Feature Map Video Coding for Adaptive Network Bitrate

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing video coding techniques struggle to efficiently adapt bitrate in network streaming, particularly in 5G radio access networks, leading to suboptimal performance and increased bandwidth consumption.

Innovation Solution

A method is introduced where processing tasks for video coding are split between a source and destination device, with the source device determining task distribution and encoding the feature map for the destination device to perform, allowing dynamic bitrate adaptation based on operating conditions and network capabilities.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If video data is transmitted over network with high bitrate to maintain quality, then processing performance is improved, but network bandwidth consumption increases

Engineering Contradiction:

Improveprocessing performanceVSAvoidnetwork bandwidth consumption

Core Design Contradiction:

ProductivityVSQuantity of substance

Solution Approach 1:

The video processing task is segmented into feature extraction at the source device and remaining processing tasks at the destination device. Only the feature map (intermediate processing result) is transmitted over the network, not the complete high-resolution video data. This segmentation allows the network to carry only essential intermediate data, reducing bandwidth consumption while maintaining processing performance at the destination.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The feature map serves as an intermediary representation between the source video data and the final processed output. By transmitting this intermediate feature representation instead of raw video data, the system reduces network bandwidth requirements while preserving the essential information needed for subsequent processing tasks at the destination device.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Productivity

If all processing tasks are performed at the source device, then processing performance is improved, but network efficiency deteriorates due to excessive data transmission

Engineering Contradiction:

Improveprocessing performanceVSAvoidnetwork efficiency

Core Design Contradiction:

ProductivityVSLoss of energy

Solution Approach 1:

Processing tasks are segmented and distributed between source and destination devices. The source device performs feature extraction to generate a feature map, which is then transmitted to the destination device for remaining processing tasks. This task segmentation reduces the amount of data that needs to be transmitted over the network, improving network efficiency while maintaining overall processing performance.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system dynamically determines the optimal split point for processing tasks based on operating conditions, network capabilities, and task requirements. This dynamic task distribution allows the system to adapt to varying network conditions and device capabilities, optimizing both processing performance and network efficiency in real-time.

Inventive Principle:

Principle #15Dynamics

3Quantity of substance

If video data is compressed to reduce bandwidth consumption, then network efficiency is improved, but processing quality deteriorates

Engineering Contradiction:

Improvenetwork bandwidth consumptionVSAvoidprocessing quality

Core Design Contradiction:

Quantity of substanceVSManufacturing precision

Solution Approach 1:

The system changes the representation parameters of the video data by transforming it into a feature map with reduced dimensionality and optimized data structure. This parameter transformation maintains the essential semantic information needed for processing while significantly reducing the data size for network transmission, thereby achieving both bandwidth efficiency and processing quality.

Inventive Principle:

Principle #35Parameter changes

4Manufacturing precision

If feature map size is increased to improve processing accuracy, then processing quality is improved, but network bandwidth consumption increases

Engineering Contradiction:

Improveprocessing accuracyVSAvoidnetwork bandwidth consumption

Core Design Contradiction:

Manufacturing precisionVSQuantity of substance

Solution Approach 1:

The feature map is optimized by changing its representation parameters including dimensionality reduction, quantization, and selective feature retention. These parameter optimizations maintain the essential information needed for accurate processing at the destination while minimizing the data size for network transmission, achieving a balance between processing accuracy and bandwidth consumption.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS12526439B2Rate adaptation for video coding for machines

Publication Date: 2026.01.13 QUALCOMM INC

AI summary

An example a first network entity for processing feature set data formed from media data includes a processing system comprising one or more processors implemented in circuitry, the processing system being configured to: determine a first set of processing tasks of a series of processing tasks to be performed on a set of media data, the first set of processing tasks corresponding to tasks to be performed by the first network entity, wherein a second set of processing tasks is to be performed by a second network entity; perform the first set of processing tasks on the set of media data to form a feature map; encode the feature map to form an encoded feature map; and send the encoded feature map to the second network entity to enable the second network entity to perform the second set of processing tasks using the feature map.