F. I. C. A.
Flexible Integrated Caching Approach for Efficient Content Delivery in Wireless Internet

---- THECB Award #003656-0108-2001 ----


Start DateJanuary 02, 2002
End Date December 31, 2003 (extended to August 2004)
Budgeted Funds $66,000 ( Intial Grant )
$8,100 ( 2002 Supplemental H.S. Teacher Grant )
$8,100 ( 2003 Supplemental H.S. Teacher Grant )
Investigator Nelson L. Passos

Abstract

Recent growth in bandwidth-intensive and interactive applications require timely access and delivery of content over the Internet. Satisfying quality of service (QoS) requirements (bandwidth, delay, delay jitter) for real-time applications on wireless Internet is even more challenging due to such limitations as low bandwidth, high error rate and frequent disconnections of wireless links. In addition, host mobility leads to dynamic topology changes and hence further challenges to successful routing and content delivery. Performance enhancing techniques like caching, prefetching, and push caching (broadcast disks) have been proposed to cope with these limited resources and the dynamics of wireless mobile networks. Prefetching allows clients to access remote data in anticipation whereas in push caching, the information source delivers data, anticipating requests from clients. Both prefetching and push caching entail extra burden on network bandwidth, necessitating prudent decision making for their exploitation.

Moreover, these static schemes do not adapt to changing resource availability. An adaptive scheme called Flexible Integrated Caching Approach (FICA) which dynamically maintains an optimal set of documents proximal to mobile clients is proposed. The optimal set is selected from a combination of cached, prefetched, and push cached items, each with a weighted profit value based on its hit ratio, delay saving ratio, prefetch gain and size, available network bandwidth, clients' mobility profile and QoS requirements. Here cache refers to the available disk space at mobile hosts and base stations. The goal of FICA is to optimize profit in terms of resource (bandwidth and disk space) utilization, and minimize overall response time by prefetching/transmitting documents in anticipation. Due to its tunability to user/system requirements, FICA is well suited for scalable wireless Internet applications. The main contributions of the project include, 1) development of algorithms for computing the utility of documents, 2) development of optimization schemes for determining the optimal set of documents in the cache, 3) development and testing of software that implements the above algorithms in an experimental wireless Internet. A software suite will be developed to to experimentally validate proposed algorithms and models under various user and networking conditions. Graduate students will be exposed to research outcomes in courses on distributed systems, wireless networks, and mobile computing. Published papers, reports and software will be shared in the public domain.

[ Published Work ] [ Top ]

Last Updated: April 6, 2004 since March 2, 2000.