Effective Testing of Android Apps Using Extended IFML Models

Minxue Pan, Yifei Lu, Yu Pei, Tian Zhang, Juan Zhai, Xuandong Li

Research output: Journal article publicationJournal articleAcademic researchpeer-review

3 Citations (Scopus)

Abstract

The last decade has seen a vast proliferation of mobile apps. To improve the reliability of such apps, various techniques have been developed to automatically generate tests for them. While such techniques have been proven to be useful in producing test suites that achieve significant levels of code coverage, there is still enormous demand for techniques that effectively generate tests to exercise more code and detect more bugs of apps. We propose in this paper the ADAMANT approach to automated Android app testing. ADAMANT utilizes models that incorporate valuable human knowledge about the behaviours of the app under consideration to guide effective test generation, and the models are encoded in an extended version of the Interaction Flow Modeling Language (IFML). In an experimental evaluation on 10 open source Android apps, ADAMANT generated over 130 test actions per minute, achieved around 68% code coverage, and exposed 8 real bugs, significantly outperforming other test generation tools like MONKEY, ANDROIDRIPPER, and GATOR in terms of code covered and bugs detected.

Original languageEnglish
Article number110433
Pages (from-to)1-17
JournalJournal of Systems and Software
Volume159
Early online date2020
DOIs
Publication statusPublished - Jan 2020

Keywords

  • Android apps
  • Interaction Flow Modeling Language
  • Model-based testing

ASJC Scopus subject areas

  • Software
  • Information Systems
  • Hardware and Architecture

Cite this