Presentation is loading. Please wait.

Presentation is loading. Please wait.

Building Scalable and High Efficient Java Multimedia Collaboration Wenjun Wu, Tao Huang, Geoffrey Fox Community Grids Computing Laboratory, Indiana University,

Similar presentations


Presentation on theme: "Building Scalable and High Efficient Java Multimedia Collaboration Wenjun Wu, Tao Huang, Geoffrey Fox Community Grids Computing Laboratory, Indiana University,"— Presentation transcript:

1 Building Scalable and High Efficient Java Multimedia Collaboration Wenjun Wu, Tao Huang, Geoffrey Fox Community Grids Computing Laboratory, Indiana University, USA

2 Global-MMCS & XGSP XGSP A common AV signaling protocol A core conference control mechanism Collaboration Web-Services Global Multimedia Collaboration System (Global- MMCS) to support scalable web-service based interoperable collaborations integrates various services including videoconferencing, instant messaging and streaming, and supports multiple videoconferencing technologies such as H.323, SIP and Access Grid clients

3 Global-MMCS & XGSP SIPH323 Access GridNative XGSP Admire Gateways convert to uniform XGSP Messaging High Performance (RTP) and XML/SOAP and.. Media Servers Filters Session Server XGSP-based Control NaradaBrokering All Messaging Use Multiple Media servers to scale to many codecs and many versions of audio/video mixing NB Scales as distributed Web Services NaradaBrokering

4 Java Media Framework Java Media Framework (JMF) A unified framework for managing the acquisition, processing, and delivery of time-based media data A common cross-platform Java API for accessing underlying media frameworks. “Write Once, Run Anywhere” to multimedia developers

5 JMF for multimedia collaboration JMF is the key building block for the implementation of Global-MMCS For both client-side and server-side Building multimedia services such as video, audio mixing, snapshot generation Implementing Global-MMCS AVportlet, a native XGSP audiovisual client Performance Issues in JMF the performance optimization of Java Media Framework has to be made especially for the conferencing application

6 Multimedia Programming Frameworks Video CaptureVideo Rendering Audio WindowsVFW or WDMDirect DrawDirect Sound LinuxVideo-for-Linux XLibALSA Mac OSSequence GrabberQuick DrawCore Audio Portability vs Performance : Platform-dependent multimedia devices and codec management a portable multimedia framework must define a good platform-independent part to cover the heterogeneous native multimedia platforms.

7 Multimedia Programming Frameworks the similar multimedia filter pipeline architecture Filter & Pipe design pattern cascaded filter chain Support high efficient media processing High-volume multimedia data is typically generated by a source filter objects and piped through one or more filter objects. Eventually, the media reaches a sink filter object and is consumed.

8 Multimedia Programming Frameworks Tcl/Tk based CMT [7], VIC [8], VuSystem [9] low-overhead control functionality implemented in a scripting language like Tcl performance-critical data handling implemented in a compiled language like C/C++ Windows DirectShow C++ COM API only enables Windows applications for multimedia programming JMF

9 JMF Filter Architecture

10 Access Grid Access Grid : a large scale audio/videoconference based on a multicast network provides the group-to-group collaborations among 150 nodes connected to Internet 2 world wide. Use improved MBONE audiovisual tools VIC and RAT Depends upon high-speed network ( each node needs 20Mbps )

11 Issues for JMF in collaboration Very good cross-platform framework No optimization for conferencing application Compared to fully performance tuned Tcl/Tk package ( Access Grid ) No implementation for Mac OS

12 Enhanced Java Media Framework Improving JMF performance by exposing the interfaces of the JMF filters and rewriting a faster one Adding the state-of-art codecs in JMF and screen capture Extending the JMF to the Mac-OS platform to extend its portability

13 Video Rendering JMF filter pipeline is composed of filter Modules and linking Connectors streaming protocols between a OutputConnector and InputConnector: Safe protocol Push protocol JMF render implementation: safe protocol, blit the whole video frame Our optimization approach: “Direct Write Through” approach

14 Direct-Write-Through optimization in JMF rendering

15 Video Capturing a video data source to abstract the real capture services which usually have different APIs in modern operating systems. added Mac video capturing into JMF package Added the desktop capture as a video data source to support remote desktop sharing

16 Add New Codec JMF 2.0 package only supports H.261 decoder, H263 and JPEG. We added H.261 encoder and MPEG-4 video codec (DivX) based on JMF codec interface. In addition, new RTP format and payload for MPEG-4 video are also added through JMF RTPManager.

17 Global-MMCS AVPortlet

18 GlobalMMCS AVPortlet, to fully make use of the services provided by Global-MMCS. The tool which can run on multiple desktop platforms such as windows, Linux and Mac OSX, integrates audio and video collaboration together Extending JMF RTP Transport over Publish/Subscribe Overlay XGSP Audiovisual Session Management

19 Performance Evaluation Compare the video rendering performance between JMF and VIC Testing video streams are captured from the same desktop of the sending machine, and multicasted to the receiver machine which runs Access Grid VIC, two JMF clients based on Sun’s JMF implementation and our improved version. two test scenarios with different source streams the CIF-size still image of the desktop area is sent to the clients. Each stream is encoded in H.261, and has average bandwidth 20kbps. the CIF-size video sequence from a 30-second movie with a lot of motions is streamed to the clients. Each stream is encoded in H.261, and has average bandwidth 500kbps.

20 Performance Evaluation Video Rendering performance (left: still desktop, right: movie sequence)

21 Q&A


Download ppt "Building Scalable and High Efficient Java Multimedia Collaboration Wenjun Wu, Tao Huang, Geoffrey Fox Community Grids Computing Laboratory, Indiana University,"

Similar presentations


Ads by Google