues such as freezing, buffering, or lagging, which often originate outside the service providers' networks.
Innovation Solution
The method involves adjusting QoE measurements obtained from a testing Content Delivery Network (CDN) entity to reflect accurately the QoE that would have been obtained from a reference video source, such as YouTube, by using a QoE adjustment factor determined through a learning process and predictive modeling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If QoE measurements are obtained directly from reference video sources (e.g., YouTube) through repeated API requests, then accurate QoE assessment is achieved, but the content provider's terms of service are violated and the system is blocked
Solution Approach 1:
The patent introduces a testing CDN entity as an intermediary between the monitoring system and the reference video source. Instead of directly querying the content provider's API repeatedly, the system uses a predictive model trained on historical QoE data from the testing CDN to estimate current QoE conditions. This intermediary approach maintains measurement accuracy while avoiding direct interference with the content provider's systems.
Solution Approach 2:
The system creates a virtual copy of the QoE measurement process by using a predictive model that replicates the relationship between CDN performance and QoE. Rather than actually querying the reference video source, the model generates estimated QoE measurements based on patterns learned from historical data, effectively copying the measurement function without the harmful side effects.
2Measurement precision
If legacy video monitoring solutions are used to assess QoE, then comprehensive video quality analysis is achieved, but the system becomes expensive and compute-intensive
Solution Approach 1:
The patent transforms the QoE assessment approach by changing the parameters being monitored. Instead of performing comprehensive video quality analysis that requires heavy computational resources, the system monitors key performance parameters from the CDN (such as buffering events, playback continuity, and network conditions) and uses these simplified parameters to infer overall QoE through the predictive model.
Solution Approach 2:
The system replaces the mechanical video analysis process with a data-driven predictive modeling approach. Rather than actually processing and analyzing video content to assess quality, the system uses machine learning models that predict QoE based on CDN performance metrics, substituting heavy mechanical computation with lighter statistical inference.
3Reliability
If real-time QoE monitoring is implemented for OTT services, then network performance issues are detected quickly, but the complexity of correlating QoE with QoS measurements increases significantly
Solution Approach 1:
The patent merges QoE measurement functionality with existing CDN performance monitoring infrastructure. By training the predictive model on historical data that correlates CDN metrics with QoE outcomes, the system combines multiple monitoring functions into a unified approach where the same data collection mechanisms serve both QoS tracking and QoE assessment purposes.
Solution Approach 2:
The predictive model serves multiple functions simultaneously: it estimates current QoE conditions, identifies performance degradation trends, and correlates CDN behavior with user experience outcomes. This multi-functional approach reduces system complexity by using a single modeling framework for what would otherwise require multiple separate monitoring and analysis systems.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
A method (100) includes obtaining (102) a first Quality of Experience, QoE, measurement (30) from a first content source (16); determining (104) a QoE adjustment factor (32) based on a predictive model; and adjusting (106) the first QoE measurement (30) using the QoE adjustment factor (32) to obtain a second QoE measurement (34), the second QoE measurement (34) reflects a QoE estimate from a second content source (18). The first content source (16) can be a Content Delivery Network, CDN, entity, and the second content source (18) is an Over-the-Top, OTT, content provider.