STEP COUNTING USING SMARTPHONE ACCELEROMETER AND FAST FOURIER TRANSFORM

Abstract

This paper suggests an easy to implement and battery friendly method to Step Counting problem, which is based on Android smartphone accelerometer, Fast Fourier Transform (FFT) and thresholding. The method, after applied preprocessing and FFT to the data gathered from smartphone's accelerometer (16hz), detects and counts steps by comparing the data to predefined thresholds. The predefined thresholds were specified by analyzing the accelerometer data obtained while users were sitting, standing, and walking. User experiments were conducted to test the method. Results showed 87.52% success for walking. Additionally, for sitting and standing the method had a success over 99%, as expected. And for running, only 41.7% of the steps were correctly counted.