应对智能移动计算平台演化的应用软件质量保障技术的研究

Mobile computing has brought a profound change to the information industry. Mobile devices equipped with intelligent operating systems have surpassed desktop PCs and become the primary computing platforms. The Android platform, with the largest global market share, provides various application software to assist users in a wide variety of daily activities (e.g., work and entertainment). Unfortunately, the Android ecosystem evolves at a fast pace and is heavily fragmented, posing serious threats to the quality of Android applications. Yet, the current research on this problem is still preliminary and the issue detection capability of existing integrated development environments is very limited. Thus, Android application developers face an unprecedented challenge of assuring the software quality of their products..In this project, we plan to systematically study how Android platform evolution affects the functionality, reliability, performance, security, and compatibility of Android applications via analyzing open-source software big data using techniques including program analysis and natural language processing. By studying the co-evolution of the Android platform and the applications running on it, we aim to mine and understand various software defect patterns to support automated quality assurance of Android applications. Based on the mined software defect patterns, we will further design static program analysis algorithms that support parallelism and incremental analysis to efficiently and effectively pinpoint software defects in Android applications as well as providing patching aids. The proposed techniques will help Android developers and markets largely improve the software quality of Android applications.

移动计算给信息产业带来深刻的变革,基于智能操作系统的移动设备已超越PC成为主流计算平台。Android是现今全球市占率最高的移动平台，其上的各类应用影响着人们生活的方方面面。然而，Android平台演化频繁，生态系统深度碎片化，给移动应用软件带来严重的质量隐患，而现有研究比较初步，集成开发环境的纠错能力也极为有限，应用开发人员因而面临前所未有的挑战。为了解决这个问题，本项目将基于开源软件大数据，结合自然语言处理和静态程序分析等方法，系统性地研究Android平台演化如何影响应用软件的功能、可靠性、性能、安全性、兼容性等质量指标。通过分析Android平台和应用软件的协同演化，本项目将深入挖掘Android平台演化导致的各类软件缺陷模式以支撑自动化的软件质量保障。最后，本项目将设计并行化增量式静态程序分析算法以实现高效精准的软件缺陷检测及修复辅助，帮助开发人员和应用市场大幅提升应用软件质量。

Android生态系统的碎片化造成大量的应用兼容性问题。本项目研究了907个真实应用中出现的设备无关API兼容性问题、设备相关API兼容性问题、配置兼容性问题、以及Android平台安全机制演化导致的动态权限误用问题，总结了19种缺陷模式。在大规模实证研究的基础上，我们进一步提出了两种Android平台演化的建模方法和一套用于兼容性缺陷检测的静态分析框架，其中包含4个实用工具：FicFinder，Pivot，ConfDroid，Aper。运用这些工具，我们在627个开源Android应用软件中发现了270个未知缺陷。本项目的研究有助于加深科研人员和软件从业人员对移动软件缺陷的理解，揭示平台演化对应用软件质量的负面影响，以及完善移动应用软件质量保障的理论体系和工具链。