Affiliation:
1. Simon Fraser University, Burnaby, BC, Canada
Abstract
Human activity identification plays a critical role in many Internet-of-Things applications, which is typically achieved through attaching tracking devices, e.g., RFID tags, to human bodies. The attachment can be inconvenient and considered intrusive. A tag-free solution instead deploys stationary tags as references, and analyzes the backscattered signals that could be affected by human activities in close proximity. The information offered by today's RFID tags however are quite limited, and the typical raw data (RSSI and phase angles) are not necessarily good indicators of human activities (being either insensitive or unreliable as revealed by our realworld experiments). As such, existing tag-based activity identification solutions are far from being satisfactory, not to mention tag-free. It is also well known that the accuracy of the readings can be noticeably affected by multipath, which unfortunately is inevitable in an indoor environment and is complicated with multiple reference tags.
In this paper, we however argue that multipath indeed brings rich information that can be explored to identify fine-grained human activities. Our experiments suggest that both the backscattered signal power and angle are correlated with human activities, impacting multiple paths with different levels. We present TagFree, the first RFID-based device-free activity identification system by analyzing the multipath signals. Different from conventional solutions that directly rely on the unreliable raw data, TagFree gathers massive angle information as spectrum frames from multiple tags, and preprocesses them to extract key features. It then analyzes their patterns through a deep learning framework. Our TagFree is readily deployable using off-the-shelf RFID devices and a prototype has been implemented using a commercial Impinj reader. Our extensive experiments demonstrate the superiority of our TagFree on activity identification in multipath-rich environments.
Publisher
Association for Computing Machinery (ACM)
Subject
Computer Networks and Communications,Hardware and Architecture,Human-Computer Interaction
Reference35 articles.
1. A Review of Smart Homes—Past, Present, and Future
2. Understanding Deep Features with Computer-Generated Imagery
3. Andreas Bulling Ulf Blanke and Bernt Schiele. 2014. A tutorial on human activity recognition using body-worn inertial sensors. Comput. Surveys 46 3 (2014). 10.1145/2499621 Andreas Bulling Ulf Blanke and Bernt Schiele. 2014. A tutorial on human activity recognition using body-worn inertial sensors. Comput. Surveys 46 3 (2014). 10.1145/2499621
4. A survey of video datasets for human action and activity recognition
5. Indoor localization without the pain
Cited by
30 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献