Decoding Of Polar Code Based On Neural Networks

With the development of the humans and the science,mobile communication is necessary for the society and the life.However,due to the interference and decline of mobile communication during transmission,errors inevitably occur.In order to enhance the ability of datum to resist all kinds of interference and ensure the reliability of the communication systems,we need the technology of error detection and correction coding.And the channel encoding means the error detection and correction coding for the digital signals to be transmitted in the channel.The research on the theory of channel encoding for digital signals to be transmitted in the channel began in Shannon's landmark paper "A Mathematical Theory of Communication".More than half a century in some practical channels such as the additive white gaussian noise channel,the central work of the academic research about the channel encoding is seeking a practical modulation and encoding scheme for approaching the channel capacity(also known as SHANNON).Polar code is a new scheme of the channel encoding first proposed by E.Arikan in 2008.According to the channel polarization,this new encoding scheme has been designed and proved to reach the capacity of the shannon channel.At the same time,as the only channel encoding technology in theory whose information transmission rate can reach the shannon limit,polar code not only has the advantages of the low complexity of decoding algorith,and also has a lower bit error rate about the long-codon transmission.Polar code is very outstanding and especially suitable for the communication systems with the high requirement for the information transmission.Based on the artificial neural network,this paper attempts to combine the neural networks and the decoding of polar code.We design a four-layer decoding network with two hidden layers about polar code based on the neural network.The network decodes the polar code with 128 length,and realizes the decoding function of polar code.Compared with the traditional Successive Cancellation algorithm of polar code,the performance has proven the feasibility and reliability about the decoding network of polar code.