An Adaptive Step Detection Method for Smartphones Based on Time-Dependent Decay Mechanism

As an indoor positioning method, pedestrian dead reckoning (PDR) is crucial for positioning and navigation in environments where satellite signals are blocked, such as shopping malls, hospitals, and tunnels. The performance of PDR is mainly influenced by step detection, step length estimation, and heading angle calculation. The accuracy and real-time performance of step detection play a crucial role in achieving high-precision indoor positioning. Most of the current step counting methods for smartphones, however, suffer from time delays. Meanwhile, the location of smartphones and pedestrian movement patterns have a significant impact on step counting accuracy. We, therefore, propose an adaptive step detection method based on a time-dependent decay mechanism to overcome the time delays and the influences of smartphone locations and pedestrian movement patterns. The proposed method first preprocesses the acceleration data of a smartphone and identifies its stationary state using a decision tree. Second, the first two peaks are identified based on the number and magnitude of acceleration increases. Third, the adaptive peak threshold and time difference threshold at the current time are calculated in real time based on the time-dependent decay mechanism to determine whether to count a step. Finally, the count of steps is corrected according to the pedestrian's end state to achieve more accurate step counting. Experimental results demonstrate that the proposed method is less affected by smartphone locations and pedestrian movements, achieving a step counting accuracy of 97.4% under complex motion conditions. Furthermore, the method exhibits good real-time performance, meeting the low-latency requirements of indoor positioning based on smartphones.