Author ORCID Identifier

0000-0002-4378-9270

Document Type

Conference Paper

Disciplines

Statistics

Publication Details

Statistical and Machine Learning: Methods and Applications (SAML-25) on June 5th and 6th, 2025.

doi:10.21427/10cc-9a48

Abstract

In recent years, WiFi-based Human Activity Recognition (HAR) has gained substantial attention due to the ubiquity of WiFi infrastructure and advancements in wireless communication. Unlike camera-based systems that raise privacy concerns or wearable sensors that require user compliance, WiFi-based HAR provides a noninvasive and practical alternative that operates seamlessly with existing infrastructure. WiFi-based HAR leverages fluctuations in wireless signals, particularly Channel State Information (CSI), to passively detect and classify human activities. WiFi-based HAR models often achieve high accuracy in a single environment but suffer significant performance drops when applied to new environments due to variations in spatial settings, human movement, and physical factors. This study investigates the environmental and subject-specific factors that influence the reliability and generalization of WiFi-based HAR. We systematically evaluate how room layout, signal interference, transmitter/receiver placement, subject appearance, and human proximity affect signal propagation and activity classification performance. Controlled experiments were conducted using ESP32 devices, collecting CSI data across various activity scenarios in different indoor configurations. A deep learning model combining a Convolutional Neural Network (CNN) with an Attention-based Bidirectional Long Short-Term Memory (BiLSTM) network was trained and evaluated. In initial experiments, we analyzed the effect of room layout changes. Training on one layout and testing on another led to a 7% drop in classification accuracy—from 94% to 87%—demonstrating a measurable impact of room layout on performance. These findings highlight the importance of considering day-to-day environmental changes in the design and deployment of WiFi-based HAR systems. Future work will expand on these results by investigating a broader range of influencing factors, including electronic interference, Tx/Rx positioning, and subject diversity, to improve the generalization and reliability of HAR in practical applications.

DOI

https://doi.org/10.21427/10cc-9a48

Creative Commons License

Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License
This work is licensed under a Creative Commons Attribution-NonCommercial-Share Alike 4.0 International License.

Download
Contact the Author


Share

COinS