Zhijun Wang

Graduation Semester and Year

2005

Language

English

Document Type

Dissertation

Degree Name

Doctor of Philosophy in Computer Science

Department

Computer Science and Engineering

First Advisor

Sajal Das

Abstract

With the tremendous growth of applications in wireless mobile and Peer-to-Peer (P2P) networks, significant research efforts have been made to improve the quality of sevice. Caching and replicating frequently used data objects or files in user's local buffers are popular mechanisms to effectively reduce the communication bandwidth requirement and thus improve the overall system performance. However, the frequent disconnections of users make data consistency a difficult task in wireless mobile and P2P networks. In this dissertation, we design and analyze a Scalable Asynchronous Cache Consistency Scheme (SACCS) for single cell wireless cellular networks. SACCS is a highly scalable, efficient, and low complexity scheme and works well in error-prone wireless mobile environments. Analytical results indicate that SACCS provides very good cache consistency in error-prone wireless environments. Comprehensive simulation results show that SACCS offers more than 50% performance gain than that of existing Timestamp (TS) and Asynchronous Stateful (AS) schemes; We also propose Dynamic SACCS (DSACCS) for multi-cell mobile environments. To the best of our knowledge, DSACCS is the first cache consistency scheme that optimizes cache performance in multi-cell mobile environments. In P2P networks, some files are heavily replicated to enhance their availability and reduce the search cost. With the dramatic growth in P2P applications dealing with dynamic files, file updates and the file consistency maintenance become critical. To effectively propagate update information to the replica peers, we propose an efficient algorithm, called Update Propagation Through Replica Chain (UPTReC), in decentralized and unstructured P2P networks to provide weak file consistency. To provide strong file consistency, we develop another algorithm, called file Consistency Maintenance through Virtual servers (CMV). In CMV, each dynamic file has a virtual server and any file update must be accepted through the virtual server to maintain one copy serilizability of the file. To the best of our knowledge, CMV is the first strong file consistency algorithm for decentralized and unstructured P2P networks. Our simulation results show that UPTReC algorithm outperforms other existing algorithms, and CMV is an efficient file consistency algorithm with very low overhead messages.

Disciplines

Computer Sciences | Physical Sciences and Mathematics

Comments

Degree granted by The University of Texas at Arlington

Download

Share

COinS