Article
A Domain-Independent Generative Adversarial Network for
Activity Recognition Using WiFi CSI Data
Augustinas Zinys *, Bram van Berlo and Nirvana Meratnia
Citation: Zinys, A.; van Berlo, B.;
Meratnia, N. A Domain-Independent
Generative Adversarial Network for
Activity Recognition Using WiFi CSI
Data. Sensors 2021, 21, 7852. https://
doi.org/10.3390/s21237852
Academic Editor: Alexander Wong
Received: 1 October 2021
Accepted: 16 November 2021
Published: 25 November 2021
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2021 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Interconnected Resource-Aware Intelligent Systems Cluster, Department of Mathematics and Computer Science,
Eindhoven University of Technology, 5600 MB Eindhoven, The Netherlands; b.r.d.v.berlo@tue.nl (B.v.B.);
n.meratnia@tue.nl (N.M.)
* Correspondence: augustinaszin@gmail.com
Abstract:
Over the past years, device-free sensing has received considerable attention due to its
unobtrusiveness. In this regard, context recognition using WiFi Channel State Information (CSI) data
has gained popularity, and various techniques have been proposed that combine unobtrusive sensing
and deep learning to accurately detect various contexts ranging from human activities to gestures.
However, research has shown that the performance of these techniques significantly degrades due to
change in various factors including sensing environment, data collection configuration, diversity of
target subjects, and target learning task (e.g., activities, gestures, emotions, vital signs). This problem,
generally known as the domain change problem, is typically addressed by collecting more data and
learning the data distribution that covers multiple factors impacting the performance. However,
activity recognition data collection is a very labor-intensive and time consuming task, and there
are too many known and unknown factors impacting WiFi CSI signals. In this paper, we propose
a domain-independent generative adversarial network for WiFi CSI based activity recognition in
combination with a simplified data pre-processing module. Our evaluation results show superiority
of our proposed approach compared to the state of the art in terms of increased robustness against
domain change, higher accuracy of activity recognition, and reduced model complexity.
Keywords:
device-free sensing; unobtrusive sensing; WiFi CSI; generative adversarial network;
domain change; domain adaptation
1. Introduction
With the accelerating development of new sensing and communication technologies,
monitoring human activities in everyday life has become more popular than ever in various
fields such as surveillance, entertainment, and healthcare. Sensing technologies in the
field of human context (e.g., activities, gestures, emotions, vital signs) recognition can be
categorised into two sub-categories: device-based and device-free. While device-based
sensing refers to a situation in which sensors are attached to the human body to measure
and monitor a specific context, device-free sensing refers to situations in which not the
human body, but the environment in which a human is present in is monitored.
Although many device-based sensing systems have become quite popular, in some
situations it is impractical and cumbersome to wear them all the time. In order to overcome
the limitations of device-based sensing, device-free sensing, such as visual-based sensing
(cameras), has been considered. Although this technology is quite popular, with computer
vision algorithms (object detection/recognition) advancing rapidly, it only operates in
scenarios in which a subject is in line-of sight and no occluding obstacles are in the view.
Additionally, it requires robust and continuous lighting conditions, as a subject may
not be visible throughout the entire day. Moreover, visual-based sensing devices are
intrusive as they impact the privacy of an individual. Therefore, in order to overcome all of
these limitations, device-free solutions, using radio signals such as WiFi, are considered.
In particular, the IEEE 802.11 protocol contains channel state information (CSI), which
Sensors 2021, 21, 7852. https://doi.org/10.3390/s21237852 https://www.mdpi.com/journal/sensors
