Options
Appamigo: An efficient middleware to record and control remote users’ interactions with their smartphone apps
Journal
ACM International Conference Proceeding Series
Date Issued
2019-02-14
Author(s)
Vyas, Parjanya
Patil, Bhagyesh
Singh, Shubham
Naik, Vinayak
Abstract
Smartphone applications are getting more complicated as they compete with their desktop counterparts regarding the functionalities they provide. Learning to use these applications can be difficult for old, differently abled, or relatively new users of the smartphones. The customer service associates cannot guide them if they don’t have visibility into users’ interactions with their phones at runtime. Software developers need such clarity while debugging the applications to understand what inputs users are providing. In this paper, we present an approach using static analysis and code synthesis to capture the users’ interactions with their smartphone applications. We propose a middleware, in terms of a library, to add record and control capabilities to the applications. We implement it as an Android library, which can be integrated in an automated manner with any Android application. While there exist solutions that grant a complete control of the smartphone to a remote person, our solution finds a middle ground as to not violate user’s privacy and yet provide effective remote interactions. Our middleware is generic enough to support Android devices across all dimensions, resolutions and orientations, yet capable enough to support a versatile set of complex GUI gestures while preserving users’ privacy. To evaluate our solution for its efficiency in terms of ease of use, resources and user-friendliness, we added it to Plaid, a popular open-source Android application. The results show that our solution changed only 0.34% of the LOC of the original source code. We compare the efficiency of our solution with that of TeamViewer using TeamViewer Remote Control and QuickSupport. As compared to QuickSupport, our solution uses 10 times less RAM, 15 times less network, 3 times less CPU, and less than half power. We evaluate the user experience of our solution by a survey which shows that none of the participants felt the UI interference or lag to be excessive.
Subjects