Empirical Evaluation and Analysis of Real-Time Detection and Localization of Unexpected Events

With the spread of smartphones, making use of crowd-sourcing approaches in everyday life became possible. In one study using smartphones, the authors previously proposed a crowd-sourced system called the Smart and Quick Unexpected-Event Detect (SQUED) system that used smartphone sensor data for the detection and localization of unexpected events. SQUED can detect the location and the time of an event using smartphone sensor data. In this paper, the authors describe our improved system named the Real-Time SQUED (RT-SQUED) system, which is based on our original SQUED system, to provide realtime detection and localization functionality. The authors also describe our empirical evaluation and analysis results on our realtime system to compensate for the lack of experimental data so far. It is shown that our real-time system can detect events faster under various conditions using a large amount of real-world data. In this paper, the usefulness of, and improvement points for the RT-SQUED system are also discussed.