A confidence propagation ldpc decoding method based on deep learning

Abstract

The present invention provides a deep learning-based confidence propagation LDPC decoding method, comprising the following steps: the first step is to establish a training sample set for LDPC decoding; the second step is to establish a deep learning decoding model; the third step is to determine The input training set of the deep learning decoding model; the fourth step is to determine the activation function of the hidden layer in the deep learning decoding model; the fifth step is to use the input training set of the third step to adopt the training method of batch gradient descent The code model is trained; the sixth step is to verify the trained deep learning decoding model, make a hard judgment on the output result of the verification and adjust the weight w accordingly, and determine the parameters of the deep learning decoding model; the seventh step is to Input the LDPC code to be decoded into the deep learning decoding model after the parameters are determined in the sixth step to decode, and complete the LDPC decoding. The invention can decode in parallel, reduce the number of iterations and complexity of decoding, and restore the data at the sending end from the sequence containing noise and interference.

Description

technical field [0001] The present invention relates to the technical field of electronic communication, and more specifically, relates to a deep learning-based belief propagation LDPC decoding method. Background technique [0002] Low Density Parity Check Code (LDPC, Low Density Parity Check Code) is a linear block error correction code with low decoding complexity and excellent performance. Based on early research, it was found that when the LDPC code length is long enough, the bit error rate can be very close to the Shannon limit, and even when it reaches a certain code length, it has an error correction capability closer to the Shannon limit than the Turbo code. Turbo codes have gained a dominant position in the channel coding schemes of the third generation mobile communication. Therefore, LDPC codes have been widely used in deep space communication, optical fiber communication, satellite digital video and audio broadcasting and other fields, and have been adopted by v...

AI Technical Summary

Problems solved by technology

[0004] The traditional LDPC maximum likelihood decoding method is very difficult to implement, and the calculation complexity is too large, especially when the code length is long, for (n, k) LDPC (n is the code word length, k is the information bit length) , with the increase of k, the codeword types 2^k show an explosive exponential growth, so there will be great difficulties in the classification of data sets

Images