Real-Time Spectrum Sensing for TDD-LTE Signals Using Deep Learning

Existing spectrum sensing techniques for Time Division Duplex-Long Term Evolution (TDD-LTE) signals primarily focus on large time-scale signals. However, the spectrum holes that exist in the small time-scale special subframes of TDD-LTE are often ignored. This paper proposes a method for spectrum sensing within TDD-LTE special subframes by applying Parallel Convolutional Neural Networks (ParallelCNN). The effective feature extraction capability of ParallelCNN enhances the spectrum sensing performance. Simulation results demonstrate superior sensing performance under various signal-to-noise ratio (SNR) conditions.