A deep neural network-based classifier using an IR-UWB sensor achieved the best results for joint vital signs and occupancy detection compared to other machine learning algorithms.
A deep neural network-based classifier using an IR-UWB sensor effectively performs joint vital signs and occupancy detection without requiring manual feature extraction.
This article proposes the machine learning (ML)-based joint vital signs (VSs) and occupancy detection (OD) with an impulse radio ultra-wideband (IR-UWB) sensor. Works that have been done on VS or OD development using an IR-UWB are related to how VS works. In the related experiments performed, the OD and state of individuals were not sufficiently verified, and the methods were computationally complex. Issues related to the use of ML for joint VS and OD (VSOD) have also not been studied in the literature. Extensive experimental scenarios involving the application of an ML-based classifier for human OD and VS classification, which we extended toward three sub-scenarios, were evaluated. We formulated a solution for VS estimation, which was aligned, so that each network input sequence received signal corresponding to respective VS over different scenarios. The performance of the proposal was evaluated with other competing ML-based classification algorithms. Compared with other techniques, our proposed deep neural network (DNN)-based classifier achieved the best results, and it also offers benefits over other algorithms, such as not needing to extract features from the data.
Numan et al. (Tue,) reported a other. Deep neural network (DNN)-based classifier with IR-UWB sensor vs. Other competing ML-based classification algorithms was evaluated on Human occupancy detection and vital signs classification. A deep neural network-based classifier using an IR-UWB sensor achieved the best results for joint vital signs and occupancy detection compared to other machine learning algorithms.