AbstractsComputer Science

As today’s smartphone hardware becomes increasingly powerful with more CPU cores, larger size of RAM, and even graphics multi-processors, the complexity of smartphone apps is approaching that of a desktop computer. However, Random smartphone rebooting and unexpected shutdown (due to reasons like overheating, battery run outs, or automatic app upgrades) can be frustrating. What can be even worse is that a phone user commonly has to restart, from beginning, the apps he or she is using when the phone gets rebooted, as all the app states would be lost, especially when the number of apps is large. Hence, a recovery service is sorely need-ed for today’s smartphones where apps are becoming increasingly complex. While checkpointing has long been used for desktop and laptop computers, such state p-reserving techniques cannot be directly applied to smartphones, due to their more stringent constraints on energy, running time, and storage space. In this paper, we propose CKPTriod, an checkpointing framework that checkpoints the apps based on the apps’ characteristics and future usage, in order to signi¿cantly reduce the amount of work that a user needs to recover the app states after restarting the smartphone. CKPTriod is prototyped on the Android platform and evaluated using real-world traces. Advisors/Committee Members: Wang, Xiaorui (Advisor).