Indoor Semantic Recognition Research

Indoor environment always plays an extremely significant role in the lives and work of people,thus there are a large number of Indoor Location-based Services(ILBs)developed.For ILBs,the most significant factor is indoor map.An valuable indoor map always includes two key elements: floor map and indoor semantic(such as an elevator,a washroom and an emergency exit,etc).An important indoor map semantic can greatly enrich the indoor map,to guide users to find their targets better.In recent years,there is much research in the field of indoor semantic recognition.However,we have to face many urgent problems required to solve.First,the existing indoor semantic recognition systems mainly depends on mobile sensing to discover the patterns of indoor facilities,so it is hard to obtain enough recognition patterns of high-level indoor semantics.In the current indoor electronic map,the richness of high-level indoor semantic need be improved.Second,there is a lack of devicefree recognition systems,leading to inconvenience for users.Mobile sensing asks users to carry devices(such as smartphone or smartwatch)and requires them to install software early for data collection,which may damage the user experience.Third,with the scope and accuracy of indoor semantic recognition being improved continually in our research,there is a security issue of indoor semantic privacy,such as the CSI based indoor privacy protection.To the best of our knowledge,there is a lack of the relevant research on it.To solve these significant problems,we perform the research consisting of three factors as follows.1.Multi-length window framework based high-level indoor semantic inference.The existing indoor semantic recognition schemes are mostly capable of discovering patterns through smartphone sensing,but it is hard to recognize enough high-level indoor semantics for map enhancement.In this work we present Deep Map+,an automatical inference system for recognizing high-level indoor semantics using complex human activities with wrist-worn sensing.Deep Map+ is the first deep computation system using deep learning based on a multi-length window framework to enrich the data source.Furthermore,we propose novel methods of increasing virtual features and virtual samples for Deep Map+ to better discover hidden patterns of complex hand gestures.We have performed 23 high-level indoor semantics(including public facilities and functional zones)and collected wrist-worn data at a Wal-Mart supermarket.The experimental results show that our proposed methods can effectively improve the classification accuracy.2.WiFi human activity and environment based high-level indoor semantic inference.Existing indoor semantic recognition schemes are capable of discovering patterns by smartphone sensing.However,there is a lack of a device-free indoor semantic recognition system.To solve it,we propose WiFi Map+ that is a first-ever automatical inference system using WiFi signals to recognize high-level indoor semantics from human activities and environments,where the high-level indoor semantics consist of indoor facilities and environments.To characterize the static indoor environments and dynamical human activities separately with channel state information(CSI),we propose a novel two-stream architecture to generate the spatial streams and the movement streams independently.Compared to the recent research on activity recognition,this two-stream architecture can make the content area of CSI samples extend from human activities to indoor environments.For obtaining accurate indoor environment detection,we propose a CSI-environment model with a spatial stream generation algorithm,which can reduce the effect of human activities on environment detection.For stable activity recognition,we also propose an environment-based testing sample representation(ETSR)method,which can utilize the environment knowledge to overcome the diversity of CSI caused by the environment changes.Finally,we implement WiFi Map+ using commercial WiFi devices and evaluate its performance for seven common semantic detection cases in six-room scenarios.The experimental results show that our proposed WiFi Map+ is robust to the multi-room scenario and can achieve the average accuracy of 92.8% and the lowest accuracy of about 82%,respectively.3.CSI signal based indoor semantic privacy protection.Existing WiFi recognition schemes are capable of discovering patterns of indoor semantics such as human activity,identity,indoor environment and so on.We note that,channel state information(CSI)presents an opportunity for hackers to learn the indoor privacy,however,currently there is a lack of security research on CSI.We are the first to discuss and define the security problem of CSI signals,which is further extended to the non-targeted protection and targeted protection problems.To solve them,we present two types of adversarial autoencoder networks(AAENs),respectively.Through replacing the original signals with the generated adversarial ones,the protected semantic features are modified effectively,and the significant features of the other semantics required to be recognized are reserved.Intensive evaluations demonstrate that,with the proposed AAENs,the recognition accuracy of the protected semantic can be significantly decreased,while still maintaining the other semantics to be identified correctly.