臺灣博碩士論文加值系統

Human activity recognition without equipment plays a vital role in smart home applications, freeing humans from the shackles of wearable devices. In this paper, by using the channel state information (CSI) of the WiFi signal, a semi-supervised transfer learning with dynamic associate domain adaptation is proposed for human activity recognition. In order to improve the CSI quality and denoising of CSI, we carried out missing packet filling, burst noise removal, Background estimation, feature extraction, feature enhancement, and data augmentation in the data preprocessing stage. This paper considers the problem of environment-independent human activity recognition, also known as domain adaptation. The pre-trained model is trained from the source domain by collecting a complete labeled data set of all CSI of human activity patterns. Then the pre-trained model is transferred to the target environment through the semi-supervised transfer learning stage. Therefore, when humans move to different target domains, partial labeled data set of the target domain are required for fine-tuning. In this paper, we propose a dynamic associate domain adaptation called DADA. By modifying the existing associate domain adaptation algorithm, the target domain can provide a dynamic ratio of labeled data set/unlabeled data set, while the existing associate domain adaptation algorithm only allow target domains with the unlabeled data set. The advantage of DADA is that it provides a dynamic strategy to eliminate different effects on different environments. In addition, we further designed an attention-based DenseNet model, or AD, as our training network, which is modified by existed DenseNet by adding the attention function. The solution we proposed was simplified to DADA-AD throughout the paper. The experimental results show that for domain adaptation in different domains, the accuracy of human activity recognition of the DADA-AD scheme is 97.4%. It also shows that DADA-AD has advantages over existing semi-supervised learning schemes.

無需設備的人類活動識別在智能家居應用中發揮著至關重要的作用，將人類從可穿戴設備的束縛中解放出來。在本文中，通過使用 WiFi 信號的信道狀態信息 (CSI)，提出了一種具有動態關聯域自適應的半監督轉移學習用於人類活動識別。為了提高CSI質量和CSI去噪，我們在數據預處理階段進行了缺失包填充、突發噪聲去除、背景估計、特徵提取、特徵增強和數據增強。本文考慮了環境無關的人類活動識別問題，也稱為領域適應。通過收集人類活動模式的所有 CSI 的完整標記數據集，從源域訓練預訓練模型。然後通過半監督遷移學習階段將預訓練好的模型遷移到目標環境中。因此，當人類移動到不同的目標域時，需要對目標域的部分標記數據集進行微調。在本文中，我們提出了一種稱為 DADA 的動態關聯域自適應。通過修改現有的關聯域自適應算法，目標域可以提供標記數據集/未標記數據集的動態比例，而現有關聯域自適應算法只允許目標域具有未標記數據集。 DADA 的優勢在於它提供了一種動態策略來消除對不同環境的不同影響。此外，我們進一步設計了一個基於注意力的 DenseNet 模型，或 AD，作為我們的訓練網絡，它通過添加註意力函數來修改現有的 DenseNet。我們提出的解決方案在整篇論文中被簡化為 DADA-AD。實驗結果表明，對於不同領域的領域適應，DADA-AD方案的人體活動識別準確率為97.4%。它還表明，DADA-AD 優於現有的半監督學習方法。

1. Introduction
2. Related Work
3. Preliminaries
4. A Semi-Supervised Transfer Learning with Dynamic Associate Domain Adaptation for HAR
5. Experimental Results
6. Conclusion

[1]A. Gupta, K. Gupta, K. Gupta, and K. Gupta, “A Survey on Human Activity Recognition and Classification, “International Conference on Communication and Signal Processing (ICCSP), pp. 0915-0919, Chennai, India, July 2020.
[2]S. M. Kwon, S. Yang, J. Liu, X. Yang, W. Saleh, and S. Patel, “Demo: Hands-Free Human Activity Recognition using Millimeter-Wave Sensors, “IEEE International Symposium on Dynamic Spectrum Access Networks (DySPAN), pp. 1-2, Newark, NJ, USA, Nov. 2019.
[3]N. S. R. Challa, P. Kesari, S. R. Ammana, S. Katukojwala, and D. S. Achanta, “Design and Implementation of Bluetooth-Beacon Based Indoor Positioning System, “IEEE International WIE Conference on Electrical and Computer Engineering (WIECON-ECE), pp. 1-4, Bangalore, India, Nov. 2019.
[4]T. F. Sanam and H. Godrich, “FuseLoc: A CCA Based Information Fusion for Indoor Localization using CSI Phase and Amplitude of WiFi Signals, “IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 7565-7569, Brighton, UK, May 2019.
[5]F. Wang, J. Feng, Y. Zhao, X. Zhang, S. Zhang, and J. Han, “Joint Activity Recognition and Indoor Localization with WiFi Fingerprints, “IEEE Access, vol. 7, pp. 80058-80068, June. 2019.
[6]S. Sigg, T. Braunschweig, S. Shi, F. Buesching, Yusheng Ji, and Lars Wolf. “Leveraging RF-Channel Fluctuation for Activity Recognition Active and Passive Systems, Continuous and RSSI-based Signal Features “, International Conference on Advances in Mobile Computing & Multimedia, New York, USA, Dec. 2013.
[7]S. Orphomma, and N. Swangmuang, “Exploiting the Wireless RF Fading for Human Activity Recognition, “International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology, pp. 1-5, Krabi, Thailand, May 2013.
[8]W. Zhang, S. Zhou, D. Peng, L. Yang, F. Li, and H. Yin, “Understanding and Modeling of WiFi Signal-Based Indoor Privacy Protection, “IEEE Internet of Things Journal, vol. 8, no. 3, pp. 2000-2010, Aug. 2020.
[9]Y. Wang, K. Wu and L. M. Ni, “Wifall: Device-Free Fall Detection by Wireless Networks, “IEEE Transactions on Mobile Computing, vol. 16, no. 2, pp. 581-594, April 2016.
[10]M. Huang, J. Liu, Y. Gu, Y. Zhang, F. Ren, and X. Wang, “Your WiFi Knows You Fall: A Channel Data-Driven Device-Free Fall Sensing System, “IEEE International Conference on Communications (ICC), pp. 1-6, Shanghai, China, May 2019.
[11]C. Han, K. Wu, Y. Wang and L. M. Ni, “Wifall: Device-Free Fall Detection by Wireless Networks, “IEEE Conference on Computer Communications, pp. 271-279, Toronto, ON, Canada, May 2014.
[12]Y. Gu, X. Zhang, C. Li, F. Ren, J. Li, and Z. Liu, “Your WiFi Knows How You Behave: Leveraging WiFi Channel Data for Behavior Analysis, “IEEE Global Communications Conference (GLOBECOM), pp. 1-6, Abu Dhabi, United Arab Emirates, Dec. 2018.
[13]S. Zhong, Y. Huang, R. Ruby, L. Wang, Y. Qiu, and K. Wu, “Wi-Fire: Device-Free Fire Detection using WiFi Networks, “IEEE International Conference on Communications (ICC), pp. 1-6, May 2017.
[14]H. Li, X. He, X. Chen, Y. Fang, and Q. Fang, “Wi-Motion: A Robust Human Activity Recognition using WiFi Signals, “IEEE Access, vol. 7, pp. 153287-153299, Oct. 2019.
[15]F. Wang, W. Gong, and J. Liu, “On Spatial Diversity in WiFi-Based Human Activity Recognition: A Deep Learning-Based Approach, “IEEE Internet of Things Journal, vol. 6, no. 2, pp. 2035-2047, Sept. 2018.
[16]Z. Wang, “A Survey on Human Behavior Recognition Using Channel State Information, “IEEE Access, vol. 7, pp. 155986-156024, Oct. 2019.
[17]D. Vasisht, S. Kumar, and D. Katabi, “Decimeter-Level Localization with A Single WiFi Access Point, “Networked Systems Design and Implementation (NSDI 16), pp. 165–178, Santa Clara, CA, March 2016.
[18]F. Wang, W. Gong, J. Liu, and K. Wu, “Channel Selective Activity Recognition with WiFi: A Deep Learning Approach Exploring Wideband Information, “IEEE Transactions on Network Science and Engineering, vol. 7, no. 1, pp. 181-192, April 2018.
[19]H. Zou, Y. Zhou, J. Yang, H. Jiang, L. Xie, and C. J. Spanos, “Deepsense: Device-Free Human Activity Recognition Via Autoencoder Long-Term Recurrent Convolutional Network, “IEEE International Conference on Communications (ICC), pp. 1-6, Kansas City, MO, USA, May 2018.
[20]Z. Shi, J. A. Zhang, R. Xu, and Q. Cheng, “Deep Learning Networks for Human Activity Recognition with CSI Correlation Feature Extraction, “IEEE International Conference on Communications (ICC), pp. 1-6, Shanghai, China, July 2019.
[21]Y. Li, T. Jiang, X. Ding, and Y. Wang, “Location-Free CSI Based Activity Recognition with Angle Difference of Arrival, “IEEE Wireless Communications and Networking Conference (WCNC), pp. 1-6, Seoul, Korea, May 2020.
[22]Z. Shi, J. A. Zhang, R. Xu, Q. Cheng, and A. Pearce, “Towards Environment-Independent Human Activity Recognition using Deep Learning and Enhanced CSI, “IEEE Global Communications Conference, pp. 1-6, Taipei, Taiwan, Dec 2020.
[23]X. Ding, T. Jiang, Y. Li, W. Xue, and Y. Zhong, “Device-Free Location-Independent Human Activity Recognition using Transfer Learning Based On CNN, “IEEE International Conference on Communications Workshops (ICC Workshops), pp. 1-6, Dublin, Ireland, June 2020.
[24]Z. Han, L. Guo, Z. Lu, X. Wen, and W. Zheng, “Deep Adaptation Networks Based Gesture Recognition using Commodity WiFi, “IEEE Wireless Communications and Networking Conference (WCNC), pp. 1-7, Seoul, Korea, May 2020.
[25]S. Arshad, C. Feng, R. Yu, and Y. Liu, “Leveraging Transfer Learning in Multiple Human Activity Recognition using WiFi Signal, “IEEE 20th International Symposium on “A World of Wireless, Mobile and Multimedia Networks “(WoWMoM), pp. 1-10, Washington, DC, USA, 2019.
[26]X. B. Peng, A. Kanazawa, S. Toyer, P. Abbeel, and Sergey Levine, “Tool Release: Gathering 802.11n Traces with Channel State Information, “ACM SIGCOMM Computer Communication Review, Volume 4, January 2011.
[27]P. Haeusser, T. Frerix, A. Mordvintsev, and D. Cremers, “Associate Domain Adaptation, “IEEE International Conference on Computer Vision (ICCV), pp. 2784-2792, Venice, Italy, Oct. 2017.
[28]Q. Wang, B. Wu, P. Zhu, P. Li, W. Zuo, and Q. Hu, “ECA-Net: Efficient Channel Attention for Deep Convolutional Neural Networks, “IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 11531-11539, Seattle, WA, USA, June 2020.
[29]J. Stuart Hunter, The Exponentially Weighted Moving Average, “Journal of Quality Technology”, 18:4, 203-210, 1986.
[30]S. Pragada and J. Sivaswamy, "Image Denoising Using Matched Biorthogonal Wavelets," Sixth Indian Conference on Computer Vision, Graphics & Image Processing, pp. 25-32, 2008.
[31]G. Huang, Z. Liu, L. Van Der Maaten and K. Q. Weinberger, "Densely Connected Convolutional Networks," IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 2261-2269, 2017.