A Deep Learning Channel Decoding Method Based on Alternating Direction Multiplier Method of Piecewise Linear Penalty Function

Abstract

The invention provides a deep learning channel decoding method based on an alternating direction multiplier method of a piecewise linear penalty function, which is mainly oriented to binary linear codes. The method includes the following steps: 1) Constructing the maximum likelihood optimization problem based on channel decoding; 2) Transforming constraints, designing an adjustable penalty function and introducing it into the objective function, transforming the maximum likelihood optimization problem of channel decoding into Decoding optimization problem with piecewise linear penalty function; 3) Introduce the alternating direction multiplier method to solve the above optimization problem, and obtain the alternating direction multiplier method channel decoder; 4) Construct deep learning according to the iterative form of the alternating direction multiplier method Network LADN‑P; 5) Obtain learning parameters through training; 6) Restore the deep learning network into a channel decoder, load learning parameters, and decode. The present invention designs an adjustable piecewise linear penalty function, and further improves the decoding performance with the help of deep learning.

Description

technical field [0001] The invention belongs to the field of wireless communication channel coding and decoding, and designs a deep learning channel decoding method based on an alternating direction multiplier method of a piecewise linear penalty function. Background technique [0002] Channel decoding is how to judge the received symbol message. In a general information transmission system, the message received by the sink is not necessarily the same as the message sent by the source, and the sink needs to know which source message is sent by the source at this time, so the message received by the sink needs to be based on a certain A rule judges that it corresponds to one of the source symbol message sets. Linear programming (LP) decoders are based on a linear relaxation of the original maximum likelihood decoding problem and are a popular decoding technique for binary linear codes. Since linear program decoders have strong guarantees on decoding performance in theory, t...

AI Technical Summary

Problems solved by technology

However, compared with the classical Belief Propagation (BP) decoder, the LP decoder has higher computational complexity and lower error correction performance in the low signal-to-noise ratio (SNR) region

Images