Time-Selective RNN for Device-Free Multiroom Human Presence Detection Using WiFi CSI

Device-free human presence detection is a crucial technology for various applications, including home automation, security, and health care. While camera-based systems have traditionally been used for this purpose, they raise privacy concerns. To address this issue, recent research has explored the use of wireless channel state information (CSI) extracted from commercial WiFi access points (APs) to provide detailed channel characteristics. In this article, we propose a device-free human presence detection system for multiroom scenarios using a time-selective conditional dual feature extract recurrent network (TCD-FERN). Our system is designed to capture significant time features on current human features using a dynamic and static (DaS) data preprocessing technique. We extract both moving and spatial features of people and differentiate between line-of-sight (LoS) and non-line-of-sight (NLoS) cases. Subcarrier fusion is carried out to provide more objective variation of each sample while reducing the computational complexity. A voting scheme is further adopted to mitigate the feature attenuation problem caused by room partitions, with around 3% improvement of human presence detection accuracy. Experimental results have revealed the significant improvement of leveraging subcarrier fusion, dual-feature recurrent network, time selection, and condition mechanisms. Compared to the existing works in open literature, our proposed TCD-FERN system can achieve above 97% of human presence detection accuracy for multiroom scenarios with the adoption of fewer WiFi APs.

I. Introduction

Smart homes have become increasingly popular due to the pursuit of convenience and innovation. Home security is a crucial issue that requires attention, including intruder alerts, anti-theft, and crisis alarms. To ensure home safety, the stability and precision of the monitoring system are indispensable. Traditionally, wearable sensors are popular for positioning due to their high accuracy. However, these approaches require users to attach sensors to themselves [1], [2], which can be inconvenient and bothersome in some situations. Moreover, wearable devices may require frequent battery recharging [3]. Instead of using device-based methods, research on device-free [4] schemes has increased, such as camera-based detection, infrared, or radar sensors. Camera-based systems [5], [6] provide particular information about human actions, height, or even facial characteristics, but they have privacy defects and blind spot problems. Infrared and radar sensor approaches overcome the privacy issue of camera-based systems [7], but they require humans to walk in a line-of-sight (LoS) path, which cannot detect their presence when staying in a non-line-of-sight (NLoS) path. Therefore, enhancing signal processing techniques in the physical layer is necessary to improve the attainable device-free detection precision at reduced computational complexity [8], [9], [10].