Efficient qos support for voice over ip applications using selective packet marking
Download
1 / 13

66-sanneck-slides-new - PowerPoint PPT Presentation


  • 200 Views
  • Updated On :

Efficient QoS Support for Voice-over-IP Applications Using Selective Packet Marking. Henning Sanneck, Long Le and Georg Carle GMD FOKUS, Berlin {sanneck,le,[email protected] First IP Telephony Workshop, Berlin April 13th, 2000. Overview. Voice over IP (VoIP) Improved quality for VoIP

Related searches for 66-sanneck-slides-new

loader
I am the owner, or an agent authorized to act on behalf of the owner, of the copyrighted work described.
capcha
Download Presentation

PowerPoint Slideshow about '66-sanneck-slides-new' - Sharon_Dale


An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.


- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript
Efficient qos support for voice over ip applications using selective packet marking l.jpg

Efficient QoS Support for Voice-over-IP Applications Using Selective Packet Marking

Henning Sanneck, Long Le and Georg Carle

GMD FOKUS, Berlin

{sanneck,le,[email protected]

First IP Telephony Workshop, Berlin

April 13th, 2000


Overview l.jpg
Overview Selective Packet Marking

  • Voice over IP (VoIP)

    • Improved quality for VoIP

  • Approach

    • Performance of the G.729 loss concealment

    • Speech Property-Based Selective Packet Marking

  • Evaluation

    • Reference packet marking/prioritization schemes

    • Simple network models

    • Objective speech quality measurement

  • Conclusions


Voice over ip l.jpg
Voice over IP Selective Packet Marking

  • Main drivers:

    • current economical incentives (Internet flat rate pricing) Internet Telephony

    • service integration, unified packet-switching infrastructure

  • One of the main problems:

    • satisfaction of real-time QoS demands in a packet-switched network (fundamental tradeoff: statistical multiplexing vs. reliability packet loss)


Qos for voice over ip flows l.jpg
QoS for Voice over IP flows Selective Packet Marking

  • low bitrate high per-flow overhead for reservations (scalability, need for multiplexing)

  • high compression (backward adaptive coding: ITU-T G.729, G.723.1)

    • no further sender adaptation / network adaptation (transcoding) possible

    • amplifies high perceptual impact of burst losses (error propagation)

  • tolerance to isolated losses (speech stationarity  extrapolation of coder state  loss concealment)  extend the loss resiliency of high-compressing codecs using prioritization (priority enforcement on flow aggregates: Differentiated Services)


Structure of an internet audio tool l.jpg
Structure of an Internet Audio Tool Selective Packet Marking


Additional components sender l.jpg
Additional components: Sender Selective Packet Marking

  • Side information available at the encoder is used

  • Decoder concealment process is taken into account

     Selective packet marking


Receiver l.jpg
Receiver Selective Packet Marking

  • No generic (PCM-level) concealment


Performance of g 729 loss concealment l.jpg
Performance of G.729 loss concealment Selective Packet Marking

  • Decoder fails to conceal losses at unvoiced/ voiced transition due to lack of state (synthesis filter parameters, excitation)


Speech property based selective packet marking l.jpg
Speech Property-Based Selective Packet Marking Selective Packet Marking

protect = 0

foreach (k frames)

classify = analysis(k frames)

if (protect > 0)

if (classify == unvoiced)

protect = 0

send(k frames, "0")

else

send(k frames, "+1")

protect = protect-k

endif

else

if (classify == uv_transition)

send(k frames, "+1")

protect = N-k

else

send(k frames, "0")

endif

endif

endfor

  • Packet marking („0“, „+1“) is application-controlled (adaptive to expected loss concealment performance):40-50% of packets marked with higher priority („+1“)


Marking scheme network model l.jpg
Marking Scheme / Network Model Selective Packet Marking


Results auditory distance l.jpg
Results: Auditory Distance Selective Packet Marking

  • Application of recent advances in objective speech quality measurement: ITU P.861A


Results perceptual distortion l.jpg
Results: Perceptual Distortion Selective Packet Marking

  • Enhanced Modified Bark Spectral Distortion(EMBSD; Temple University)


Conclusions l.jpg
Conclusions Selective Packet Marking

  • SPB-(DIFF)MARK exploits differences in „concealability“ to adjust the amount of needed network prioritization

  • simple network models & objective speech quality measures showed the reduction of number of marked packets while maintaining a good output quality

  • selective marking requires appropriate socket interface (or application-level gateway)

  • future work: employ speech quality measurement for the marking decision

  • http://www.fokus.gmd.de/glone


ad