Earthquake detection using a smartphone

Nowadays, many transactions are achieved with smartphones. Through its holdings in accelerometer sensors it is also used in the various studies. Accelerometer sensor is a sensor that measures the size of a power applied onto a mass. First of all data was gathered with the help of accelerometer sensor on mobile phones. In order to produce more natural results, the smartphones in the study were carried in side pockets of trousers during data gathering. Data was gathered during a variety of activities like running, walking, going upstairs and downstairs, holding still and so on via accelerometer sensor. In order to avoid orientation problem in raw acceleration data, amplitudes of these raw data were used. Pieces of data whose amplitudes were calculated according to the pre-determined slider range were divided into windows with certain length. A fuzzy system has been developed to decide whether the the shake is due to earthquake movements or human movements using AR (Autoregressive) model with Matlab. Output data were chosen in such a way that they would be positive for earthquake data and negative for human activity. Data were divided as training and test data. In this way, training and set data groups of acceleration data gathered from various conditions were formed. During experiments with test data, the system was able to differentiate between all earthquake-based data and human-caused data successfully. Keywords: Accelerometer sensor; Autoregressive model; Smartphone; Earthquake detection