Monitoring Smart Home Traffic under Differential Privacy

Recent years have witnessed the proliferation of smart home ecosystems. Well-characterized traffic generated by smart home devices has promoted the development of security enhancing techniques for smart homes but exposes users to the privacy disclosure risk at the same time. Malicious eavesdroppers can infer working states of smart home devices and user activities based on spatial-temporal traffic characteristics. Existing countermeasures towards this kind of side channel attack ignore the utility of smart home traffic profiles and signatures for network management and attempt to completely eliminate them. In this paper, we give a comprehensive study on the trade offs between the usability of smart home traffic for security monitoring and its privacy threat. We propose to monitor the smart homes under differential privacy. Based on our solution, decoy traffic can be generated in a controlled manner so as to confound the attackers without disturbing the running monitoring systems. We prototyped our proposal and demonstrate its effectiveness empirically. An interview study is also conducted to learn the user acceptance of the proposed privacy preserving mechanism.