Analysis of 1/f fluctuation in walking using gyro sensor system

Abstract

Presents a useful walking analyser system (WAS). The set-up for acquiring data is a small wearable personal computer (WPC) assisted system. It employs a lithium battery-powered gyro-sensor. While walking, the WPC is worn like a wrist-watch, the small sensor is fixed to the person's back, near the body's centre of gravity, and the batteries are inserted into a waist pouch. When the WPC switch is turned on, three relative angles (i.e. roll, pitch and yaw) and accelerations in the X, Y and Z axes of the person's back are recorded at 30 Hz. People who wear dentures were selected as subjects at the department of geriatric dentistry in a dental hospital. It was observed that, in the case of people wearing dentures, the angular displacements and the accelerations of the back in walking had a better rhythm compared to those without dentures. As an effective analyser system for walking stability, power spectral analysis is used, utilizing autoregressive modelling. In the case of people not wearing dentures, their power spectrum densities of the three angles and accelerations of back fluctuation were wavy, showing unstable walking. In the case of people wearing dentures, their power spectrum densities were approximately close to a 1/f/sup 3/ fluctuation (f = frequency), i.e. they were stable when walking in all directions. Using this WAS, it is easy to discover people's walking stability in rehabilitation and physical fitness.