Passive Wi-Fi Radar Fall And Non-Fall Motion Classification Using Hybrid PCA-SVM Classifier

Falls present substantial risks to the health and well-being of the elderly population. To address the limitations of wearable devices, non-contact methods for indoor fall detection have gained popularity. This paper focuses on the utilization of passive Wi-Fi radar as a non-contact sensing technique for detecting and classifying falls in the elderly. The proposed approach combines Principal Component Analysis (PCA) and Support Vector Machine (SVM) to enhance precision and dependability. The study involves analysis of signatures in both the time and frequency domains of fall and non-fall activities, with the aim of identifying distinguishing characteristics. By employing PCA, the dimensionality of the data is effectively reduced. Subsequently, SVM is employed to determine the optimal boundary between fall and non-fall activities, enabling accurate classification. The experimental results demonstrate a high classification accuracy, with a rate of 97.7% achieved for PC=3 and 97% for PC=4. These results underscore the potential of the proposed method for real-time fall detection in elderly individuals. The use of passive Wi-Fi radar, in conjunction with PCA and SVM, presents a promising avenue for mitigating the risks associated with falls among the elderly, thereby contributing to their overall safety and well-being.