Predictive Frame Dropping for Streaming Data Quality

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Real-time streaming video applications face challenges in maintaining video stream quality due to bandwidth limitations, where frame dropping is necessary to prevent delays, but existing methods are inefficient and can adversely affect video quality by dropping critical frames.

Innovation Solution

Implementing predictive frame dropping on the sender side, where frames not yet transmitted are dropped from the queue except for the currently transmitting I-frame, and using a drop counter to determine which frames to drop from subsequent frames groups based on previous frame group drops, ensuring maximum bandwidth usage and even distribution of dropped frames.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If bandwidth limitation is enforced on the video stream, then bandwidth usage is controlled within limits, but video stream transmission rate is slowed down causing frames to arrive late and be dropped

Engineering Contradiction:
Improvebandwidth usageVSAvoidvideo frame delivery
Core Design Contradiction:
Quantity of substanceVSReliability

Solution Approach 1:

The system performs preliminary actions by proactively dropping frames from the transmission queue before they are transmitted, specifically when a new I-frame is queued. This predictive dropping prevents bandwidth overflow and ensures that only frames that can be transmitted within the bandwidth limit are sent, thereby controlling bandwidth usage while maintaining reliable delivery of transmitted frames.

Inventive Principle:
Principle #10Preliminary action

2Productivity

If frames are dropped to prevent delays, then bandwidth limitations are adhered to, but video quality is adversely affected by dropping critical frames

Engineering Contradiction:
Improvetransmission efficiencyVSAvoidvideo quality
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The system applies local quality by differentiating between frame types and applying selective dropping rules. I-frames are protected from dropping as they are critical for video quality, while P-frames and B-frames are dropped based on their dependency relationships. The drop counter mechanism locally tracks and distributes drops across frame groups to maintain overall video quality while adhering to bandwidth limitations.

Inventive Principle:
Principle #3Local quality

3Device complexity

If the video stream is encoded once for several recipients of differing bandwidth capability, then encoding complexity is reduced, but the system cannot adapt to individual bandwidth limitations of each recipient

Engineering Contradiction:
Improveencoding complexityVSAvoidbandwidth adaptation
Core Design Contradiction:
Device complexityVSAdaptability or versatility

Solution Approach 1:

The system introduces dynamics by implementing adaptive frame dropping at the transmission queue level for each recipient. While the video stream is encoded once, the QoS module dynamically adjusts which frames are transmitted to each recipient based on their specific bandwidth limitations. The drop counter and predictive dropping mechanism enable real-time adaptation to individual bandwidth conditions without requiring separate encodings.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP2030448B1Predictive frame dropping to enhance quality of service in streaming data
Publication Date: 2016.10.19 SONY INTERACTIVE ENTERTAINMENT LLC
  • EP2030448B1 patent drawingFigure 1A~2A
  • EP2030448B1 patent drawingFigure 2B~3
  • EP2030448B1 patent drawingFigure 4A~4D

AI summary

Methods and apparatus for optimizing quality of streaming data transmitted between a first node and a second node over a network, the streaming data including one or more frame groups is disclosed. A processor-readable medium having embodied therein processor readably instructions for implementing a method for optimizing quality of streaming data transmitted between a first node and a second node over a network is also disclosed.