دانلود رایگان مقاله انگلیسی بهبود پایداری سیستم عامل اندروید سریع برای مشاهده برنامه - IEEE 2018

Abstract

Application analysis is important for smartphone markets and application developers. However, dynamic analysis involving the execution of the actual application is highly time consuming; thus, reducing the observation time is important issue. In this study, we focus on the Android operating system, which is based on a Linux kernel, and a method for accelerating the time flow speed as recognized by the processes in the system. The method accelerates the speed by modifying the implementation of time management in the kernel, thereby reducing the time required to monitor the behavior of applications. We discuss the stability and accuracy of the method with high acceleration ratios. First, we evaluated the method with a variety of ratios. The evaluation demonstrated that the method does not function accurately and stably with high acceleration ratios, whereas it is effective with low ratios. Second, we investigated the reason for this instability and showed that accelerated time flow causes several timers to experience timeouts, which results in system restarts. Third, we explored a method to prevent restarting, which we tested by disabling these timers and observing the system behavior.

VI. CONCLUSION

In this paper, we introduced an important problem relating to the dynamic observation of Android applications, namely that this analysis is highly time consuming. We presented a method for reducing the observation time by modifying the implementation of time management in the kernel. We evaluated the method by using high acceleration ratios. Our experiments revealed the adverse effect of the system stability to be attributable to the system process rebooting in the user space. We then investigated the causes of system failure and found that process terminations caused by timeouts that occurred during accelerated time were an important cause. For testing purposes, we implemented the operating system such that these terminating functions were disabled, thereby improving the system stability, as our tests confirmed. In future, we plan to evaluate our method with even higher accelerating ratios to enable us to propose a method for improving the system stability.