Method for establishing SCMA codec model based on noise reduction auto-encoder

Abstract

The invention relates to a method for establishing an SCMA codec model based on a noise reduction auto-encoder, and the method comprises the steps: building the SCMA encoder based on the noise reduction auto-encoder and a full-connection neural network, and enabling the original input data of a user to be mapped into a code word; superposing and transmitting the code words of all the users on eachresource block, and superposing channel noise on the signal on each resource block; establishing an SCMA decoder based on a full-connection neural network at a receiving end, and decoding original input data of all users; training an SCMA codec model based on a noise reduction auto-encoder; and testing the BER performance of the SCMA codec model based on the noise reduction auto-encoder. Comparedwith a traditional SCMA system, the method has the advantages that the encoding and decoding complexity is reduced; compared with an existing SCMA system model based on deep learning, the method hasthe advantages that the bit error rate is further reduced; compared with the existing SCMA system model based on deep learning, the SCMA system model based on deep learning has a faster training convergence speed.

Description

technical field [0001] The invention relates to the technical fields of wireless communication and deep learning, in particular to a method for establishing a SCMA codec model based on a noise reduction autoencoder. Background technique [0002] Sparse code multiple access (SCMA) is a non-orthogonal multiple access method that can provide high spectral efficiency and large-scale connectivity that meet the requirements of 5G wireless communication systems. The encoding end and decoding end of the SCMA system constitute the main components of the system. Each user has its own dedicated codebook, and the encoded bits are directly mapped to a multi-dimensional codeword in the SCMA codebook, and non-orthogonal superposition is performed on the resources. Taking advantage of the sparseness of user codewords in the SCMA system, multi-user Detection, while the algorithm used in the current multi-user detection scheme is mainly the Message Passing Algorithm (MPA). [0003] However,...

AI Technical Summary

Problems solved by technology

But there are still problems: 1) Although the bit error rate of the D-SCMA system is lower than that of the traditional SCMA system, it is because the neural network has also learned a better codeword mapping, but its decoding performance is still inferior to that of the traditional SCMA system. MPA algorithm; 2) The slow convergence speed of the deep neural network used in the D-SCMA system results in longer training time

Images