WUSTL High-Definition Video Trace Library

Presented here are our results of encoding more than 50 high definition (HD) video traces collected from the popular web video streaming website: YouTube. Additionally, we provide a simple statistical analysis of the video traces and support it with different graphs. As a part of this contribution, we evaluate the capability of the SAM model [JSAC10, MMSys10] of representing these traces accurately. SAM parameters for each movie traces is then calculated and presented. Our approach can be used not only to model and generate different video traces but also to predict resource demands and to optimize scheduling and network coding in real time. The figure blow shows the steps we took in analyzing and modeling the selected videos with their corresponding outputs.

Modeling, analyzing, and generating video traces processes

The process starts with a YouTube video, which is then converted to a YUV raw video. The raw video is consequently encoded with AVC to produce an encoded movie file, an encoding statistics file, and a full verbose description of the encoding process. The verbose output is then parsed to get the video frames information, which is then modeled using SAM. The video frames' vector is used to produce the autocorrelation function (ACF) and the partial autocorrelation function (PACF) graphs. The video's SAM parameters then can be used either to predict future traffic, or to produce a movie trace. To produce a trace, SAM frame generator uses these parameters to generate a movie trace that is statistically close to the original movie trace.

Click here to view/download the SAM Model and High-Definition Video Traces Data.

References

[MMSys10] A. Al-Tamimi, R. Jain, C. So-In, "Modeling and Generation of AVC and SVC-TS Mobile Video Traces for Broadband Access Networks," In Proc. of the First Annual ACM SIGMM Conference on Multimedia Systems (Phoenix, AZ, USA, Feb 22-23, 2010). MMSys'10.

[JSAC10] A. Al-Tamimi, C. So-In, R. Jain, "Modeling and Resource Allocation for Mobile Video over WiMAX Broadband Wireless Networks,"IEEE Journal on Selected Areas in Communications (JSAC), Special Issue on Wireless Video Transmission, vol. 28, no. 3, pp. 354-365, April 2010.

Tools Used in Converting and the Analysis of the Video Traces

Parser Tool to Parse JM reference software verbose output, JMReferenceTraceParser.zip
JM Reference Software Website Link, http://iphome.hhi.de/suehring/tml/
R Project Software Website Link, http://www.r-project.org/
FFMPEG The leading Audio/Video Codec Library, http://ffmpeg.org/

Scripts Used in Converting the Video Files

Modeling Traces : R Command [For Period of 24]
VideoTrace.SAM = arima(VideoTrace, order=c(1,0,1),
seasonal=list(order=c(1,1,1),period=24));
Convert YUV to AVC Using JM Reference Software
lencod.exe -p InputFile="VideoName.YUV" -p StartFrame=0 -p
FramesToBeEncoded=N -p FrameRate=24 -p SourceWidth=1280 -p
SourceHeight=720 -p OutputWidth=1280 -p OutputHeight=720 -p
OutputFile="VideoName.H264" -p StatsFile="VideoName.Stats.txt" -p
ReportFrameStats=1 -p DisplayEncParams=1 -p Verbose=3 -p ProfileIDC=100
-p NumberBFrames=2 -p IDRPeriod=24 -p NumberReferenceFrames=3 >>
EncodingResults.txt
Convert Mp4 to YUV Using FFMPEG
ffmpeg -i "VideoName.mp4" -vcodec rawvideo -s 1280x720 -f rawvideo -pass
1 -r 24 -an -aspect 16:9 "VideoName.YUV"

Back to Raj Jain's home page