Cut-Load: Application-Unaware Content Partitioning for Web-based Information Access in Wireless Data Networks

Tae-Young Chang, Aravind Velayutham, and Raghupathy Sivakumar
GNAN Research Group
School of Electrical and Computer Engineering
Georgia Institute of Technology

IEEE International Conference on Commnucations (ICC), Istanbul, Turkey, June 2006.


Abstract

Web-based information access suffers tremendously in low-bandwidth wireless data networks due to the non-correlation between the content transferred across the wireless links and the actual data that is used to serve the user requests. As a result, the current web-access mechanisms face such problems as unnecessary bandwidth consumption, large response times, no service for partial disconnections, and low system utilization in wireless networks. In order to solve these problems with web-transfer in wireless networks, we present a new middleware for wireless web-access called Cut-Load, which performs application unaware content-partitioning in the graphical domain residing at both the mobile client and the proxy server that the mobile client communicates with. Cut-Load uses dynamic mode selection, opportunistic hoarding, transparent mode transfer, and display caching for efficient wireless web-access. Through simulations, we compare the performance of Cut-Load with that of the current web-access mechanisms and show that the proposed middleware brings significant performance benefits both in terms of bandwidth consumption and user-perceived response times.


Presentation: [pdf (788KB)]      Full Paper: [pdf (296KB)]