A framework to assess Android mobile applications energy consumption and privacy policies.

点击翻译按钮获取中文摘要

Why is my cell phone battery already low? How did it happen that almost all the data of my monthly phone plan is already used up? Why is the new app tracking my location? These are non-obvious questions. Choosing apps requires effort, patience to examine pertinent information, and technical knowledge. In fact, the best app combination depends on user preference, frequency of use of different features, single feature performance, user privacy concerns, and app (hidden) costs. If we consider the Android ecosystem, for any given app category (e.g., e-mail clients, weather forecast, business, communication, tools, games), the user has plenty of choices for different apps (often implementing the same, or very similar, features). However, quite often, while an app category lists dozens of apps, the information supplied is limited to the advertisement for a specific app, the Android market meta-data, and the app ranking. When setting up an Android phone environment, the user has to try different apps, verify implemented features, evaluate intrusiveness of advertisements, ponder pros and cons of different apps, hoping that, in doing so, no personal or confidential information is leaked. This research program is directed to the development of a framework that will support users and developers. We aim at supporting users in selecting the best set of apps, given the user needs, and supporting developers in designing and implementing efficient and energy aware apps. Specifically, we will support the analysis, assessment and verification of mobile apps’ 1) energy consumption per specified scenario; 2) app implemented features (and feature specific energy consumption); and 3) app privacy implemented policies. In the short term, we will make available a WEB portal to users and selected developers. Users will be able to select some app categories, set their energy, network, and confidentiality concerns (high, medium, low or don’t care), specify the list of grants they are willing to give to apps, and, automatically, obtain a list of optimal, or close to optimal, apps to be installed in their phones. We will generalize this with a scenario definition language. Also, we will offer selected developers the possibility to upload an app, select a scenario (define an app specific scenario if needed), and produce an app energy consumption profile. In the long term, we aim at extending the tool chain supporting 1) mixed language apps (native C and Java or pure C/C++ apps); 2) app and app execution scenarios co-evolution; and 3) different hardware and Android versions. These are all non-trivial and challenging goals. Different hardware implies different screen geometry, device drivers and Android API. App and Android evolution may substantially modify the way in which the user and the app interact: scenarios, privacy policies, and setup to gather information leading to the need of app and scenarios co-evolution.