An accurate step detection algorithm using unconstrained smartphones

In recent years, mobile device (e.g., smartphone, tablet and etc.) equipped with various inertial sensors is increasingly popular in daily life, and a large number of mobile applications have been developed based on such built-in inertial sensors. In particular, detecting and counting steps is a prerequisite for many applications, such as smart healthcare, smart home, tracking and location, and etc., and thus has attained much attention. Peak detection is known to be one of the simplest and most efficient solutions in this field, but suffers from the drift in the orientation and position of the device if it is not tightly fixed on the human's body. In this paper, we present a novel method to accurately detect and count steps of a human who carries on a smartphone in an unconstrained manner. To be specific, the proposed method fuses the signals from the accelerometer, magnetometer and gyroscope of the smartphone to transform the device reference frame to the earth reference frame, and then employs the vertical acceleration to implement the peak detection algorithm. Extensive simulations are carried out and confirm that the proposed method is more robust than the existing algorithms.