Fall Detection Based on Millimeter-Wave Radar Feature Extraction Using LSTM

Radar-based fall detection technology demonstrates significant application value in healthcare, especially for ensuring the safety of elderly individuals living alone at home. To improve the accuracy and real-time performance of fall detection, this paper proposes a long short-term memory network (LSTM) based fall detection method that extracts features from radar signals. Specifically, this method utilizes pulse compression and Doppler fast Fourier transform (FFT) to extract three features from radar echo signals: velocity spectrum bandwidth, velocity spectrum area, and direction of movement. These three features of millimeter-wave signals are then used for fall detection through a convolutional neural network in deep learning. This approach achieves efficient and accurate detection of fall events. Experimental results validate the effectiveness of the proposed method, providing a new technical means for ensuring the safety of elderly individuals living alone at home.