Abstract: As a typical application of Internet of things (IoT) technology, smart home platform is gradually entering thousands of households. However, the security issues of smart home platforms have hindered their further market expanding. The research on the security of smart home platforms is still at early stages, but many security threats have been proposed by researchers recently. In this paper, we review the typical the architectures of current popular smart home platforms, and analyze their attack interfaces on all components. In the area of information interface security, this paper analyzes the data steganography problems existing in the smart camera’s image interface, as well as the adversarial examples aiming at the voice control interface. In the aspect of cloud backend security, this paper analyses the security rulers’ breaches and the data leakage when running the smart applications. For these security problems, this paper reviews some existing solutions and their limitations. Furthermore, for smart home platform, this paper proposes a third-party malicious application detection system using wireless traffic analysis without any modification on the current platform. This paper introduces the design details of the proposed system, and demonstrates its effectiveness on the Samsung SmartThings platform, and proposes a privacy protection mechanism for this system.

Key words: smart home platform, security analysis, privacy protection, adversarial machine learning, malicious application detection