eGEN: An Energy-saving Modeling Language and Code Generator for Location-sensing of Mobile Apps

TL;DR

eGEN is a tool that helps developers create energy-efficient location-based mobile apps by generating battery-aware code during the design phase, significantly reducing GPS usage and battery consumption with minimal impact on location accuracy.

Contribution

This paper introduces eGEN, a novel energy-saving modeling language and code generator specifically for location sensing in mobile apps, supporting native Android development.

Findings

Average GPS usage reduced by 4.35 minutes per hour

Battery consumption decreased by 188 mA

Location accuracy degraded by only 97 meters over 3 km

Abstract

The demand for reducing the energy consumption of location-based applications has increased in recent years. The abnormal battery-draining behavior of GPS makes it difficult for the developers to decide on battery optimization during the development phase directly. It will reduce the burden on developers if battery-saving strategies are considered early, and relevant battery-aware code is generated from the design phase artifacts. Therefore, we aim to develop tool support, eGEN, to specify and create native location-based mobile apps. eGEN consists of Domain-specific Modeling Language (DSML) and a code generator for location-sensing. It is developed using Xtext and Xtend as an Eclipse plug-in, and currently, it supports native Android apps. eGEN is evaluated through controlled experiments by instrumenting the generated code in five location-based open-source Android applications. The experimental results show 4.35 minutes of average GPS reduction per hour and 188 mA of average reduction in battery consumption while showing only 97 meters degrade in location accuracy over 3 kilometers of a cycling path. Hence, we believe that code generated by eGEN would help developers to balance between energy and accuracy requirements of location-based applications. The source code, documentation, tool demo video, and tool installation video are available at https://github.com/Kowndinya2000/egen.