Detecting and diagnosing performance impact of smartphone software upgrades

Smartphone manufacturers often release software upgrades to their users for improving service performance, patching security vulnerabilities, enhancing device stability, fixing bugs, increasing battery life, or even enriching the graphical user interface. It is crucial to monitor the smartphones after software upgrades to either confirm their expected impacts, or quickly identify any undesirable behaviors. In this paper, we focus on automatically detecting the software upgrades on smartphones and analyzing their service performance impacts. The complex interactions between the software on the smartphones and the cellular networks make it hard to differentiate if the impacts are smartphone-centric, or network-centric. We propose a new approach, SSM (Smartphone Specific Monitoring) for conducting pre/post impact analysis of multiple service performance metrics across smartphones aggregated by their type, make, model and network locations. Using one-year worth of operational network data, we demonstrate the effectiveness of SSM in accurately detecting and diagnosing the performance impact of smartphone software upgrades.