Adaptive Bitrate Streaming Encoding Optimization via Quality Index Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing models for estimating watching time in adaptive bitrate video streaming are impractical due to shallow decision tree structures and high correlation issues with playback stop times and counts, leading to nonsensical estimates.
Innovation Solution
A streaming design assistance method that estimates the relationship between watching time and application quality indices using a mathematical model, applying these indices to optimize encoding conditions for increased watching time through abandonment rate and application quality index estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a decision tree model is used to estimate watching time, then the model structure is simple, but the estimation accuracy deteriorates due to shallow tree depth and discrete value concentration
Solution Approach 1:
The patent transforms the watching time estimation problem from a discrete classification task to a continuous prediction task by changing the model output parameter from discrete categories to continuous time values, thereby improving estimation accuracy while maintaining model simplicity
Solution Approach 2:
The patent replaces the traditional decision tree mechanical structure with a neural network-based mathematical model that can capture complex non-linear relationships between video quality parameters and watching time, achieving better accuracy without proportionally increasing complexity
2Reliability
If playback stop time and playback stop count are used as features, then the model can capture quality issues, but the estimation becomes nonsensical due to high correlation with session time
Solution Approach 1:
The patent extracts and removes the problematic high-correlation features (playback stop time and playback stop count) from the model input, keeping only the essential quality-related features that do not create spurious correlations with session time
Solution Approach 2:
Instead of using playback stop metrics that directly correlate with session time, the patent inverts the approach by using features that capture quality issues indirectly through their impact on user engagement, thereby avoiding the correlation problem while maintaining quality assessment capability
Data Source
AI summary
This streaming design assistance method assists in the selection of an encoding condition for increasing the watching time in adaptive bitrate video streaming by causing a computer to execute: estimating a relationship between a watching time or an abandonment rate and an application quality index according to a mathematical model on the basis of an actual value of a watching time of each of a plurality of first sessions related to video streaming and an actual value of an application quality index of each of the sessions; estimating the application quality index of each of the sessions for each of a plurality of encoding conditions related to video streaming on the basis of the plurality of encoding conditions and an actual value of time-series throughput of each of a plurality of second sessions; and applying the application quality index estimated under each of the encoding conditions to the relationship to estimate an average watching time of the plurality of second sessions.


