Device-Free Activity Recognition for Underground Spaces Based on Convolutional Neural Network
Inspired by the excellent work of wireless sensing, we propose a non-invasive activity recognition system, Under-Sense, for underground space sensing with a pair of commodity Wi-Fi devices. Firstly, by extracting relative phase information from all 90 subcarriers, we construct fine-grained images and then compress the rectangle images into k-dimension by singular value decomposition (SVD). A nine-layer convolutional neural network (CNN) is designed to automatically extract important features from constructed images and classify five human activities. Our results show Under-Sense could achieve 99.5% average accuracy in the empty meeting room and 96.7% in complex student studio environment.
KeywordsSmart underground space Wi-Fi-based sensing Activity recognition Convolutional neural network
Acknowledgements, Ethical Approval and Informed Consent
This work is supported by the National Key Research and Development Project of China No. 2017YFC0704100 (entitled New Generation Intelligent Building Platform Techniques).
This study has been conducted with ethics approval obtained from the ethics committee: Prof. Qiliang Yang and Prof. Ping Wang of Department of National Defense Engineering, Army Engineering University of PLA, Jiangsu, China.
Participants are recruited from Army Engineering University of PLA, Jiangsu, China, and all participants have given their consent to use the dataset and disclose information relevant for research in this study.
- 2.Liu, T., Lu, D.: The application and development of IoT. In: Information Technology in Medicine and Education (ITME), 2012 International Symposium, vol. 2, pp. 991–994. IEEE (2012)Google Scholar
- 3.Qian, K., Wu, C., Yang, Z., Liu, Y., Zhou, Z.: PADS: passive detection of moving targets with dynamic speed using PHY layer information. In: Parallel and Distributed System, 2014 20th IEEE International Conference, pp. 1–8. IEEE (2014)Google Scholar
- 4.Wang, X., Wang, X., Mao, S.: Cifi: deep convolutional neural networks for indoor localization with 5 GHz Wi-Fi. In: Communications (ICC), 2017 International Conference, pp. 1–6. IEEE (2017)Google Scholar
- 5.Qian, K., Wu, C., Yang, Z., Liu, Y., Jamieson, K.: Widar: decimeter-level passive tracking via velocity monitoring with commodity Wi-Fi. In: Proceedings of the 18th ACM International Symposium on Mobile Ad Hoc Networking and Computing. ACM (2017)Google Scholar
- 6.Zhou, Q., Wu, C., Xing, J., Li, J., Yang, Z., Yang, Q. Wi-Dog: Monitoring school violence with commodity Wi-Fi devices. In: International Conference on Wireless Algorithms, Systems, and Applications, pp. 47–59. Springer, Cham (2017)Google Scholar
- 7.Wang, X., Gao, L., Mao, S., Pandey, S.: CSI-based fingerprinting for indoor localization: a deep learning approach. IEEE Trans. Veh. Technol. 66(1), 763–776 (2017)Google Scholar
- 8.Wu, C., Yang, Z., Zhou, Z., Qian, K. PhaseU: real-time LOS identification with WiFi. In: Computer Communications, pp. 2038–2046. IEEE (2015)Google Scholar
- 9.Krizhevsky, A., Sutskever, I., Hinton, G.: ImageNet classification with deep convolutional neural networks. In: International Conference on Neural Information Processing Systems, vol. 60, pp. 1097–1105. Curran Associates Inc. (2012)Google Scholar
- 10.Qian, K., Wu, C., Zhou, Z., Zheng, Y., Yang, Z., Liu, Y.: Inferring motion direction using commodity Wi-Fi for interactive exergames. In: Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems, pp. 1961–1972 (2017)Google Scholar