TriSense: RFID, radar, and USRP-based hybrid sensing system for enhanced sensing and monitoring

Khan, Muhammad Zakir (2024) TriSense: RFID, radar, and USRP-based hybrid sensing system for enhanced sensing and monitoring. PhD thesis, University of Glasgow.

Full text available as:
[thumbnail of 2024KhanPhD.pdf] PDF
Download (11MB)


This thesis presents a comprehensive approach to contactless human activity recognition (HAR) using the capabilities of three distinct technologies: radio frequency identification (RFID), Radar, and universal software-defined radio peripheral (USRP) for capturing and processing Wi-Fi-based signals. These technologies are then fused to enhance smart healthcare systems. The study initially utilises USRP devices to analyse Wi-Fi channel state information (CSI), choosing this over received signal strength for more accurate activity recognition. It employs a combination of machine learning and a hybrid of deep learning algorithms, such as the super learner and LSTM-CNN, for precise activity localisation. Subsequently, the study progresses to incorporate a transparent RFID tag wall (TRT-Wall) that employs a passive ultra-high frequency (UHF) RFID tag array. This RFID system has proven highly accurate in distinguishing between various activities, including sitting, standing, leaning, falling, and walking in two directions. Its effectiveness and non-intrusiveness make it particularly suited for elderly care, achieved using a modified version of the Transformer model without the use of a decoder. Furthermore, a significant advancement within this study is the creation of a novel fusion (RFiDARFusion) system, which combines RFID and Radar technologies. This system employs a long short-term memory networks variational autoencoder (LSTM-VAE) fusion model, utilising RFID amplitude and Radar RSSI data. This fusion approach significantly improves accuracy in challenging scenarios, such as those involving long-range and non-line-of-sight conditions. The RFiDARFusion system notably improves the detection of complex activities, highlighting its potential to reduce healthcare costs and enhance the quality of life for elderly patients in assisted living facilities. Overall, this thesis highlights the significant potential of radio frequency technologies with artif icial intelligence, along with their combined application, to develop robust, privacy-conscious, and cost-effective solutions for healthcare and assisted living monitoring systems.

Item Type: Thesis (PhD)
Qualification Level: Doctoral
Subjects: T Technology > T Technology (General)
Colleges/Schools: College of Science and Engineering > School of Engineering > Electronics and Nanoscale Engineering
Supervisor's Name: Abbasi, Professor Qammer H.
Date of Award: 2024
Depositing User: Theses Team
Unique ID: glathesis:2024-84222
Copyright: Copyright of this thesis is held by the author.
Date Deposited: 10 Apr 2024 13:55
Last Modified: 10 Apr 2024 13:56
Thesis DOI: 10.5525/gla.thesis.84222
Related URLs:

Actions (login required)

View Item View Item


Downloads per month over past year