Convolutional Neural Network (CNN)-Based Frame Synchronization Method

Abstract

A new frame synchronization technique based on convolutional neural network (CNN) is proposed for synchronized networks. To estimate the exact packet arrival time, the receiver typically uses the correlator between the received signal and the preamble or pilot in front of the transmitted packet. The conventional frame synchronization technique searches the correlation peak within the time window. In contrast, the proposed method utilizes a CNN to find the packet arrival time. Specifically, in the proposed method, the 1D correlator output is converted into a 2D matrix by reshaping, and the resulting signal is inputted to the proposed 4-layer CNN classifier. Then, the CNN predicts the packet arrival time. To verify the frame synchronization performance, computer simulation is performed for two channel models: additive white Gaussian noise and fading channels. Simulation results show that the proposed CNN-based synchronization method outperforms the conventional correlation-based technique by 2dB.

ELM-Based Frame Synchronization in Nonlinear Distortion Scenario Using Superimposed Training

“…The DL-based FS has been investigated in [26]- [29]. In [26], an artificial neural network (ANN)-based synchronization method was proposed.…”

“…In [27], to achieve the task of FS, a deep neural network (DNN) was employed to auto-encoder, and a convolutional neural network (CNN) was developed in [28] to compensate impairments introduced by timing offset and timing error of sampling. As for [29], a CNN-based FS method was proposed to convert the one dimension (1D) correlator to two dimension (2D) matrix to find the frame offset. From [26]- [29], the DL technology provides effective approaches for FS.…”

“…As for [29], a CNN-based FS method was proposed to convert the one dimension (1D) correlator to two dimension (2D) matrix to find the frame offset. From [26]- [29], the DL technology provides effective approaches for FS. Nevertheless, these DL-based FSs are still challenged by many issues, such as long training time, complex parameter tuning, large memory requirements, etc.…”