Passive profiling of mobile engaging behaviours via user-end application performance assessment

Abstract

Benefiting from the booming mobile industry, our daily connections have become instant and ubiquitous. In the meanwhile, user experience in mobile services becomes an important consideration for service providers to cultivate customers’ loyalty, or for network operators to seek profit chances from existing infrastructure. To meet these requirements, we specify engaging behaviours to characterize the dynamics of user participation in mobile applications in real contexts. Leveraging a five-month collection of backbone traffic in an operational WiFi mesh network, we make comprehensive insights on the variation of engaging behaviours, as well as their non-linear and counter-intuitive interactions with user-perceived application performance. With the Hidden Markov Modelling of individual’s engaging trajectory, two distinctive behavioural clusters are identified and investigated. These achievements will be contributed to the development of multiple pioneer R&D areas, such as mobile network simulation and user experience optimization.