Published Paper Details:

Software energy use is now a measurable, growing slice of global electricity demand, and much of that footprint comes down to how code is written, not what hardware it runs on. Green refactoring, restructuring existing source code to cut its energy footprint without changing what it does, has been proposed as a practical way to close that gap without waiting on new hardware or new languages. This paper reviews the literature on green refactoring for Java, a language that still runs most enterprise and Android backend systems even though sustainability research has largely moved on to mobile and cloud-native platforms. We searched five major databases, applied a structured inclusion and exclusion protocol, and screened the results down to a final set of primary studies covering refactoring techniques, profiling tools, and their reported energy effects. The review sorts green refactoring into six technique categories, compares five widely used Java profiling tools, and lays out what the empirical literature agrees and disagrees on. The clearest finding is that results are inconsistent: energy savings from the same technique vary by workload, JVM version, and measurement tool, and few studies isolate cause from effect cleanly. We identify six open gaps, among them the lack of multi-metric evaluation standards and the near-total absence of automated, profiling-guided refactoring tools for Java, and close with six directions for future work, including AI-assisted refactoring and energy-aware practices for microservice architectures.