AN INTEGRATED STATIC AND DYNAMIC ANALYSIS FRAMEWORK FOR ANDROID APPLICATION MAINTAINABILITY AND SUSTAINABILITY ASSESSMENT

The paper describes the design and development of ASDA, a comprehensive analysis framework for Android applications, which employs code analysis techniques, machine learning, and performance analysis at run time to offer end-to-end results on code quality and maintainability. The maintainability debts of applications are analyzed at design time by extracting metrics from Java code using the CK metrics tool to build machine learning models for predicting the Maintainability Debt Index (MDI) for classes. Four regression analysis techniques, namely Random Forest, Gradient Boosting, Extra Trees, and Voting Ensemble, are used to identify the best-suited regression model for maintainability prediction, which is Random Forest (RMSE = 0.082, R² = 0.764). The results of dynamic analysis techniques are obtained from Perfetto traces and used to measure CPU, context switches, method performance times, and estimated energy consumption. The performance results are incorporated into the design of the maintainability and sustainability debt metrics framework for run-time analysis of applications, which include aspects of execution, energy, responsiveness, and concurrency. The results clearly show direct correlations between the predicted and actual energy consumption of applications (R² = 0.93), thereby confirming the effectiveness of the framework. The combination of design-time and run-time analyses helps to identify classes for which maintainability and resource intensity are desirable goals and enables prioritizing of changes to Android applications. The results of the design and development of ASDA clearly illustrate its utility in Android applications as a comprehensive analysis framework that reduces maintainability and sustainability debt and reduces energy consumption in applications through tool-driven changes by up to 89.2% in maintainability and 90.3% in energy consumption in applications for redeveloping them as environmentally sound and resource-efficient applications. The design and development of ASDA clearly illustrate its utility as a comprehensive framework for Android applications at design time that provides end-to-end results on maintainability and resource efficiency of applications for developing resource-efficient and environmentally sound applications. The design and development of ASDA clearly illustrate its utility as a comprehensive framework at design time that provides end-to-end results on maintainability and resource efficiency of applications for developing resource-efficient and environmentally sound applications. KEYWORDS