In this study, we propose a method to identify Wi-Fi devices by analyzing the inter-arrival time and periodicity of probe request frames, without requiring any device information in the frames. First, we group the probe request frames according to the number of frames within three different time intervals. Second, we analyze the periods of the frames through a correlation function. Third, we estimate a threshold to divide each group into smaller groups. Finally, we identify the devices by tracking the small groups. For a small group, if it can be found again after a certain period, we consider that one device is identified. We utilized smartphones to conduct experiments and evaluated the proposed method based on detection and identification rates. When there was only one sniffer, in small-scale experiments, the average detection rate was 100.0% and identification rate was approximately 96.0%, whereas in middle-scale experiments, the average detection rate was approximately 89.0% and identification rate was approximately 73.8%. Although the detection and identification rates decreased with the increase in scale, we also conducted experiments utilizing multiple sniffers, and the results show that by increasing the sniffer, the detection and identification rates can be improved.
Qi et al. (Sat,) studied this question.