An autonomic indoor positioning application based on smartphone

Abstract

Nowadays positioning and navigation technologies based on smartphone are sprouting up for numerous application scenarios. In this paper a more self-contained approach is introduced by which merely inertial units within the smartphone are utilized. By the Pedestrian Dead Reckoning technique, all kinds of indoor location information are provided at users' disposal. With the gyroscope, the attitude of smartphone is measured. So the real time accelerations in standard coordinate system without gravity component can be calculated. Here only vertical acceleration signals are made use of to extract the features for steps counting as well as step lengths estimation. A series of algorithms are employed to eliminate the noise and deviation, such as Zero Velocity Compensation, Moving Average Filter, Kalman Filter, and Successive Peaks Merging. Particularly the whole walking process is divided into small segments in each of which only straight walking, no stop, no turn is contained. So, different segments are processed respectively with distinctive parameters. The breakpoints are determined by moving variance analysis for accelerations and rotation angles, after which the heading and length of every step are acquired so that the mileage and position can be updated, closely followed by moving trajectory. In experiments, the average deviation of our approach is 0.48 m.