Abstract:Spectrum resources have become a strategic and scarce national resource, and spectrum space has become the sixth dimensional battle space. Drone-based 5G mobile communications have a very narrow bandwidth of available frequencies, and mobile spectrum resources are particularly scarce. With the explosive growth of drone-based 5G mobile users, mobile spectrum space is facing serious challenges. Spectrum sensing is a new technology for rational allocation of mobile communication resources, improving mobile spectrum utilization and tapping into high-dimensional spatial communication resources. However, the drone-based 5G mobile channel is complex and variable, the drone-based mobile communication network is highly dynamic, and the received signal of drone-based mobile communication is highly susceptible to interference by noise, which leads to the problem of low accuracy rate of drone-based 5G mobile spectrum sensing. Therefore, this paper proposes an intelligent spectrum sensing algorithm based on Principal Component Analysis (PCA) and Recurrent Neural Network (RNN). Using the PCA method to extract the singular spectral entropy features of mobile wireless signals, the dimensionality of the features is reduced while the noise is removed to improve the quality of the features; the intrinsic connection between the mobile signal features and the spectrum sensing classifier is explored, and the RNN network model is designed to establish an intelligent spectrum sensing. Experimental results show that the proposed PCA-RNN sensing model achieves more than 95% sensing accuracy above -3dB, and improves about 30% at low signal-to-noise ratios (SNRs) compared with traditional algorithms Learning Vector Quantization (LVQ) and Support Vector Machine (SVM).

PCA-RNN-based intelligent mobile drone spectrum sensing algorithm