Titulo Estágio
Mining large datasets to extend the battery uptime of Android devices
Áreas de especialidade
Engenharia de Software
Local do Estágio
DEI-FCTUC
Enquadramento
It is really difficult to describe the gigantic growth of mobile device usage that we observe these days, with, e.g., Zenithmedia.com revealing that mobile devices are expected to drive 80% of global internet usage. This, of course, leads the smartphone application market to follow up. Indeed, according to Statista.com, the total numbers of mobile app downloads in 2017 were 197 billion, a number that is expected to increase to 352 billion by 2021.
Many smartphone applications make use of modern features to provide immersive experiences. For instance, by location sensing, Google Maps can detour drivers to avoid traffic jams or accidents. However, this kind of operations is often energy costly.
Indeed, smartphones in general, and Android smartphones, in particular, are getting increasingly more powerful, resulting in a steady and documented increase in the energy consumption of the devices. However, there has not been a significant major breakthrough in battery capacity for years.
As such, energy efficiency is one of the most critical concerns for smartphone users. Inappropriate battery use can be a reason for users to uninstall applications, or give negative feedback on them.
Locating energy problems in Android applications is difficult and reproducing these energy problems is labor-intensive. Developers have to continuously and extensively test their applications on different environments using various devices, conducting detailed energy profiling in order to know how much their application use the devices hardware components, such as the CPU to figure out the root causing the energy anomalies.
These methods can quickly become cumbersome and typically time-consuming and most importantly, the results obtained not always reflect the usage of the single application that the developers want to profile, but rather the usage of the whole device.
After taking this into consideration, it seems important to address these challenges, and this has to lead to a research initiative, that the interested reader may know more about here:
https://greenhubproject.org/
This is an effort whose insight is to gather (energy-related) inputs from a large collaborative community. This is achieved by creating a comprehensive platform to gather (strictly energy-related) data from the devices used by the members in this (always open) community.
So far, the initiative has collected over 19 million data samples whose systematic exploration can reveal information on how to extend the battery uptime of Android devices.
The way to give back to the community is to afterward make all the collected data, and the information extracted from it, publicly available. This is expected to enhance further research on the topic of Android energy analysis and optimization.
Objetivo
The interested candidate will join the previously mentioned research initiative and will be focused on exploring the data that is gathered by the public platform. The goal is to conduct large scale mining of such data to establish patterns that can be considered good/bad in terms of energy consumption.
These patterns are expected to provide smartphone users insights and feedback on how to increase their battery duration without changing too much their usage pattern.
Not only will the candidate inspect the GreenHub dataset itself, as he/she will enrich the data that it makes available using data from other public sources.
The main goals to achieve within this project are:
- to study and critically analyze the state of the art in mining large data repositories;
- to realize how information from multiple datasets can enrich each other;
- to devise and implement a framework for identifying energy saving patterns within Android;
- to propose a catalog of such patterns.
Plano de Trabalhos - Semestre 1
- Trimester 1: Study the literature, the tools and the frameworks that are available for mining large data repositories.
- Trimester 2: Design a framework that, combining the data of the GreenHub dataset with other datasets, can identify a series of patterns that reflect the usage of Android smartphones. Write an intermediate report.
Plano de Trabalhos - Semestre 2
- Trimester 3: Implement the designed framework; propose a catalog of Android energy-usage patterns.
- Trimester 4: Extensively analyze the catalog to provide strategies for improving Android usage patterns in terms of reducing energy consumption. Thesis writing.
Condições
The eligible student will have at disposal all the necessary computational platforms, tools and devices, including the use of Android devices, if agreed necessary.
He/she will work within a research team that is working on related issues.
Orientador
João Paulo Fernandes (jpf@dei.uc.pt) Bruno Cabral (bcabral@dei.uc.pt)
jpf@dei.uc.pt 📩