Definition, implementation and validation of energy code smells: an exploratory study on an embedded system

Optimizing software in terms of energy efficiency is one of the challenges that both research and industry will have to face in the next few years.We consider energy efficiency as a software product quality characteristic, to be improved through the refactoring of appropriate code pattern: the aim of this work is identifying those code patterns, hereby defined as Energy Code Smells, that might increase the impact of software over power consumption. For our purposes, we perform an experiment consisting in the execution of several code patterns on an embedded system. These code patterns are executed in two versions: the first one contains a code issue that could negatively impact power consumption, the other one is refactored removing the issue. We measure the power consumption of the embedded device during the execution of each code pattern. We also track the execution time to investigate whether Energy Code Smells are also Performance Smells. Our results show that some Energy Code Smells actually have an impact over power consumption in the magnitude order of micro Watts. Moreover, those Smells did not introduce a performance decrease.