Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Data Augmentation Techniques for Cross-Domain WiFi CSI-based Human Activity Recognition

Julian Strohmayer, Martin Kampel

TL;DR

Data augmentation techniques commonly used in image-based learning are applied to WiFi CSI to investigate their effects on model generalization performance in cross-scenario and cross-system settings, showing that specific combinations of simple data augmentation techniques applied to CSI amplitude data can significantly improve cross-scenario and cross-system generalization.

Abstract

The recognition of human activities based on WiFi Channel State Information (CSI) enables contactless and visual privacy-preserving sensing in indoor environments. However, poor model generalization, due to varying environmental conditions and sensing hardware, is a well-known problem in this space. To address this issue, in this work, data augmentation techniques commonly used in image-based learning are applied to WiFi CSI to investigate their effects on model generalization performance in cross-scenario and cross-system settings. In particular, we focus on the generalization between line-of-sight (LOS) and non-line-of-sight (NLOS) through-wall scenarios, as well as on the generalization between different antenna systems, which remains under-explored. We collect and make publicly available a dataset of CSI amplitude spectrograms of human activities. Utilizing this data, an ablation study is conducted in which activity recognition models based on the EfficientNetV2 architecture are trained, allowing us to assess the effects of each augmentation on model generalization performance. The gathered results show that specific combinations of simple data augmentation techniques applied to CSI amplitude data can significantly improve cross-scenario and cross-system generalization.

Data Augmentation Techniques for Cross-Domain WiFi CSI-based Human Activity Recognition

TL;DR

Data augmentation techniques commonly used in image-based learning are applied to WiFi CSI to investigate their effects on model generalization performance in cross-scenario and cross-system settings, showing that specific combinations of simple data augmentation techniques applied to CSI amplitude data can significantly improve cross-scenario and cross-system generalization.

Abstract

The recognition of human activities based on WiFi Channel State Information (CSI) enables contactless and visual privacy-preserving sensing in indoor environments. However, poor model generalization, due to varying environmental conditions and sensing hardware, is a well-known problem in this space. To address this issue, in this work, data augmentation techniques commonly used in image-based learning are applied to WiFi CSI to investigate their effects on model generalization performance in cross-scenario and cross-system settings. In particular, we focus on the generalization between line-of-sight (LOS) and non-line-of-sight (NLOS) through-wall scenarios, as well as on the generalization between different antenna systems, which remains under-explored. We collect and make publicly available a dataset of CSI amplitude spectrograms of human activities. Utilizing this data, an ablation study is conducted in which activity recognition models based on the EfficientNetV2 architecture are trained, allowing us to assess the effects of each augmentation on model generalization performance. The gathered results show that specific combinations of simple data augmentation techniques applied to CSI amplitude data can significantly improve cross-scenario and cross-system generalization.
Paper Structure (13 sections, 4 figures, 6 tables)

This paper contains 13 sections, 4 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Experimental Setup
  4. Hardware
  5. Environment
  6. Data
  7. Augmentations
  8. Model Training
  9. Ablation Study
  10. Cross-scenario Generalization
  11. Cross-system Generalization
  12. DISCUSSION
  13. CONCLUSION

Figures (4)

  • Figure 1: CSI spectrograms of a person walking and a person walking while simultaneously waving their arms, showing the characteristic patterns in WiFi subcarrier amplitudes caused by these activities.
  • Figure 2: Systems overview, showing (a) the PIFA with a plane reflector system, and (b) the BQ antenna system.
  • Figure 3: Floor plan of the test environment, showing the transmitter and receiver placement in LOS and NLOS scenarios.
  • Figure 4: (a) LOS and (b) NLOS CSI amplitude spectrograms of the classes no presence, walking, and walking + arm-waving, captured with BQ antenna and PIFA with plane reflector systems at a distance of 9.4m (room 3 in the NLOS scenario). The spectrograms show the amplitudes of 52 L-LTF subcarriers over a time interval of 4 seconds ($\sim$400 packets).