JY Tigli Homepages

• CHIARAMELLO
• Julien
• SI5 / CASPAR
• chiarame@polytech.unice.fr

• STIRBU
• Cristina
• Master IFI/IAM
• mariacristina.stirbu@gmail.com

• Model : HTC 8S
• It is the school's one
• IMEI : 358 721 050 415 149

This project is about tracking battery usage in phones running WP8. For that, we use the Battery class, in the .Net namespace “Windows.Phone.Devices.Power”. This class fires an event whenever the battery percentage changes. Depending on the direction of the change, we know if the phone is discharging or charging, and by using the time of changes we can calculate the (dis)charging rate of the phone.

From that data, we can then establish profiles or “clusters” about both battery usage stats and charging habits of the user (e.g. the battery drops faster every day from 5 PM to 6 PM indicating heavy use and the phone charges from 11 PM to 7 AM every day)

• README File

The readme file is located at the root of the archive. It is named “readme.txt”.

• Project zip file

The project can be downloaded here. Simply click the grey “Download” button, enter the captcha and click “Download Now”.

Please note that filedropper has an expiry period of 30 days since last download. Should the file expire, please contact us so we can re-upload it quickly.

• All required software

Other than Windows 8, Visual Studio 2013 SP2 and the SDK for WP8.0+, no additional software is required.

The program behavior is simple. In order to use it, you only need to launch the application from your phone. Once it is on, simply wait for the phone to charge or discharge.

Whenever the battery level changes, the app will automatically record it. After a while, you can press the “Show clusters” button to see the cluster ID followed by whether the battery is charging or discharging and the time of the change.

Should you wish to test the clusters, you can press the “Add (dis)charge” buttons to simulate a charging or discharging event.

The application is working and successfully records the battery level changes. The application is currently using 2 clusters, but the algorithm works for more of them.

One problem of using the battery as a sensor is that unlike other types of sensors such as accelerometers or GPS, the battery charging or discharging cannot be directly controlled, merely quickened with background apps (or lack thereof). Points for the clusters are also pretty scarce, typically one every few minutes, which means that in order to have meaningful data, we need to have many points, which can take a really long time. This in turn means that we cannot make tests with many cluster quantities.