Spectrum Sensing Based on STFT-ImpResNet for Cognitive Radio

Spectrum sensing is a crucial technology for cognitive radio. The existing spectrum methods generally suffer from certain problems, such as insufficient signal feature representation, low efficiency, high sensibility to noise uncertainty, and drastic degradation in deep networks. In view of these challenges, we propose method based on short-time Fourier transform improved residual network (STFT-ImpResNet) this work. Specifically, STFT, the received transformed into two-dimensional time-frequency matrix which normalized gray image input network. An designed classify samples, dropout layer added block mitigate over-fitting effectively. We conducted comprehensive evaluations proposed method, demonstrate that—compared with other current algorithms—STFT-ImpResNet exhibits higher accuracy lower computational complexity, well strong robustness it can meet needs real-time detection.