Chunk-Based Filtering for Video Streaming Quality Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing adaptive bitrate (ABR) video streaming technologies do not effectively consider data efficiency and perceptual quality variability across video chunks, leading to suboptimal video quality and inefficient data usage, particularly in dynamic network conditions like cellular networks.
Innovation Solution
The method involves selecting video chunks based on a desired perceptual quality, using a manifest file to filter out chunks with higher perceptual quality than desired, and choosing the chunk with the lowest encoding bitrate that meets the quality threshold, thereby optimizing data usage and maintaining consistent video quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If the highest encoding bitrate video chunk is selected, then the video quality is maximized, but the data consumption increases significantly
Solution Approach 1:
The system changes the selection parameter from purely bitrate-based to perceptual quality-based. By using perceptual quality metrics (such as VMAF, PSNR, or SSIM) to evaluate and select video chunks, the system achieves better quality efficiency, downloading only the necessary bits required for acceptable perceived quality rather than maximizing raw bitrate.
Solution Approach 2:
The patent applies local quality assessment by evaluating individual video chunks based on their specific perceptual quality characteristics rather than treating all chunks uniformly. This allows selective downloading of chunks with higher perceptual quality at lower bitrates, optimizing the balance between quality and data consumption for each local chunk.
2Reliability
If all video chunks are downloaded to ensure quality, then the video playback reliability is improved, but the data usage becomes inefficient
Solution Approach 1:
The system replaces traditional mechanical bitrate-adaptation mechanisms with a perceptual quality-aware selection mechanism. Instead of relying solely on network conditions and bitrate thresholds, the system uses perceptual quality metrics to predict which chunks will provide acceptable quality, enabling more intelligent data selection that maintains playback reliability while improving efficiency.
Solution Approach 2:
The system incorporates feedback loops that continuously monitor actual perceived quality and adjust future chunk selections accordingly. By using quality metrics to guide selection and learning from actual playback performance, the system optimizes the balance between reliability and efficiency over time.
3Loss of substance
If perceptual quality metrics are collected and used for selection, then the data efficiency is improved, but the system complexity increases
Solution Approach 1:
The system performs preliminary quality assessment and chunk selection before actual video playback begins. By pre-evaluating chunks using perceptual quality metrics and pre-selecting optimal chunks for download, the system avoids the complexity of real-time quality adjustment during playback, shifting the computational burden to a more efficient pre-processing phase.
Solution Approach 2:
The patent uses quality metrics and models as copies or representations of actual video quality characteristics. Instead of analyzing every pixel and bit in real-time, the system uses pre-computed quality models and metrics that capture essential quality information, reducing computational complexity while maintaining selection accuracy.
4Ease of manufacture
If traditional ABR protocols are used, then the implementation simplicity is maintained, but the quality consistency across different chunks deteriorates
Solution Approach 1:
The system segments the video stream into individual chunks and evaluates each chunk's perceptual quality independently. This segmentation allows for granular quality control and selection, enabling consistent quality across different chunks by assessing and selecting based on actual chunk-level quality metrics rather than relying on aggregate bitrate profiles.
Data Source
AI summary
A processing system including at least one processor may obtain a selection of a desired perceptual quality for a video and selects a first video chunk from among a first plurality of video chunks, wherein the first plurality of video chunks is associated with a first time block of the video, wherein the selecting is based upon a manifest file of the video, wherein each video chunk of the first plurality of video chunks is associated with a respective perceptual quality and a respective encoding bitrate, wherein the first video chunk is part of a subset of the first plurality of video chunks for which the respective perceptual quality matches the desired perceptual quality within a predefined threshold tolerance, and wherein the respective encoding bitrate of the first video chunk is lower than a highest encoding bitrate associated with the subset of the first plurality of video chunks. The processing system may obtain the first video chunk from a server and store the first video chunk in a buffer.