Classifying smartphone screen ON/OFF state based on wifi probe patterns
Résumé
WiFi enabled smartphones regularly send probe request messages to actively discover nearby access points. Previously, these probes have been exploited to understand users' mobility patterns and were also identified to be a privacy threat. To highlight the increased threats to user's privacy, we aim to identify the user behavior by characterizing the changes in probe patterns, which occur as an effect of user's smartphone usage. In this article, we developed a browser based interactive front-end client, with a Node JS backend server, to interact with the sniffing hardware and to display realtime packets visualization with smartphone screen state predictions. We were able to detect the screen ON and OFF states by analyzing the probe patterns using a Decision-Tree classifier with accuracy ranging from 93% to 100% on specific smartphone models.