Timing‐deviation and frequency‐offset estimations for multicarrier transmission in high mobility environments using deep neural network

Abstract

SummaryThe multicarrier technology used in fourth‐generation (4G) and fifth‐generation (5G) communications is quite susceptible to multipath fading and Doppler frequency shift, which is more prominent in high mobility environments. Therefore, the accuracy of timing and frequency synchronization impacts the overall system performance significantly. Existing synchronization methods rely on the correlation of the preamble sequences and are, hence, vulnerable to severe multipath effect and Doppler effect. In this article, we propose a novel scheme leveraging deep neural network (DNN) to achieve high‐precision synchronization in high mobility environments. Concretely, a convolutional neural network (CNN) architecture is proposed to extract the hidden features of received signals for timing deviation estimation. In the following, a fully connected (FC) model is demonstrated to classify the optimal carrier frequency offset (CFO) estimate from several CFO candidates. The proposed synchronization scheme is assessed under extended vehicular A (EVA) channel with various Doppler frequency shifts. Simulation results corroborate that the proposed DNN‐based scheme achieves a significant performance gain over the conventional correlation methods, and the proposed timing deviation estimation scheme exhibits an excellent complexity reduction; wherefore, it is extremely promising for multicarrier transmission systems in high mobility environments.