Abstract
There are many established methods to collect traffic data used for vehicle counting, speed checking and vehicle classification, such as induction loops, piezoelectric sensor, bending plates, etc. This paper presents a new method for traffic data collection. It uses vibration signals which are generated when vehicles pass the roadside. In order to extract useful information, such as the corresponding time when the vehicle was passing the vibration sensor, wavelet analysis was employed for data mining of the vibration signal observed from accelerometers (vibration sensors). The applications of wavelet transform and decomposition for vibration signals were reviewed to identify a suitable approach for the vibration signal data mining. The paper concluded with discussing the impact of wavelet analysis on data mining and outlining potential future research directions and applications.