In the design of Cognitive Radio Mobile Networks, the detection of attacks such as primary user emulation is essential as it avoids secondary users interfering with primary users of the cellular mobile network. In addition, it allows a primary user to be distinguished from a primary user emulator attacker, considering that the mobile cognitive radio user makes use of dynamic spectrum allocation and performs variable frequency hopping over time, which makes it vulnerable to attack. The research problem consists in establishing a model that allows obtaining information from different layers of the network to detect the primary user emulation attack in Cognitive Radio Mobile Networks where the attacking user has a dynamic location, whose received power is variable with the change of position and its objectives can be to transmit in the primary frequency or generate noise to prevent transmission from other users. On the other hand, detection systems have been developed for attackers with a static location whose power and position are fixed, and for the mobile cellular network users are in motion. In this research work, the cross-layer design is used to detect the primary user emulation attack with a static and dynamic location in Cognitive Radio Mobile Networks, through the study of the characteristics of the signal and its behavior over time and frequency. The model design integrates techniques such as energy detection, entropy, trilateration for location, and application-level information in the sensors to differentiate the attacker from a primary user. A real-case test of the model is performed on software-defined radio devices, allowing a comparison between theory, simulation, and practical results. Results found in the real case test with software-defined radio equipment allowed proving the applicability of the detection model in Cognitive Radio Mobile Networks. The results show that by using the crosslayer design, a technique for PUE detection is generated for static and dynamic cases that allows an exchange of information between the layers combining the energy detection technique that has PUE detection results over 93%, the detection technique by location that allows detecting the user's movement, with 97% of effectiveness using RSSI and trilateration with the least squares technique, the entropy detection technique that improves results by 8dB for low SNR signals as it is less sensitive to noise and the application technique that using information from the application layer extracted from primary users, allows the short name and operator data to be compared with the PUE, achieving 100% of PUE detection. These techniques were implemented in SDR equipment, using GNURadio and OpenBTS as base software.