The Key Security Techniques For Smart Home Based Internet Of Things

As a typical application of the Internet of things(Io T),the smart home system is widely deployed in people’s production and life.The smart home is becoming a popular lifestyle.Different from the traditional Io T system,in the smart home environment,smart devices are connected through wireless networks,managed uniformly through smart applications,and provide users with non-contact interfaces such as the voice interface.Smart terminal devices(such as smartphones,tablets,and smart sensing devices)provide users with richer functions and can sense the environmental changes of the smart home in real-time.With the popularization of the voice interface,users can interact with the smart home system without contact.Smart applications can automatically control devices and adjust the status of the smart home system.However,the current researches show that smart home still faces a wide range of security threats.Firstly,although the wireless communication technology adopted by the terminal devices speeds up the network communication rate and expands the network range,it also increases the privacy risks of users when encountering a side-channel attack.Secondly,due to the open nature of the voice channel,the voice interface faces various voice spoofing attacks.Lastly,the smart application platform suffers from the abnormal behavior of smart applications.To ensure the security and privacy of a smart home system,this dissertation studies the corresponding key technologies of smart home security following the logic of “terminal device — user interface— application platform”.Firstly,at the layer of the terminal device,this dissertation studies the side channel privacy threats caused by the side-channel attack aiming at the wireless communication protocol.This dissertation reveals a cross-layer privacy attack mechanism based on wireless signal channel state information Wind Talker.The new attack can integrate the channel state information from the physical layer and the side channel information of the network traffic to infer the user’s keystroke information on the device screen,and then steal the mobile payment password used by the user on the terminal device.To resist this security risk,this dissertation proposes a defense scheme based on the physical layer signal obfuscation,which effectively protects the privacy and security of the terminal device.Secondly,at the layer of voice interface,aiming at the vulnerability of voice interface to voice deception attack,this dissertation first studies the living detection system based on two-factor authentication.To relieve previous works’ requirement of carrying sensing devices,this dissertation proposes a liveness detection scheme named WSVA via leveraging the Wi-Fi signal which is ubiquitous in the smart home environment.The system uses the wireless Wi-Fi signal to characterize the user’s mouth movement and then determines whether the voice command received by the voice interface is an authentic voice command or a spoofing command by judging the consistency between the user’s mouth movement and the voice signal.In this dissertation,for the two different signals of wireless signal and speech signal,the features are designed from the macro level and mouth motion level,and the accurate speech in liveness detection performance is realized through feature combination.Then,at the voice interface layer,considering that all two-factor authentication based liveness detection schemes need to collect additional data(e.g.,Wi-Fi signal,ultrasonic signal),to further improve the universality of liveness detection,this dissertation considers the passive liveness detection schemes that only depend on the collected voice signal.To overcome the defect that the traditional passive liveness detection schemes are vulnerable to performance degradation in different environments and the limitation of strict requirements on user posture,we designed Array ID.This system uses the microphone array commonly equipped with smart speakers to achieve more robust liveness detection performance.Compared with traditional mono or dual-channel based passive liveness detection schemes,leveraging the microphone array can overcome the influence of environmental changes and user posture changes.Array ID is a lightweight detection system.In this dissertation,14 different devices are used to conduct voice spoofing attacks,and both authentic and spoofing voice samples are collected.The experimental results show that Array ID outperforms existing passive liveness detection schemes,and retains high robustness under many factors including distance and direction.Finally,at the layer of smart application platform,to solve the threat of the abnormal behavior of the smart application,this dissertation proposes a third-party application anomaly monitoring system Ho Monit.Ho Monit neither requires modifies the application platform nor installs patches on the smart applications.Instead,Ho Monit can accurately monitor the abnormal behavior of applications by analyzing the side channel information of wireless communication traffic.This dissertation introduces the design overflow of Ho Monit,and verifies the accuracy of Ho Monit on a smart home platform named Samsung Smart Things.Considering that the wireless side-channel information may also cause privacy disclosure,this dissertation further proposes a dummy traffic based privacy protection mechanism to thwart the potential privacy inference attack.