Research And Platform Implementation Of OFDM Channel Estimation Based On Neural Network In Low SNR

Considering the rapid development of wireless communication technology,the requirements for transmission rate,spectrum utilization and stronger robustness of communication system is higher than formers.In the new generation communication transmission system with large-scale wireless channel,the channel with massive data can be easily disturbed by noise,which leads to serious distortion of performance in low SNR environment.In order to obtain high reliability and transmission rate in communication system,it is necessary for wireless channel to calculate the attribute information of wireless channel,which is the Channel State Information(CSI).The CSI can be obtained directly by using channel estimation technology.However,the effectiveness and performance of channel estimation are determined by whether CSI can be obtained effectively.Because of its defaults,the traditional channel estimation is hard to obtain excellent performance in the new generation communication system.Therefore,it is very important to find a new estimation method to calculate CSI.With the progress of Artificial Intelligence and Deep Learning Network,using Deep Learning Network to optimize the communications problems has attracted the attention of academia.Based on the channel estimation of OFDM,using Deep Learning Network for channel estimation is proposed to improve the performance of channel estimation in low SNR environment by this paper.The main work is as follows:Firstly,the traditional channel estimation algorithm is described in detail,including OFDM communication system,pilot structure and interpolation algorithm.Then,the Deep Learning Network is described,which provides reference for the following research in this paper.Secondly,in order to solve the problem that the traditional channel estimation algorithm is not capable of acquiring CSI in low SNR environment,Bi-LSTM network is used for training and learning to obtain CSI.Compared with classical estimation algorithm,Bi-LSTM has a performance gain of 2-4d B and the time complexity is as low as LS algorithm,which is about().Then,to solve the problem of the distortion problem under low SNR situation using neural network,in this paper,we research and use CGAN network to denoise the data set with increasing the amount of training data to enhance the channel estimation performance,through the constellation diagram and BER curve simulation result,in lower SNR range using CGAN network for the performance of channel estimation is about 3-5 d B gain,and can significantly improve deep learning network under the low SNR data analysis ability,and it also proved that deep learning technology can be combined with channel estimation technology.Finally,in order to further explore the application of deep learning technology in channel estimation in the actual scene,this paper uses edge computing technology to develop the basic discrimination system,and configures the function to the edge node through the edge computing platform,so that it can judge whether to use CGAN network model parameters to improve the estimation performance of channel estimation algorithm,which provides reference for the implementation of the algorithm for technical support.