Evaluating BiLSTM and CNN+GRU Approaches for Human Activity Recognition Using WiFi CSI Data

TL;DR

This study compares BiLSTM and CNN+GRU deep learning models for WiFi-based human activity recognition, highlighting how dataset characteristics influence model performance and demonstrating real-world applications.

Contribution

It provides a comparative analysis of BiLSTM and CNN+GRU models on two WiFi CSI datasets, emphasizing the impact of dataset features and preprocessing on accuracy.

Findings

CNN+GRU achieves 95.20% accuracy on UT-HAR.

BiLSTM achieves 92.05% accuracy on NTU-Fi HAR.

Model performance is heavily influenced by dataset characteristics.

Abstract

This paper compares the performance of BiLSTM and CNN+GRU deep learning models for Human Activity Recognition (HAR) on two WiFi-based Channel State Information (CSI) datasets: UT-HAR and NTU-Fi HAR. The findings indicate that the CNN+GRU model has a higher accuracy on the UT-HAR dataset (95.20%) thanks to its ability to extract spatial features. In contrast, the BiLSTM model performs better on the high-resolution NTU-Fi HAR dataset (92.05%) by extracting long-term temporal dependencies more effectively. The findings strongly emphasize the critical role of dataset characteristics and preprocessing techniques in model performance improvement. We also show the real-world applicability of such models in applications like healthcare and intelligent home systems, highlighting their potential for unobtrusive activity recognition.