Communication Signal Modulation Recognition Technology Using Global Context Residual Recurrent Neural Network

Signal modulation recognition is a substantial technology in non cooperative wireless communication systems. It identifies a modulation type of an interrupted signal in the absence of prior knowledge which provides parameter data for consecutive demodulation. However, modulation recognition for radar signals has become a vital issue in electronic counter quantity systems in low Signal-to-Noise Ratio (SNR) because of enhancing amount of radar signals with intricate environments. This research proposes the Global-Context Residual Recurrent Neural Network (GC-RRNN) to recognize type of modulation signals in communication technology. The residual connections reduce vanishing gradient issues and make better training and variations in signals which makes them greatly efficient for complex modulation schemes. Different input signal sequence is utilized like Binary Frequency-Shift Keying (BFSK), Binary Amplitude Shift Keying (BASK), Continuous Wave (CW), Binary Phase Shift Keying (BPSK), Linear Frequency Modulation (LFM), and Stepping Frequency Wave (SFW) signals to evaluate GC-RRNN. The proposed GC-RRNN achieves a better accuracy of 98.67% for BPSK signal compared to Convolutional Neural Network (CNN), RNN, and RRNN.