Demonstrating device-free localization based on radio tomographic imaging through simulations and experimentation

Abstract

Location Based Services (LBSs) have led to an increase in demand for Device-Free Localization (DFL) in the indoor environment. Radio Tomographic Imaging (RTI) is one well-known localization algorithm that can estimate the position of a physical object on the basis of the received signal strength indicator (RSSI) changes in the environment that are acquired by Wi-Fi or Wireless Sensor Networks (WSNs). This thesis makes two significant contributions. The first contribution involves the demonstration of the RTI approach by utilizing full electromagnetic simulations of an indoor environment. In particular, we considered dielectric cylinders with varying permittivities but fixed radius in which 20 Wi-Fi nodes are deployed on the perimeter (3 m by 3 m) of a region to perform RTI. The RTI approach is based on the line-of-sight (LOS) weighting model with total variation-based regularization which is used to reconstruct and localize the position of the cylinder at high resolution. The second contribution of the thesis is to utilize the knowledge and fundamentals from the simulations to develop an experimental setup by using inexpensive Wi-Fi hardware modules with directional antenna. We performed experiments on the Wi-Fi hardware modules itself by comparing the mean RSSI of two Wi-Fi hardware modules with the two-ray ground-reflection model at incremental distances in order to explore the impact of multipath fading from the ground and its reflection coefficient. We also introduced a physical object between two Wi-Fi hardware modules to observe the interference caused on the Wi-Fi signal and changes in the RSSI measurements. Finally, we developed a domain of interest identical to the simulations and placed physical objects in the region at different coordinates in order to reconstruct its image and localize its position. Through both the simulations and experimentation, the RTI methodology is suitable for reconstructing and localizing objects with high permittivity such as humans in the domain of interest. However, RTI tends to face issues with reconstructing images and localizing physical objects with lower permittivities thus motivating further future investigations.

Date of Award	2020
Original language	English
Awarding Institution	The Hong Kong University of Science and Technology