Deep-Learning-Based Signal Enhancement of Low-Resolution Accelerometer for Fall Detection Systems

In the last two decades, fall detection (FD) systems have been developed as a popular assistive technology. To support long-term FD services, various power-saving strategies implemented. Among them, reduced sampling rate is common approach for an energy-efficient system in real world. However, performance of diminished owing to low-resolution (LR) accelerometer signals. improve accuracy with LR signals, several technical challenges must be considered, including mismatch effective features and degradation effects. this work, deep-learning-based signal enhancement (ASE) model proposed front-end processor help typical LR-FD achieve better performance. The ASE based on deep denoising convolutional autoencoder architecture reconstructs high-resolution (HR) signals from by learning relationship between HR results show that using vector machine (SVM) at extremely low (sampling < 2 Hz) achieved 97.34% 90.52% accuracies SisFall FallAllD data sets, respectively, while those without models only 95.92% 87.47% respectively. also demonstrate mode can suitably combined systems.