With the continuous development of RFID technology, its applications in logistics management, cargo monitoring, conference security and other fields are becoming broader and broader. In wireless communication technology, the air interface defines the technical specification of the radio link between the terminal device and the network device. Most of the current RFID devices use common standardized communication protocol for data transmission, which makes the RFID systems suffer air interface intrusion by malicious devices. The air interface intrusion can then cause the security threats and data privacy protection problem in RFID systems. This study accomplishes the detection of air interface intrusion based on the real-time discovery of malicious readers. It avoids theft of data and ensures the security of data transmission. The paper mainly uses passive sensing technology for the analysis and calculation of the RFID signal, channel state and throughput information. Parameters are extracted from the received signal strength, phase and other state information to describe the wireless channel state. The extracted parameters and finite-state machine theory are used to build a perception data inference model of RFID signal and to obtain an initial steady state according to the adaptive algorithm. As a result, the specific change of RFID signal can be analyzed to complete the detection of RFID air interface intrusions.