Fault Tolerance Technique Based on Monitoring and Pattern for Reliable Resource Management in Mobile Cloud Computing

Abstract

Recently, with the rapid growth in the smart phone market and advancements in wireless communication environments, mobile devices are being considered as resources for the processing of fine-grained tasks. However, mobile devices have several problems such as unstable wireless connections, low communication bandwidth, and frequent location changes. As resource providers, mobile devices can join and leave the distributed computing environment in an unpredictable manner. This frequent movement interrupts the operation in progress, and the delay or failure in completing the operation may cause a system failure. Therefore, we should consider volatility, one of the dynamic characteristics of mobile devices, for reliable resource management. In this study, we define a resource availability model for reliable resource management, and we propose a monitoring system and a fault tolerance technique based on the proposed model. In our proposed monitoring system, we determine the operation availability state dynamically according to the operation availability state of predicted resource information. We define a set of patterns of operation availability states, which describe the state of mobile devices. Our proposed fault tolerance technique creates checkpoints and replications according to the patterns of operation availability states.