An optimal application partitioning and computational offloading framework for mobile cloud computing

---

Tezin Türü: Doktora

Tezin Yürütüldüğü Kurum: Orta Doğu Teknik Üniversitesi, Enformatik Enstitüsü, Bilişim Sistemleri Anabilim Dalı, Türkiye

Tezin Onay Tarihi: 2016

Öğrenci: MAHİR KAYA

Danışman: ALTAN KOÇYİĞİT

Özet:

The use of mobile applications is increasing every day and they offer more functionality on mobile devices. However, these devices are inferior to server computers in terms of memory and processor capacity. Furthermore, rapid depletion of mobile devices’ energy resources is still a major problem. Performance and energy shortcomings of mobile devices can be improved by using surrogate or cloud computing technologies. In this thesis, an offloading framework is proposed to improve the performance and efficiency of mobile applications. The framework seamlessly handles offloading and provides distribution transparency via the Inversion of Control mechanism. In particular, computation intensive components of an application are run on a remote server. It is possible to migrate different combinations of components to remote servers. Indeed, offloading some combinations of components are productive and others are counterproductive. Experimental results show that offloading the optimal combination of components to remote servers reduces the execution time and energy consumption of mobile devices. Hence, a call graph model is proposed to decide on the components to be offloaded. Offloading decisions are made by finding the best partitioning in the graph. The graph model has been validated by extensive experiments.