Browse Prior Art Database

CONTENT-ADAPTIVE DIFFERENTIAL MULTISTREAM RATE ADAPTATION

IP.com Disclosure Number: IPCOM000238667D
Publication Date: 2014-Sep-10

Publishing Venue

The IP.com Prior Art Database

Related People

Paul Bright-Thomas: AUTHOR

Abstract

A solution is presented herein that utilizes simple statistics describing motion and content scale within coded sequences to decide which members of a multistream aggregate forwarded to an endpoint X can be decimated, e.g., by procurement at an alternate resolution from the source, or discard of spatial or temporal layers, by the conference server with the least impact on the received experience at endpoint X.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 20% of the total text.

Page 01 of 12

CONTENT-ADAPTIVE DIFFERENTIAL MULTISTREAM RATE ADAPTATION

AUTHOR:

Paul Bright-Thomas

CISCO SYSTEMS, INC.

ABSTRACT

    A solution is presented herein that utilizes simple statistics describing motion and content scale within coded sequences to decide which members of a multistream aggregate forwarded to an endpoint X can be decimated, e.g., by procurement at an alternate resolution from the source, or discard of spatial or temporal layers, by the conference server with the least impact on the received experience at endpoint X.

DETAILED DESCRIPTION

    A rich multiparty conferencing experience may be implemented with low cost by multistream forwarding, where a conference server allows each endpoint to receive N video streams from N other endpoints. When there is congestion on the distribution path from a mediating conference server to a single endpoint (X), oversubscription of the available distribution bandwidth should be met by rate adaptation (bandwidth reduction) of the assemblage of streams delivered to that endpoint. The solution presented herein discloses how to perform rate adaptation having the least impact on the user experience of that receiving endpoint, and of other endpoints receiving the same or other streams, by differential action on the streams forwarded to X, prioritizing those streams and those actions that have the least impact on experience for bandwidth reduction.

    The following reference, http://simula.no/research/nd/publications/ Simula.ND.265 states that "[o]ur study reveals that QoE degrades nonmonotonically with bitrate and that scaling order preferences are content-dependent," confirming that sensitivity to temporal decimation is dependent on the motion in the sequence. Another scheme described in http://www.princeton.edu/~chiangm/caf.pdf, states that "CAF [Content-Aware distortion-Fair video-streaming] leverages content-awareness to prioritize packet dropping during congestion," but is not based on real-time media nor the

Copyright 2014 Cisco Systems, Inc.
1


Page 02 of 12

provision of metrics from a remote source to be acted upon by a central server. The authors of http://ieeexplore.ieee.org/stamp/stamp.jsp?tp =&arnumber=4543842, remark that "we formulate and study the problem of joint scheduling and playout control in the framework of Markov decision processes … we propose a novel content-aware adaptive playout control, that takes into account the content of a video sequence, and in particular the motion characteristics of different scenes." While this scheme proposes motion- aware frame-rate variation of playout time, to avoid loss using pre-coded video and maintain sensitivity to scene content, it does not describe frame or layer discard, as does the solution presented herein. Finally, reference http://www0.cs.ucl.ac.uk/ staff/d.miras/thesis/thesis.pdf, addresses multistream statistical multiplexing for pre- recorded sequences and uses high layer counts and pre-computed profiles using a neural network, wi...