On the Performance of Cloud-Based mHealth Applications: A Methodology on Measuring Service Response Time and a Case Study

With the increasing use of smartphones, performance monitoring and the analysis of mobile applications (apps) are gaining momentum. Smartphones are resource-constrained devices. Thus, mobile apps typically rely on cloud services for the execution of resource-intensive functionalities, storage, and computation power. Measuring the user experience is crucial for the development and maintenance of mobile apps. Such characterization requires testing specific traits such as network connectivity, battery levels, server loads, and operating conditions. This paper presents a technique for the measurement-based performance assessment of cloud backend and mobile networks that support mobile app services. The feasibility of the technique is demonstrated through a representative case study of an app developed for medication adherence management among breast cancer patients undergoing endocrine hormone therapy (EHT). The app leverages cloud technologies to provide a portable, cost-effective, and convenient monitoring environment. Nonfunctional performance and load testing is performed by modeling third-party cloud backend services. The experimental results of the case study demonstrate the feasibility of the approach for monitoring and analyzing the backend service response times with different mobile device configurations, such as regular or power-saving battery modes and LTE or Wi-Fi mobile network connectivity, under server loading. The methodology is validated through statistical analyses of the experimental performance data involving confidence intervals, tail latencies, and analysis of variance. The results address the occurrence of server loading and its impact on the response times which relates to the quality of the user experience. We establish the effect of server loading on the responsiveness of the user interface (UI) of the mobile app considered in this case study. The proposed technique will allow developers to conduct similar measurement-based performance studies for various mobile apps leveraging cloud-based backend services.