This work considers watch faces for Android Wear devices such
as smartwatches. Watch faces are a popular category of apps that
display current time and relevant contextual information. Our
study of watch faces in an app market indicates that
energy efficiency is a key concern for users and developers.

The first contribution of this work is the definition of several
energy-inefficiency patterns of watch face behavior, focusing on two
energy-intensive resources: sensors and displays. Based on these
patterns, we propose a control-flow model and static analysis
algorithms to identify instances of these patterns. The algorithms
use interprocedural control-flow analysis of callback methods and the
invocation sequences of these methods. Potential energy inefficiencies
are then used for automated test generation and execution, where the
static analysis reports are validated via run-time execution. Our
experimental results and case studies demonstrate that the analysis
achieves high precision and low cost, and provide insights into
potential pitfalls faced by developers of watch faces.