Résumé :

The energy consumption in Information and Communications Technology (ICT) systems is growing and currently rivals the aviation sector in CO2 emissions and could consume up to 20% of global energy production by 2030. While existing research has established relationships between software structure and energy consumption, most remediation approaches can only be applied during or after system construction, when architectural changes become prohibitively expensive.
We propose a modeling language inspired by MoBiCa that enables software architects to quantify, model, and formally analyze the energy consumption of different architectural choices before implementation. Our approach translates system descriptions into Timed Automata for automated analysis using the UPPAAL model checker, allowing architects to compare alternative designs by considering the energy costs of clients, servers, and communication.
The methodology is demonstrated through application to the Java Petstore application, providing architects with a tool to evaluate what-if scenarios and identify optimal or near-optimal energy-efficient architectures during the design phase.