TrCMP: An app usage inference method for mobile service enhancement

Abstract

In order to improve the quality of life and the efficiency of work, users need timely and accurate services provided by mobile devices. However, for the same service, different users have various personalized use styles, such as usage time, invoking frequency, etc. As a result, the accuracy of real-time service recommendations often depends on effective user behavior analysis. Technically, user behaviors associated with a certain service could be reflected with traffic, CPU, memory and energy consumption during app running. In this paper, an app usage inference method, named TrCMP, is investigated. This method takes Traffic, CPU, Memory and Power into consideration in a comprehensive way for analyzing user behaviors. Extensive experiments are conducted to validate the efficiency and effectiveness of our method.