DFTIMP: Distributed Fake Traffic Injection from Multiple Points for Obfuscation of IoT Traffic

DFTIMP: Distributed Fake Traffic Injection from Multiple Points for Obfuscation of IoT Traffic

연구 분야: Analysis

학회: 2025 IEEE International Conference on Consumer Electronics (ICCE)

The rapid proliferation of Internet of Things (IoT) devices in smart homes has raised user privacy concerns through various attacks by analyzing wireless communication traffic. Such attacks often use machine learning algorithms to identify device categories and working states to map them to user behavior. Traffic obfuscation such as packet padding, packet shaping, and fake traffic injection can address these kinds of user privacy inference attacks. However, such traffic obfuscation methods generally rely on a single injection device and cannot scale effectively as network size increases. To overcome this issue, we propose DFTIMP (Distributed Fake Traffic Injection from Multiple Points) for obfuscation of IoT traffic. We conducted experiments to evaluate the effectiveness of traffic obfuscation reducing the overhead of each device. Experimental results show that the proposed method reduces the overhead of individual injection devices without degrading their obfuscation effectiveness, making it suitable for application to large networks.