Indoor Radio Tomographic Imaging with Sparse Device Topologies

Abstract

Abstract Device-Free Localization (DFL) using Wi-Fi signals has emerged as a compelling solution for indoor tracking without requiring individuals to carry any device. This thesis investigates the use of Channel State Information (CSI) for Radio Tomographic Imaging (RTI) in sparse network deployments. By developing a custom round-robin protocol over ESP-NOW and a lightweight data processing pipeline, it was shown that e!ective localization can be achieved with as few as four ESP32 nodes. CSI-based models significantly outperformed traditional Received Signal Strength Indicator (RSSI) methods, particularly in non-line-of-sight (NLOS) and multipath-rich environments. Feature importance analysis highlighted the stability and informativeness of CSI amplitude features for localization tasks. Although limitations such as environmental sensitivity and payload constraints were identified, the study proposes future improvements including phase calibration, domain adaptation, protocol enhancements, and the use of advanced machine learning models. This research contributes to the development of practical, low-cost, and scalable DFL systems suitable for indoor environments.