Forward and Backward Extrapolation for Packet Loss Concealment

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current packet loss concealment techniques in real-time communication services, such as VoIP, often produce unnatural sounds due to limitations in forward prediction and interpolation methods, especially when network packet loss occurs, leading to suboptimal user experience.

Innovation Solution

The implementation of forward and backward extrapolation methods to reconstruct lost data by using network packets before and after the loss period, allowing for a more natural and effective filling of gaps in audio or video streams, reducing the need for repetitive data and minimizing distortion.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If forward prediction and interpolation methods are used to compensate for network packet loss, then some lost data can be recovered, but the reconstructed data produces unnatural sounds and poor user experience

Engineering Contradiction:

Improvepacket loss concealment effectivenessVSAvoidunnatural sounds in audio output

Core Design Contradiction:

ReliabilityVSObject-affected harmful factors

Solution Approach 1:

The patent applies backward extrapolation from future packets to reconstruct lost data, inverting the conventional forward-only approach. By extrapolating backward from packets that arrive after the loss period and combining with forward extrapolation from packets before the loss, the method generates more natural-sounding audio compared to traditional forward prediction alone

Inventive Principle:

Principle #13The other way round (Inversion)

Solution Approach 2:

The patent segments the reconstruction of lost data into two distinct portions: a first portion reconstructed using forward extrapolation from pre-loss packets, and a second portion reconstructed using backward extrapolation from post-loss packets. This segmentation allows each direction to handle different parts of the lost data, improving overall naturalness

Inventive Principle:

Principle #1Segmentation

2Loss of information

If traditional packet loss concealment methods are used, then lost data can be partially recovered, but the quality of audio playback deteriorates

Engineering Contradiction:

Improverecovered data completenessVSAvoidaudio quality degradation

Core Design Contradiction:

Loss of informationVSObject-affected harmful factors

Solution Approach 1:

The patent changes the fundamental parameter of extrapolation direction by implementing backward extrapolation from future packets in addition to forward extrapolation. This parameter change in the reconstruction approach produces more natural-sounding audio while maintaining data completeness

Inventive Principle:

Principle #35Parameter changes

3Reliability

If more data is transmitted to compensate for packet loss, then reliability improves, but bandwidth requirements increase

Engineering Contradiction:

Improvecommunication reliabilityVSAvoidbandwidth consumption

Core Design Contradiction:

ReliabilityVSQuantity of substance

Solution Approach 1:

The patent enables the receiving device to self-service by reconstructing lost data locally using extrapolation techniques applied to received packets. This eliminates the need to transmit additional redundant data or retransmissions, reducing bandwidth consumption while maintaining reliability

Inventive Principle:

Principle #25Self-service

Data Source

PatentUS10803876B2Combined forward and backward extrapolation of lost network data

Publication Date: 2020.10.13 MICROSOFT TECHNOLOGY LICENSING LLC

AI summary

Techniques are described for performing forward and backward extrapolation of data to compensate for data that has been lost due to network packet loss. The forward and backward extrapolation can be used to perform packet loss concealment. For example, when network packet loss is detected, network packets before and after the lost data can be identified. Forward and backward extrapolation can then be applied to cover the period of lost data. For example, the network packets before the period of lost data can be used to perform forward extrapolation to cover a first portion of the period of lost data. The network packets after the period of lost data can be used to perform backward extrapolation to cover a remaining portion of the period of lost data. The period of lost data can be reconstructed based at least in part on the extrapolation.