Detection and Authentication for Cross-Technology Communication

Cross-Technology Communication (CTC) introduces novel security challenges, demanding urgent mitigation strategies. Although recent literature offers the possibility of detecting malicious CTC, they commonly require access to In-phase and Quadrature (IQ) signals, thus not compatible with existing billions of IoT devices. In this paper, we propose a lightweight model to detect unauthorized CTC signals at ZigBee devices by analyzing CTC's inherent chip error patterns that are available on commodity ZigBee devices. To further enhance the detection accuracy, we propose an augmentation framework that integrates both theoretical analysis and channel fading models, and adopts an LSTM-based deep learning method. As for authentication of authorized CTC, we introduce a dynamic method that flips the error-prone chips. This enables authentication without introducing additional chip errors, further ensuring the transparency of existing CTC. Evaluation with testbeds demonstrates transparent detection and reliable authentication.