Optimized Communication For Mobile Multimedia Collaboration Applications Sangyoon Oh Community Grids Laboratory Indiana University CTS`05 St. Louis May 6 th 2005

2 Contents Introduction Related Works Handheld Flexible Representation Examples and Experiments Conclusion

Introduction

4 Why Interesting? Prospective of popularity Recent prominent data sharing: mobile computing and other distributed system. SOAP Interoperability  Connects disparate and distributed resources. Open Standard Community (W3C), Industry (Nokia, Sun, etc) Binary Interchange of XML Infoset Workshop (Sep. 2003).

5 Problems Possible performance overheads in Web services XML needs considerable resources for transmitting and parsing. Information storing in textual format XML’s descriptive tags and structure (increasing size)  Add parsing and transmission overheads. Mobile environment Battery-constraint, limited computation High-latency, intermittent communication Compression of XML (GZip XML) would increase a processing time and size of small message

6 Proposal of Flexible Representation Purpose of Research: To design and develop an optimized communication framework preserving SOAP semantics for mobile computing. Characteristics of research Separate data presentation format and data content Loss of self description Saving parsing computation and data binding >> Latency Saving  High latency of cellular network

Related Works

8 Classification of SOAP Alternatives Some data don’t fit in XML serialization : MTOM/XOP Media data Standard format with compression JPEG, GIF, MP3 Data that include digital signature Binary integrity would not be preserved after serializing into textual format Some application domains use a binary representation to avoid verbose XML encoding.  Fast Web Services/Fast Infoset, Extreme! Lab Send pre-parse data over binary communication

9 Summary of projects Extreme! LabMTOM/XOP Fast Web Services Properties Negotiation, Binary Application Protocols Support binary contents GZip for XML Target Scientific Data Media File, Data that includes digital signature Java Application (No changes on Application level) Lack of mobile issues QoS Performance, Optimized message scheme for in-memory representation Fast Infoset is not interoperable, QoS

Handheld Flexible Representation

11 Flexible Representation in general Using optimized data representation to improve communication performances preserving SOAP semantics Binary format stream Negotiation Quality of Service (reliability and security) Imposing XML problems Hard to address on individual message, but appears to be possible to combine SOAP structure and binary format in stream processing. Work best for a stream Designed for high-frequency small messages. Sharing message structure and WSDL information. SOAP headers are largely unchanged.

12 Example of Filtering Data Representation <representation repType="binary“ byteOrder="bigEndian"/> <xs:sequence minOccurs="number of points" maxOccurs="number of points"> (XML Representation)(DFDL Description)

13 Simple Scenario with negotiation 1.Mobile device which is capable of Flexible Representation initiates a negotiation with the mobile gateway. 2.SOAP message for negotiation is sent over the conventional SOAP. 3.mobile gateway responds to the negotiation request with its capability in the SOAP message. 4.response with message template/XML schema and quality of service issues. 5.agreed message format (Flexible Representation of SOAP Infoset) throughout the session.

14 Software architecture ContextHandler Schema Parser - Data Structure generator Stream Reader/Writer Context server Stores SOAP header of initial message (negotiation request message) Message schema URI of Flexible Representation Specification ( ) Each ContextBlock has a pointer to URI to Context server. Currently, Ad-hoc scheme to store context  WS-Context or any community-specific.

15 Relationship among different forms of SOAP messages and their defining context

16 HHFR in GlobalMMCS 1. Up-streaming the captured Image Periodically captured images are converted into stream using JMF library. 2. Down-streaming Video Session Porting the conventional GlobalMMCS client on mobile. Filtering the frame rate and image size. MMAPI (JRS 135) in J2ME MIDP 2.0

17 Snap Shots

Conclusion and Future Works

19 Conclusion and future works Conclusion Negotiation and byte stream which is preserving SOAP infoset help to overcome possible performance overhead of mobile web services Help the integration of mobile device and collaboration systems in web services. Future works QoS Handler Reliable Messaging – WS-WRM Security – Light weight security model Up-stream client porting on mobile

Appendix

21 Current status of 3G in the US Cingular GPRSCingular EDGECingular UMTSVerizon EV-DO Rate 56kbps (max.) 30~40 kbps 120kbps (aver.) Download: 384kbps (max.) 290kbps (aver.) Upload: 56kbps Download: 300~500kbps Upload: 90kbps CoverageNation-wide Nation-wide (except Montana, Dakotas and Virginia) Only six cities 32 cities (expensive plans) Voice and data -- Simultaneous usage --

22 Wireless Tech. Comparison Bandwidth Comparison of IR, WLAN and Bluetooth Serial Infrared115.2kbps Bluetooth kbps Bluetooth Mbps b11Mbps g54 Mbps

23 Handheld Messaging Service General mobile communication framework by CGL Provides a core pub/sub API for mobile application. Lightweight user library Server-side gateway TCP and HTTP transport HHFR is a part of HHMS

24 GlobalMMCS architecture