This study presents a robust, privacy-preserving method for human location estimation in bathrooms using 25-kHz spatial ultrasound and the Doppler effect, without relying on machine learning. The system consists of a ceiling-mounted speaker and a 16-channel Fermat-spiral microphone array that transmits DSSS-modulated ultrasound signals. Received signals are processed through bandpass filtering, cross-correlation, beamforming, and FFT to track Doppler-induced phase shifts caused by human movement. The floor is divided into spatial zones, and temporal phase variations are analyzed to estimate a person's location every 0.5 seconds. Experimental results demonstrate an average localization accuracy of 95.2% when the subject performs slight periodic torso movement. The method requires no training data, offering strong robustness to environmental variation and high potential for applications such as posture recognition, fall detection, and drowning prevention in real-world bathroom environments.
