Code Offloading and Resource Management Algorithms for Heterogeneous Mobile Clouds
AffiliationComputing and Information Systems
Document TypePhD thesis
Access StatusOpen Access
© 2018 Dr. Bowen Zhou
The recent innovation and development of software and hardware on mobile devices such as smartphones and tablets have made them evolve as the primary tool of our digital life. Mobile device users have increased their demands on more PC-like user experiences such as mobile gaming, augmented reality, and mobile version of legacy PC applications. However, the battery lifetime remains as a weakness due to the advanced display, camera and sensors that drain the battery quickly. Since the battery technology seems unlikely to have a significant improvement in the foreseen future, other solutions are needed. The heterogeneous mobile cloud paradigm emerges as a promising solution. It defines a wireless, shared computing resource environment that consists of ad-hoc connected mobile devices, nearby form-factor servers (cloudlets), and cloud computing services. Mobile devices can leverage the shared resource environment to offload their computation intensive tasks in order to conserve battery lifetime and accelerate application performances. However, with such a loosely coupled and mobile device dominating network, new challenges and problems emerge such as how to enable task offloading in such environment, how to achieve minimum time and energy consumption through task offloading and scheduling, how to maintain the service reliability and recover from failures, and how to incentivize ad-hoc mobile users to use mobile cloud offloading services. This thesis studies algorithms and technologies to solve these problems and provides a task offloading framework for the heterogeneous mobile cloud service. It extended the state-of-the-art by making the following key contributions: 1. A system architecture for heterogeneous mobile cloud services and a system framework implemented on Android platform to enable the mobile cloud offloading service. 2. A context-aware offloading algorithm for individual mobile devices to make task offloading decisions based on the context changes in the heterogeneous mobile cloud network. 3. An optimal offline algorithm and an online algorithm based on the ski-rental framework with near-optimal performance guarantee for providing task offloading and scheduling decisions on minimizing overall task execution time, considering the load balancing of all devices in the heterogeneous mobile cloud network as well as the unique constraints such as battery limit and offloading enhancement. 4. A group-based fault tolerant mechanism for improving the service reliability of heterogeneous mobile cloud offloading. It classifies mobile devices into groups based on its processing capacity. mobility, and reliability. Different fault tolerant technologies such as checkpointing and replication are devised adaptively based on the task offloading schedules and the specific group of machines it’s offloaded. 5. A reverse auction based incentive mechanism and an optimal offline model for increasing the participation and utilization of the proposed mobile cloud offloading services. The proposed algorithm guarantees computational efficiency, truthfulness, individual rationality and near-optimal auction results.
Keywordsmobile cloud computing; code offloading; resource management
- Click on "Export Reference in RIS Format" and choose "open with... Endnote".
- Click on "Export Reference in RIS Format". Login to Refworks, go to References => Import References