SNR Estimation Based on Machine Learned Modulation Classification for Link Adaptation

The evolution of artificial intelligence has contributed to the use of deep learning in Automatic Modulation Classification (AMC) and has produced excellent results. Current wireless communication systems depend on the AMC, which accurately evaluates the modulation type of receiving signals. This work adopts a deep learning method for implementing radio signal identification tasks. Signal-to-Noise Ratio (SNR) estimation considerably affects the efficiency of various wireless communication techniques. This work proposes a novel link adaptation technique with AMC and SNR detection. A Convolutional Neural Network (CNN) is used to recognize the modulation type of receiving signals. The moment-based method is applied to detect the SNR of the QPSK sig nal, and a non-data-aided SNR estimation algorithm is used to detect the SNR of 16 QAM and 64 QAM signals. The suggested method is simulated for signals received through the AWGN channel. An overall confusion matrix of the test data was obtained for performance analysis.