Energy Consumption Fuzzy Estimation for Object-Oriented Code

Abstract

The energy consumption (EC) estimation of a computing system is a primitive approach for evaluating its energy efficiency and for further optimization. Existing studies adopt the runtime-dependent approach to measure the EC of software; however, in this paper, the source-level and fuzzy estimation approach are employed to measure the EC of software code, especially object-oriented code, before it is executed. This approach is beneficial to source code quality improvement and EC static optimization. For runtime independence, a fuzzy energy consumption (FEC) model is proposed, in which the code and the EC model, as well as the mappings between them, are first defined; then, the process for estimating the FEC of an arbitrary statement is described, including the statement features, data preparation, fuzzy comprehensive evaluation, and fuzzy pattern matching. Finally, experiments are performed, including the regularities verification, the statement features analysis, the EC and FEC comparison, and the FEC application. The experimental results show that the mean values of EC/FEC for the selected test cases are stable, their standard deviation is approximately 0.00064, and their mean is approximately 0.0059. By FEC, it is feasible to compare the ECs of code statically with reasonable accuracy.