The rapid growth of mobile communication and pervasive computing technology has expanded the demand for location-based services in urban areas, which necessitate accurate localization and tracking of the user. Due to very poor performances of global positioning system (GPS) in indoor areas and urban canyons, various radio frequency (RF) technologies have been exploited by many researchers to design indoor positioning systems (IPSs) over the years. However, such RF based IPSs would work only in the presence of certain infrastructure. The emergence of low-cost inertial sensors like accelerometer, gyroscope etc., on the other hand, has motivated the researchers to design pedestrian dead reckoning (PDR) based IPSs as they do not depend on availability of any sort of infrastructure in the surrounding environment. Step length estimation (SLE) is an integral part of PDR based localization. We, thus, aim to design a SLE technique in this paper, by processing signal measurements collected from the inertial sensors attached to some body parts.
Our proposed SLE method requires only a simple foot-mounted inertial sensor unit integrating three-axis accelerometer and gyroscope and it uses Butterworth low-pass filter to remove the noises from both acceleration and angular velocity signals. A normalized threshold and minimum time constraint-based peak-valley detection algorithm is also employed by our proposed method to calculate the swing phase duration. The proposed SLE technique uses only one user-specific parameter which is also calibrated for five different users in this paper. The performances of the proposed SLE technique are then evaluated and compared with some existing method in terms of accuracy to demonstrate the efficacy of the proposed method. Moreover, the relationships between the gait features and step length of the pedestrians as well as the same between their body mass index and the calibrated parameter are also examined in this paper.