Study On LDPC Decoder Based On Deep Learning Under Vehicle Channel

With the proliferation of vehicles in recent years,Vehicles communication,as an important part of the Internet of Vehicles,can effectively help us alleviate a series of problems caused by the increase of cars.This is to ensure that Vehicles communication can achieve real-time communication and high fidelity communication quality.Compared with the indoor,due to the relatively high speed movement of both transmiters and receivers,the Vehicles channel has a larger Doppler shift and delay,which makes higher requirements for the reliability and real-time performance of the communication system.This article mainly focuses on the improvement of bit error rate performance of decoding algorithm in vehicle channel.Compared with indoor,vehicular channel has larger Doppler frequency shift and delay due to the relative high-speed movement of the transmitter and the receiver,which makes higher requirements for the reliability and real-time performance of the communication system.Deep residual shrinkage networks(DRSN)has strong feature extraction ability.The feature extraction ability of DRSN is used to assist LDPC decoding algorithm.Experiments show that the scheme of using DRSN feature extraction ability to assist LDPC decoding is not only feasible in fast fading channel,but also the decoding performance is improved.The main contents and contributions are as follow:(1)Analyze the basic theory and classical decoding algorithm of LDPC code,build a LDPC decoder based on deep learning with reference to Paper[40],and carry out simulation experiments on the python platform.After proving the feasibility of LDPC decoding algorithm based on deep learning,DRSN is used to instead of convolutional neural network(CNN).The value of loss funcation proved that the channel gain prediction ability of DRSN is better than that of CNN.Moreover,compared with the traditional decoding algorithm,the LDPC decoding algorithm based on DRSN has better decoding performance in Rayleigh channel.(2)Based on ieee802.11 p standard,DSRC(dedicated short range communication)vehicular channel is used and simulated to evaluate the feasibility of the proposed scheme.Simulation results show that the decoding performance of the original scheme is inferior to that of the traditional decoding algorithm.By redefining the log likelihood ratio(LLR)formula,the simulation results show that the decoding performance of the proposed algorithm is better under high SNR.