Wi-Fi signal-based human action acknowledgement using channel state information with CNN-LSTM: A device less approach
| dc.authorid | 0000-0003-1840-9958 | en_US |
| dc.contributor.author | Kumar, V. Dhilip | |
| dc.contributor.author | Rajesh, P. | |
| dc.contributor.author | Polat, Kemal | |
| dc.contributor.author | Alenezi, Fayadh | |
| dc.contributor.author | Althubiti, Sara A. | |
| dc.date.accessioned | 2023-09-06T13:30:32Z | |
| dc.date.available | 2023-09-06T13:30:32Z | |
| dc.date.issued | 2022 | en_US |
| dc.department | BAİBÜ, Mühendislik Fakültesi, Elektrik Elektronik Mühendisliği Bölümü | en_US |
| dc.description.abstract | Human action acknowledgment is an abundant and significant area for machine learning-based researchers due to the level of accuracy in identifying human actions. Due to the rapid growth of technologies in the machine and deep learning techniques, wireless sensors, handy Internet of Things (IoT) devices, and Wireless Fidelity (Wi-Fi), the activity recognition process is made effective with higher accuracy. By using those booming technologies and preserving the privacy of the test person we propose a novel human action recognition model that uses the channel state information (CSI) from Wi-Fi and the most prominent machine learning model, CNN with LSTM. Initially, CSI is introduced, the changes in CSI signals are assessed, and the obtained data samples are made as input to the CNN-LSTM model. To make the recognition more accurate, we also incorporated Kalman filters for noise removal and smoothed the data sample. Furthermore, we have used an image segmentation procedure to identify the initial and end times of all the activities considered and to fragment the image obtained, which is further fed as input to the CNN-LSTM model. Getting a dataset for the experiment is a herculean task. Hence a self-collected dataset is used to assess, or model proposed. Finally, the results obtained are verified and validated for their correctness with appropriate machine learning metrics and parameters like accuracy, F1 score, etc. Our proposed model affords the accuracy of 98.96% for all the considered activities. The model can adapt itself even for a minimum sampling rate and subcarriers found in the test bed. | en_US |
| dc.identifier.citation | Kumar, V. D., Rajesh, P., Polat, K., Alenezi, F., Althubiti, S. A., & Alhudhaif, A. (2022). Wi-Fi signal-based human action acknowledgement using channel state information with CNN-LSTM: a device less approach. Neural Computing and Applications, 34(24), 21763-21775. | en_US |
| dc.identifier.doi | 10.1007/s00521-022-07630-6 | |
| dc.identifier.endpage | 21775 | en_US |
| dc.identifier.issn | 0941-0643 | |
| dc.identifier.issn | 1433-3058 | |
| dc.identifier.issue | 24 | en_US |
| dc.identifier.scopus | 2-s2.0-85135263550 | en_US |
| dc.identifier.scopusquality | Q1 | en_US |
| dc.identifier.startpage | 21763 | en_US |
| dc.identifier.uri | http://dx.doi.org/10.1007/s00521-022-07630-6 | |
| dc.identifier.uri | https://hdl.handle.net/20.500.12491/11669 | |
| dc.identifier.volume | 34 | en_US |
| dc.identifier.wos | WOS:000833451900005 | en_US |
| dc.identifier.wosquality | Q2 | en_US |
| dc.indekslendigikaynak | Web of Science | en_US |
| dc.indekslendigikaynak | Scopus | en_US |
| dc.institutionauthor | Polat, Kemal | |
| dc.language.iso | en | en_US |
| dc.publisher | Springer London Ltd | en_US |
| dc.relation.ispartof | Neural Computing and Applications | en_US |
| dc.relation.publicationcategory | Makale - Uluslararası Hakemli Dergi - Kurum Öğretim Elemanı | en_US |
| dc.rights | info:eu-repo/semantics/closedAccess | en_US |
| dc.subject | Activity Recognition | en_US |
| dc.subject | CNN | en_US |
| dc.subject | Kalman Filters | en_US |
| dc.subject | Fall Detection | en_US |
| dc.subject | Recognition | en_US |
| dc.subject | EEG | en_US |
| dc.title | Wi-Fi signal-based human action acknowledgement using channel state information with CNN-LSTM: A device less approach | en_US |
| dc.type | Article | en_US |
