Physical Activity Sensing via Stand-Alone WiFi Device

Abstract

WiFi signals for physical activity sensing shows promising potential for many healthcare applications due to its potential for recognising various everyday activities, non-invasive nature and low intrusion on privacy. Traditionally, WiFi-based sensing uses the Channel State Information (CSI) from an offthe-shelf WiFi Access Point (AP) which transmits signals that have high pulse repetition frequencies. However, when there are no users on the communication network only beacon signals are transmitted from the WiFi AP which significantly deteriorates the sensitivity and specificity of such systems. Surprisingly WiFi based sensing under these conditions have received little attention given that WiFi APs are frequently in idle state. This paper presents a practical system based on passive radar technique which does not require any special setup or preset firmware and able to work with any commercial WiFi device. To cope with the low density of beacon signal, a modified Cross Ambiguity Function (CAF) has been proposed to reduce redundant samples in the recorded. In addition, an external device has been developed to send WiFi probe request signals and stimulate an idle AP to transmit WiFi probe responses thus generate usable transmission signals for sensing applications without the need to authenticate and join the network. Experimental results prove that proposed concept can significantly improve activity detection and is an ideal candidate for future healthcare and security applications.

Using RF Transmissions From IoT Devices for Occupancy Detection and Activity Recognition

“…• WiFi router provides a gateway to the internet for other devices. WiFi router constantly broadcasts beacon signals to announce its presence which is normally fixed at 10Hz [22]. It can achieve high frame rates when there are is internet usage [12], [34].…”

“…In fine-grained sensing, tasks like activity recognition and event classification, require advanced signal processing and classifiers to achieve sufficient performance. The signal processing for generating the Doppler spectrogram has been designed with a low-complexity design [22], which can be processed in real-time within LabVIEW. Classifiers due to their high computational complexity are implemented off-line processing.…”

“…In work [19], authors demonstrate the feasibility of tracking drones, whilst work [20] counts the pedestrians in urban traffic scenarios. PWR systems have also been used in human presence detection [21] and activity recognition [22]. However, these PWR systems cannot be directly applied to IoT networks as the mechanisms of signal broadcast from IoT devices are rather different from that of a WiFi router.…”

Design of high‐speed software defined radar with GPU accelerator

“…Cross-correlation evaluates the level of similarity between two functions or signals [20], and is widely used to detect the time delay and Doppler shift for a known transmitted signal and reflected signal from objects. More specifically, according to our previous works [15,21], cross-correlation consumes the major computational power in signal processing. In such scenarios, the speed of the cross-correlation is crucial to the overall performance of the system.…”

A High-Speed Multi-Purpose Software Defined Radar for Near-Field Applications

“…The work [9] further explores SDR-based UWB radar for landmine detection. Our previous work [10], [11] presented a Passive WiFi Radar (PWR) that used the WiFi AP as the signal source to illuminate the enviornment. The system was designed for real-time operation, however it was limited by single surveillance channel and relatively low sampling rate.…”