Smartphone users judge their apps and devices based on the performance they experience. Yet, because of the subjective and multi-dimensional nature of quality of experience (QoE), objectively measuring it is a serious challenge. Without metrics capable of capturing aspects of interactive performance, mobile operating systems cannot understand how their decisions affect QoE. Instead, we are left to design static OS policies that provide best-effort QoE, and measure overhead in terms of percentage, without understanding the effects it has on QoE.
To quantify aspects of user interaction that have an effect QoE, we propose computing QoE metrics at the point where instances of poor QoE often manifest themselves, namely the screen. Frozen apps, laggy scrolling or typing, and long user transaction latencies are examples of poor QoE that users experience through interacting with the screen. These are also examples of where the OS should understand how its resource allocation and policy decisions affect QoE.