Ashish Bijlani

Ph.D. Student

School of Computer Science

Georgia Institute of Technology

Date: Thursday, September 13th, 2018

Time: 11:00am EDT

Location: Klaus 3126

Committee

Dr. Umakishore Ramachandran (Advisor, School of Computer Science, Georgia Institute of Technology)

Dr. Mostafa Ammar (School of Computer Science, Georgia Institute of Technology)

Dr. Ada Gavrilovska (School of Computer Science, Georgia Institute of Technology)

Dr. Vivek Sarkar (School of Computer Science, Georgia Institute of Technology)

Dr. Raghupathy Sivakumar (School of Electrical and Computer Engineering, Georgia Institute of Technology)

Abstract

Smart mobile devices have largely evolved as primary tools for everyday personal computing needs, including communication, travel and planning, gaming, health tracking, social networking, home automation, media, and entertainment. Personal mobile devices are also increasingly being used for work. The versatility of these devices, however, poses high, and often conflicting, storage demands. While historically small in size (25MB), modern mobile apps are feature-rich, large universal packages, each capable of consuming over 4GB of storage space. As a result, devices quickly run out of space given their limited storage capacity. Yet, apps are not constrained by storage quota limits. 

This work first shows that mobile apps are heavily bloated: not all installed apps (or features) are equally important to the user at all times and proposes a context-sensitive quota model for automatic storage management on smart mobile devices. To understand the design challenges and performance implications, we are building a platform storage service that performs proactive management of local storage. Storage space consumed by contextually unwanted apps/data is temporarily reclaimed by employing multiple standard techniques, such as compression, deletion, content adaptation, deduplication, and cloud-backed hierarchical management. Reclaimed data is reconstructed proactively under predictive usage or served on-demand either by computation on the device or fetching it from the cloud.