Multiple-Description Decoding with Layered Quantization Under Packet Loss

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing multiple-description coding techniques suffer from significant quantization precision loss and increased errors during packet loss, particularly at high packet loss rates, which degrades user experience in internet multimedia applications like real-time video-audio communication and video on demand.

Innovation Solution

A method and system for multiple-description coding and decoding that processes signals using a combination of multiple-description parity separation, dual transformation, and scalar quantization methods to generate and decode description signals, distributing voice quality deterioration across different aspects to improve user experience during packet loss.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If four-description coding is implemented to resist against three-packet loss, then reliability is improved, but device complexity increases due to 15 packet loss scenarios requiring separate decoding

Engineering Contradiction:
Improvepacket loss resistanceVSAvoiddecoding complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent divides the signal into multiple descriptions (at least two) with different quantization precisions. Each description is independently encoded and can be separately decoded, allowing the system to handle packet loss scenarios without requiring complex joint decoding for all possible loss combinations. This segmentation reduces decoding complexity while maintaining reliability.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies different quantization precisions to different descriptions locally. Some descriptions use higher precision while others use lower precision, allowing the system to optimize for specific packet loss scenarios. This local quality differentiation enables efficient resource allocation without requiring uniform high precision across all descriptions, reducing overall complexity.

Inventive Principle:
Principle #3Local quality

2Manufacturing precision

If higher quantization precision is used to maintain signal quality, then manufacturing precision is improved, but loss of information increases during packet loss due to greater quantization errors when descriptions are lost

Engineering Contradiction:
Improvequantization precisionVSAvoidquantization error
Core Design Contradiction:
Manufacturing precisionVSLoss of information

Solution Approach 1:

The patent applies different quantization precisions to different descriptions based on their importance and the packet loss scenarios they are designed to handle. Critical descriptions use higher precision while less critical ones use lower precision, optimizing the balance between quality and information loss without uniformly increasing precision across all descriptions.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent changes the quantization precision parameter for different descriptions rather than using a single precision level. This allows the system to adapt the precision parameter to match the specific requirements of each description and the expected packet loss patterns, reducing overall information loss while maintaining necessary quality.

Inventive Principle:
Principle #35Parameter changes

3Reliability

If multiple-description coding is used to prevent packet loss impact, then reliability is improved, but loss of energy increases due to encoding overhead and redundancy

Engineering Contradiction:
Improvepacket loss resistanceVSAvoidcoding overhead
Core Design Contradiction:
ReliabilityVSLoss of energy

Solution Approach 1:

The patent optimizes the redundancy parameter by using different quantization precisions for different descriptions. This parameter change allows the system to achieve reliable packet loss resistance with reduced overall redundancy compared to uniform high-precision coding, thereby reducing the energy overhead associated with encoding and transmitting redundant information.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP2146436B1The method, apparatus and system for multiple-description coding and decoding
Publication Date: 2013.12.04 HUAWEI TECH CO LTD
  • EP2146436B1 patent drawingFigure 1
  • EP2146436B1 patent drawingFigure 2
  • EP2146436B1 patent drawingFigure 3

AI summary

A method, apparatus and system for multiple-description coding (MDC) and multiple-description decoding (MDD) are provided in the present invention. The MDC method comprises the following steps: receiving signals, processing the signal according to the first multiple-description method to generate at least two first description signals, processing at least one of the first description signals through the second multiple-description method to generate at least two second description signals, and coding the second description signal to generate the multiple-description bit stream. The MDD method comprises the following steps: receiving the multiple-description bit stream, decoding the multiple-description bit stream to generate the second description signal, processing the second description signal through the second multiple-description inverse transformation method to generate the first description signal, processing the first description signal through the first multiple-description inverse transformation method to generate the reconstruction signal. This invention simplifies the MDC and MDD, and distributes the voice quality deteriorations to different parts, thus improving user experience.