Cost adaptive replica management scheme for S-PQRM system in wireless grid

Abstract

Wireless communication is a rapidly evolving and promising sector of the communications arena, and even in a challenging time for the telecommunications industry. The increasing reliance on wireless networks for information exchange makes it critical to maintain reliable communications in the instants of a component failure. To configure wireless Grids, several key technologies are essential. First, surrogate architecture technology to interconnect resource constrained mobile devices and wired Grids has been proposed. Second, a QoS guaranteeing job processing scheme in Grids using policy quorum based resource management system. Third, replica generation schemes have been proposed to reduce access latency, improve data availability, and increase robustness, scalability and performance. We consider an e-Health care service as a primary application for wireless Grids. The key technology in the wireless Grid application is to keep reliability and transparency in data processing for budget constraints. In previous works, however, little has been studied about how to integrate mobile devices and Grid network with reliability Surrogate architecture does not provide solutions for failures, and QoS constrained policy quorum based resource management scheme does not present reliability and cost optimized scheme in wireless Grids, either. Moreover, in the existing replica generation algorithms, the complexity grows from O($n^{2}$) to polynomially with respect to the number of binary variables. In this thesis, we propose a cost adaptive replica management scheme for reliable surrogate integrated PQRM system in wireless Grids. First, we can improve QoS guaranteed resource management scheme efficiently by integrating surrogate and PQRM logically. Second, we introduce an analytical model for cost adaptive replica management scheme, which will show how to configure wireless Grids using replicas to guarantee reliability. Third, we present a heuristic algorithm to find opt...