Performance Evaluation of a CNN-Based Channel Estimation for OFDM Systems in High-Mobility Scenarios

Next-generation intelligent transportation systems demand communication technologies that deliver low latency and high reliability. However, a significant challenge for wireless communication systems is channel estimation in high-mobility scenarios. Thanks to the innovation of computer science, numerous applications of machine learning, particularly deep learning, in wireless communication systems are deployed to handle more diverse obstacles of the future mobile networks. Recently, deep learning-based channel estimations for wireless applications have garnered lots of attention from both researchers and industry. Among these efforts, a novel convolution neural network (CNN)-based channel estimation approach named ChannelNet was proposed for OFDM systems. ChannelNet treats the channel response matrix in the time-frequency domain as 2D images then the deep image processing techniques are used to enhance the intelligibility and reliability of channel estimation. Inspired by this reason, this paper gives a brief study of the abovementioned technique with the performance evaluation and comparison with current existing channel estimation techniques in high-mobility scenarios.