Video Frame Segmentation for Adaptive Bitrate Streaming

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Adaptive bitrate streaming in video transmission faces challenges such as high end-to-end latency, jitter, and network congestion due to the frequent transmission of large Intra-coded Picture Frames (I-frames), which are resource-intensive and cause spikes in network traffic.

Innovation Solution

The technique involves dividing video frames into lower quality Long Term Reference (LTR) frames and enhancement layers, allowing for incremental quality improvement during transmission, reducing the variance in frame size and network traffic spikes by combining LTR frame enhancement layers with inter-coded frames.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If Intra-coded Picture Frames (I-frames) are transmitted frequently to maintain video quality, then video quality is improved, but network congestion and latency increase

Engineering Contradiction:
Improvevideo qualityVSAvoidlatency
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent segments I-frames into multiple smaller units called mini-I-frames, which are then transmitted incrementally over time. This segmentation allows the video stream to maintain quality reference frames without transmitting large I-frame data packets all at once, thereby reducing latency and network congestion while preserving video quality.

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If large I-frames are transmitted to ensure reference frame quality, then video quality is improved, but network traffic spikes and jitter increase

Engineering Contradiction:
Improvereference frame qualityVSAvoidnetwork traffic stability
Core Design Contradiction:
Measurement precisionVSStability of the object's composition

Solution Approach 1:

Large I-frames are divided into multiple mini-I-frames that are transmitted as smaller, more manageable data packets distributed over time. This segmentation smooths out network traffic patterns, preventing the sharp spikes and jitter that occur when large I-frame packets are transmitted all at once, while still delivering the complete reference frame data.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The transmission of mini-I-frames is dynamically controlled based on network conditions and timing requirements. The system adjusts the transmission rate and timing of mini-I-frames to match network capacity, creating a dynamic transmission pattern that stabilizes network traffic while ensuring reference frame quality is maintained.

Inventive Principle:
Principle #15Dynamics

3Measurement precision

If I-frames are transmitted at regular intervals to maintain video quality, then video quality is improved, but end-to-end latency increases

Engineering Contradiction:
Improvevideo qualityVSAvoidend-to-end latency
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

By segmenting I-frames into mini-I-frames, the system reduces the time required to transmit each reference frame. Instead of waiting for a complete large I-frame to be transmitted, the decoder can begin processing smaller mini-I-frames incrementally, significantly reducing end-to-end latency while maintaining the quality benefits of regular I-frame transmission.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentUS10681382B1Enhanced encoding and decoding of video reference frames
Publication Date: 2020.06.09 AMAZON TECH INC
  • US10681382B1 patent drawing
  • US10681382B1 patent drawing
  • US10681382B1 patent drawing

AI summary

Techniques are generally described for compression encoding and decoding of video frames. In various examples, the techniques may include generating first reference frame data, first enhancement layer data corresponding to the first reference frame data, and second enhancement layer data corresponding to the first reference frame data. In various examples, the techniques may further include sending the first reference frame data, the first enhancement layer data, and the second enhancement layer data to at least one recipient computing device. The techniques may further include determining that the recipient computing device did not receive first unreceived data. In some examples, the first unreceived data may comprise at least one of the first reference frame data, the first enhancement layer data, and the second enhancement layer data. In other examples, the techniques may include resending the first unreceived data to the recipient computing device.