NavyDroid: Detecting Energy Inefficiency Problems for Smartphone Applications

Abstract

Many smartphone applications suffer from energy inefficiency problems, but locating these problems is quite difficult and labor-intensive. Automated tools for detecting energy inefficiency bugs have been shown to be effective. Existing approaches generally consist of two parts, namely the simulation part and the monitor part. The simulation part explores an application’s state space guided by an application execution model, and the monitor part checks for occurrences of energy inefficiency patterns. However, existing approaches might miss energy inefficiency bugs due to their imprecise application execution models and oversimplified energy inefficiency diagnosis policies. In this paper, we proposed NavyDroid, an approach to diagnosing energy inefficiency problems more effectively. We summarized a comprehensive application execution model from Android specifications and expressed it as a state machine. By considering multiple patterns of wake lock misuses, our approach is able to detect more complex energy bugs caused by wake lock misuses. We implemented NavyDroidon top of Java Pathfinder (JPF) and applied it to real-world applications. We evaluated NavyDroid with 17 real-world Android applications, and NavyDroid located more energy inefficiency bugs in these applications than the existing work E-GreenDroid did. The results of our experiments demonstrate that our approach can effectively locate real energy inefficiency bugs in Android applications, suggesting its effectiveness.