Optimized Application Level Checkpoint Based Load Sharing Model for Heterogeneous Mobile Grid Computing

Abstract

Objectives: Recent technical advances have fueled the popularity of mobile grid computing. Mobile devices such as cellular phones and PDAs are becoming more common due to the diminution in their size and increase of computational power. In addition, wireless networks are also beginning to fill the environment. With these advances, mobile devices are becoming available to act as service providers in Grid. But the mobile environment presents a number of challenges. Analysis: The range of mobile execution platforms now available which introduces the problem of heterogeneity. Heavy weight checkpoints also provide hindrance to achieve this integration. At present, Grid Computing standards, neither state any load sharing architecture and model that integrates mobile devices in Grid computing nor does it provide any policy that hides heterogeneity and overcome memory limitations of mobile devices thus it is still an open research problem. Findings: Mobile Grid computing solutions must be developed that are lightweight, independent of specific platform and a load sharing model for mobile grid computing that distributes computational tasks on heterogeneous mobile devices. Our simulation results show the effectiveness of data optimization techniques for mobile devices, interoperability and proxy performance in heterogeneous mobile environment. Novelty: We propose a novel layered architecture that adjusts the data size of checkpoints at the minimum possible level and a load sharing Mobile Proxy algorithm.