Classification of abnormal behavior in the elderly based on Wi-Fi

Abnormal behavior in daily life is a great threat to the health of the elderly, so it is of great significance to detect the abnormal behavior of the elderly. Using ubiquitous commercial Wi-Fi devices to collect information, behavior detection can be performed in an indoor environment without carrying other devices. The Wi-Fi based abnormal behavior detection for the elderly proposed in this paper is aimed at the fall and wandering behaviors that the elderly are prone to. At the same time, squatting and sitting behaviors those are similar to the fall behaviors are distinguished. Several classical classification models are used to compare the effect and recognition time. The experimental comparison shows that the SVM and KNN machine learning methods after adjusting the model parameters can achieve the same recognition effect as the deep learning model LSTM, shorten the running time and improve the efficiency.