Wi-Fi CSI-Based Camera-Free Human Presence and Activity Detection Using Low-Cost ESP32 Devices

This paper presents a literature-based analytical study of Wi-Fi Channel State Information (CSI) for camera-free human presence and activity detection using low-cost ESP32 devices. It reviews commodity CSI extraction, preprocessing, machine-learning approaches, semi-supervised inference, one-sided through-wall sensing, and related privacy and security considerations. Published evidence indicates that CSI-based sensing can achieve strong performance in controlled environments, but results vary substantially across environments due to factors such as room geometry, walls, multipath propagation, device placement, and dataset differences. The study does not claim original physical CSI measurements, a newly collected dataset, or experimental validation. Instead, it synthesizes existing research evidence and proposes a reproduction-oriented architecture for future implementation and cross-environment evaluation.