Disaster Detection by Statistics and SVM for Emergency Rescue Evacuation Support System

The authors have proposed the Emergency Rescue Evacuation Support System (ERESS) for reducing human damage at disasters. The ERESS primarily intended to reduce the number of victims in panic-type disasters (e. g., fire, terrorism) and works under mobile ad-hoc networks (MANET). This system uses ERESS Mobile Terminals (EMTs) like smart phones and tablets. EMTs have an advanced disaster detection algorithm and sensors. They get data from sensors such as acceleration, angular velocity, and terrestrial magnetism. EMTs classify behavior of EMTs holders like walking and running, and detect disasters from escape actions. In this paper, the authors propose a new effective disasters detection method using statistics and a support vector machine (SVM). In this method, we introduce weight coefficient for each EMT holder according to his behavior statistics. Disasters simulation experimental results show the effectiveness of the proposed method.