Measuring the Capability of Smartphones for Executing Context Algorithms

Abstract

While the rise of context aware apps remains, the question arises whether algorithms for context processing can be applied to any smartphone and for any user. In this paper, we propose to test context algorithms for each specific smartphone, user, and situation to which a context algorithm is applied. Towards this, we present our framework which automatically measures all relevant information about an app´s runtime properties, i.e. resource utilization and the time taken to process sensor values by a context algorithm. Both information can be automatically returned to the app developers in order to improve the app. We implemented the framework on Android and tested it using the example of a state-of-the-art context algorithm to find out whether our test smartphone is capable of running the context algorithm without delays and deadlocks.