Device free human gesture recognition using Wi-Fi CSI: A survey
Device free human gesture recognition using Wi-Fi CSI: A survey
复制标题
DOI：
10.1016/j.engappai.2019.103281
复制
发表时间：
2020-01-01
影响因子：
8
通讯作者：
Aravind, C., V
中科院分区：
文献类型：
--
作者：
Ahmed, Hasmath Farhana Thariq;Ahmad, Hafisoh;Aravind, C., V
随着无线技术的最新进展，无设备感知人体手势得到了极大的研究关注。信道状态信息(CSI)是用于设备无关检测的Wi-Fi设备的指标，可实现更好的识别性能。这项调查将最先进的识别任务分为基于设备的和无设备的感知方法，并重点介绍了Wi-Fi CSI的进步。本文还全面总结了CSI在基于模型和基于学习两种方法下的无设备感知的识别性能。讨论了基于学习的方法下的机器学习和深度学习算法，并给出了相应的识别精度。讨论了手势识别中广泛采用的各种信号预处理、特征提取、选择和分类技术，以及影响识别精度的环境因素。这项调查得出了结论，指出了在使用Wi-Fi设备的CSI指标进行设备无关手势识别方面可以探索的挑战和机遇。
Device-free sensing of human gestures has gained tremendous research attention with the recent advancements in wireless technologies. Channel State Information (CSI), a metric of Wi-Fi devices adopted for device-free sensing achieves better recognition performance. This survey classifies the state of the art recognition task into device-based and device-free sensing methods and highlights advancements with Wi-Fi CSI. This paper also comprehensively summarizes the recognition performance of device-free sensing using CSI under two approaches: model-based and learning based approaches. Machine Learning and Deep Learning algorithms are discussed under the learning based approaches with its corresponding recognition accuracy. Various signal pre-processing, feature extraction, selection, and classification techniques that are widely adopted for gesture recognition along with the environmental factors that influence the recognition accuracy are also discussed. This survey presents the conclusion spotting the challenges and opportunities that could be explored in the device free gesture recognition using the CSI metric of Wi-Fi devices.