A Scalable Wearable Wireless Sensor Network for Human Activity Detection

View at publisher Summarise with VIRAT
Authors: S. H. Gunawardena, P. M. Amarasinghe, S. D. H. C. Gunawardhana, S. B. N. M. Sabaragamuwa
Cited by: 0
References: 15

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Human activity recognition is a crucial component in athlete performance monitoring and healthcare applications. This work introduces a scalable Wearable Wireless Sensor Network (WWSN) relying on tri-axial accelerometers and biometric sensors. The scalability and device synchronization are supported by the modular primary-secondary design and communication protocol. Data is collected from 41 subjects across six different activities by placing the sensor modules on wrists, legs, and torso, in order to identify the most significant sensor positions. Machine learning models such as Random Forest and K-Nearest Neighbors achieve over 99% accuracy using all five sensors, while data from the wrist sensors resulted in over 97% accuracy, showcasing their significance in human activity recognition. Further, Convolutional Neural Network showed marginally better performance by using a sequence of instances, rather than a single instance, as the input. Findings of this work emphasize that the paired wrist sensors can effectively classify the target activities, reducing both burden and cost. This WWSN provides a platform for the HAR research and athlete performance monitoring.

Fields of Study & Subjects

Context-Aware Activity Recognition Systems Wireless Sensor Networks for Data Analysis Wireless Body Area Networks

Publication details

Journal / source
IEEE International Conference on Industrial and Information Systems (ICIIS)
Pages
298–303
Publication date
16 January 2026
Language
EN