Mobile computing algorithms and systems for user-aware optimization of enterprise applications
Moravapalle, Uma Parthavi
MetadataShow full item record
The adoption of mobile devices, particularly smartphones, has grown steadily over the last decade, also permeating the enterprise sector. Enterprises are investing heavily in mobilization to improve employee productivity and perform business workflows, including smartphones and tablets. Enterprise mobility is expected to be more than a $250 billion market in 2019. Strategies to achieve mobilization range from building native apps, using mobile enterprise application platforms (MEAPS), developing with a mobile backend as a service (mBaaS), relying on application virtualization, and employing application refactoring. Enterprises are not yet experiencing the many benefits of mobilization, even though there is great promise. Email and browsing are used heavily, but the practical adoption of enterprise mobility to deliver value beyond these applications is in its infancy and faces barriers. Enterprises deploy few business workflows (<5 percent). Barriers include the heavy task burden in executing workflows on mobile devices, the irrelevance of available mobile features, non-availability of necessary business functions, the high cost of network access, increased security risks associated with smartphones, and increased complexity of mobile application development. This dissertation identifies key barriers to user productivity on smartphones and investigates user-aware solutions that leverage redundancies in user behavior to reduce burden, focusing on the following mobility aspects: (1) Workflow Mobilization: For an employee to successfully perform workflows on a smartphone, a mobile app must be available, and the specific workflow must survive the defeaturization process necessary for mobilization. While typical mobilization strategies offer mobile access to a few heavily-used features, there is a long-tail problem for enterprise application mobilization, in that many application features are left unsupported or are too difficult to access. We propose a do-it-yourself (DIY) platform, Taskr, that allows users at all skill levels to mobilize workflows. Taskr uses remote computing with application refactoring to achieve code-less mobilization of enterprise web applications. It allows for flexible mobile delivery so that users can execute spot tasks through Twitter, email, or a native mobile app. Taskr prototypes from 15 enterprise applications reduce the number of user actions performing workflows by 40 percent compared to the desktop; (2) Content sharing (enterprise email): An enterprise employee spends an inordinate amount of time on email responding to queries and sharing information with co-workers. This problem is further aggravated on smartphones due to smaller screen real estate. We consider automated information suggestions to ease the burden of reply construction on smartphones. The premise is that a significant portion of the information content in a reply is likely present in prior emails. We first motivate this premise by analyzing both public and private email datasets. We then present Dejavu, a system that relies on inverse document frequency (IDF) and keyword matching to provide relevant suggestions for responses. Evaluation of Dejavu over email datasets shows a 22 percent reduction in the user’s typing burden; (3) Collaboration: Even though many business processes within enterprises require employees to work as a team and collaborate, few mobile apps allow two employees to work on an object from two separate devices simultaneously. We present Peek, a mobile-to-mobile remote computing protocol for collaboration that lets users remotely interact with an application in a responsive manner. Unlike traditional desktop remote computing protocols, Peek provides multi-touch support for ease of operation and a flexible frame compression scheme that accounts for the resource constraints of a smartphone. An Android prototype of Peek shows a 62 percent reduction in time to perform touchscreen actions.