Efficient Real-Time Devices Based on Accelerometer UsingMachine Learning for HAR on Low-PerformanceMicrocontrollers

作     者：Vi M.-T. Tran D.-N. Thuong V.T. Linh N.N. Tran D.-T. 

作者机构：Faculty of Electrical and Electronic Engineering Phenikaa University Hanoi City 100000 Viet Nam Institute of Information Technology Vietnam Academy of Science and Technology Hanoi City 100000 Viet Nam Faculty of Information Technology and Communication Phuong Dong University Hanoi City 100000 Viet Nam Graduate University of Sciences and Technology Vietnam Academy of Science and Technology Hanoi City 100000 Viet Nam International School-Vietnam National University Hanoi City 100000 Viet Nam 

出 版 物：《Computers, Materials and Continua》 (计算机、材料和连续体（英文）)

年 卷 期：2024年第81卷第1期

页      面：1729-1756页

核心收录：

学科分类：08[工学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：National Foundation for Science and Technology Development  NAFOSTED  (02/2022/TN) 

主　　题：Accelerometer activity recognition classification HAR wearable computing 

摘      要：Analyzing physical activities through wearable devices is a promising research area for improving health assessment. This research focuses on the development of an affordable and real-time Human Activity Recognition (HAR) system designed to operate on low-performance microcontrollers. The system utilizes data from a bodyworn accelerometer to recognize and classify human activities, providing a cost-effective, easy-to-use, and highly accurate solution. A key challenge addressed in this study is the execution of efficient motion recognition within a resource-constrained environment. The system employs a Random Forest (RF) classifier, which outperforms Gradient Boosting Decision Trees (GBDT), Support VectorMachines (SVM), and K-Nearest Neighbors (KNN) in terms of accuracy and computational efficiency. The proposed featuresAverage absolute deviation (AAD), Standard deviation (STD), Interquartile range (IQR), Range, and Root mean square (RMS). The research has conducted numerous experiments and comparisons to establish optimal parameters for ensuring system effectiveness, including setting a sampling frequency of 50 Hz and selecting an 8-s window size with a 40% overlap between *** conducted on both theWISDMpublic dataset and a self-collected dataset, focusing on five fundamental daily activities: Standing, Sitting, Jogging,Walking, andWalking the stairs. The results demonstrated high recognition accuracy, with the system achieving 96.7% on the WISDM dataset and 97.13% on the collected dataset. This research confirms the feasibility of deployingHAR systems on low-performance microcontrollers and highlights the system s potential applications in patient support, rehabilitation, and elderly care. © 2024 The Authors. Published by Tech Science Press.