Automated Intermediate Frequency Carrier Estimation by Using Deep Neural Network

Abstract

The intermediate frequency (IF) carrier is at most important parameter in an adaptive sixth-generation (6G) communication system for automated and reliable detection. In the adaptive communication system, the wide range of antennas is unable to provide accurate IF carrier. In addition to this, due to the Doppler effect or oscillator mismatch, the desired IF carrier gets changed and unable to convert the passband signal to an error-free baseband signal. In this paper, we have estimated IF carrier using a deep neural network (DNN). In statistical-based IF carrier estimation methods, signal bandwidth must be known prior to the IF carrier estimation, but the proposed scheme does not require knowledge of it. The proposed scheme also does not require any information on modulation schemes, channel state information (CSI), and synchronization parameters. The performance of the proposed scheme is evaluated in the presence of Rayleigh and Rician fading environment for five different modulation classes and compared with a statistical method by normalized mean square error (NMSE) performance metric. Simulation results show that the proposed method outperforms the statistical method at a low signal-to-noise ratio (SNR).