Step Detection Enhanced by Anomaly Filtering
Sayyaf MI, Zhu N, Renaudin V, Feigl T (2025)
Publication Language: English
Publication Type: Conference contribution, Original article
Publication year: 2025
Pages Range: 1-4
Conference Proceedings Title: Proc. Intl. IEEE Applied Sensing Conference (APSCON)
Event location: Hyderabad, India
Abstract
Here’s a corrected and refined version of the abstract:
---
This paper presents a novel anomaly detection algorithm designed to enhance the accuracy of step detection by eliminating false positives caused by signals that mimic walking. Such signals include hand movements, vibrations, phone drops, rapid movements, and actions like standing up or sitting down, which can compromise reliability. The challenge lies in differentiating true walking signals from a wide variety of non-walking movements, highlighting the need for robust anomaly detection methods capable of distinguishing walking from other activities with minimal uncertainty.
The proposed algorithm achieves a high detection rate of **95.2% ± 0.7%** for normal walking across a 40 km dataset collected from eight participants using a smartphone in various modes, including texting, pocket, and swing. It effectively identifies anomalous movements, attaining a detection rate of **82.5% ± 1.25%**, thereby filtering out false signals and improving overall reliability. Additionally, the algorithm is optimized for mobile devices, with computational and energy efficiency demonstrated through metrics such as floating point operations per second (FLOPs), CPU processing time, and energy consumption. This ensures its practicality for real-world, resource-constrained applications.
Authors with CRIS profile
Involved external institutions
France (FR)How to cite
APA:
Sayyaf, M.I., Zhu, N., Renaudin, V., & Feigl, T. (2025). Step Detection Enhanced by Anomaly Filtering. In Proc. Intl. IEEE Applied Sensing Conference (APSCON) (pp. 1-4). Hyderabad, India.
MLA:
Sayyaf, Mohamad Issam, et al. "Step Detection Enhanced by Anomaly Filtering." Proceedings of the 3rd International IEEE Applied Sensing Conference (APSCON), Hyderabad, India 2025. 1-4.
BibTeX: Download