Utilizing Sensing Coil for Device Model Recognition in Qi-Compatible Wireless Charging

Wireless charging, particularly supported by the Qi standard, is becoming increasingly important as a power supply method for mobile electronics. It is crucial to accurately identify the device model, as it affects the maximum charging power and plays a significant role in optimizing charging and enhancing the user experience. However, the embedded query response communication-based device model recognition in the Qi specification is unreliable when facing security and privacy risks. Current research on hardware fingerprinting primarily focuses on identifying individual devices by using features related to hardware imperfections or communication traffic, which are inadequate for identifying the device model. This paper introduces SenModel, a communication-independent and cost-efficient system for identifying the device model. Our research exploits the differences between device models and how they impact electromagnetic fields, taking into account factors such as charging control and device structures. For the hardware implementation, we design an array of magnetic sensing coils that can capture the magnetic field characteristics of each device model, and we use demodulation-based sampling to obtain induced signals from these coils. For the software implementation, we develop a unique feature extraction process and implement a bagging algorithm for accurate device classification. Our experiment results demonstrate that SenModel can accurately recognize up to 23 device models with 97.87% precision, while also showcasing the consistency, scalability, robustness, and stability of our approach.