All aspects of our products and services should both meet and surpass our customer’s expectations. To achieve this we have to prioritise the end user experience.
Industry PhD and data scientist Anders Skretting has done extensive research into the importance of optimisation and performance. Especially when developing software for resource constrained devices such as smartphones.
Skretting has published the paper “Baseline for Performance Prediction of Android Applications”, following a presentation at the IEEE Big Data 2020 conference. The conference paper investigates what could be expected of different apps on different devices in terms of energy consumption, memory usage, graphical performance, and network traffic. He profiled and compared various performance metrics of applications in the Play Store that had been given the Android Excellence award.
Android Excellence is awarded to applications that deliver incredible user experiences on Android, as well as using many of the best practises, having great design, technical performance, localisation, and device optimisation. It is thus reasonable to assume that the metrics profiled from these applications would serve as a baseline for how other well-performing applications should behave in terms of energy consumption, memory usage, graphical performance, and network traffic.
The metrics presented in the paper can form the foundation of machine learning algorithms that may be able to automate and improve the current manual and non-trivial effort that developers face today when evaluating the performance of their applications. Another aspect here is that of malware detection. By taking advantage of the baseline provided in this paper, malware detection algorithms can be developed and improved. The performance metrics can also serve as a reference for both researchers and developers who are interested in knowing whether their solutions deviate significantly from what is expected of well performing Android applications.
You can find the published paper here.
Skretting continues to conduct research into performance related topics. He is currently investigating how distributed computation offloading to nearby smartphones can enhance performance of mobility solutions. This is especially interesting and important for his work with colleagues and partners in the ongoing research project Be-Insight (automated ticketing). For instance, by sharing contextual data with nearby devices, computations can infer which vehicle you are in, or what activity you are doing, which in turn potentially could be enhanced both in terms of computation speed and a reduced performance impact. Preliminary results are very promising.
Kogenta regards optimisation and the performance of our products, along with end user privacy, of the utmost importance, thereby putting efforts and resources into innovative research. We hope to continue to provide the industry with novel insights, as well as providing clients with the best products possible.
Stay tuned for more insights and upcoming research findings.