	US 12,452,113 B1
	Transceiver decoding method and system based on protograph differential chaos shift keying
Yi Fang, Guangdong (CN); Yuyi He, Guangdong (CN); Liang Lv, Guangdong (CN); Liuguo Yin, Guangdong (CN); and Pingping Chen, Guangdong (CN)
Assigned to Guangdong University of Technology, Guangzhou (CN)
Filed by Guangdong University of Technology, Guangdong (CN)
Filed on Jun. 11, 2025, as Appl. No. 19/235,482.
Claims priority of application No. 202410773836.9 (CN), filed on Jun. 17, 2024.
Int. Cl. H04L 27/00 (2006.01); H04L 1/00 (2006.01)

CPC H04L 27/001 (2013.01) [H04L 1/0057 (2013.01)]

9 Claims

1. A transceiver decoding method based on protograph differential chaos shift keying, a transceiver comprising a protograph differential chaos shift keying transmitter, a wireless channel, a target extrinsic information-aided network and an a-priori calculation decoder, wherein the method comprises:

acquiring an information bit sequence, performing a coded modulation on the information bit sequence via the protograph differential chaos shift keying transmitter, and outputting a plurality of modulated symbols;

inputting each modulated symbol into the wireless channel for channel interference, and outputting a received symbol corresponding to each modulated symbol;

using the target extrinsic information-aided network to determine a target a-posteriori probability vector corresponding to each received symbol according to each received symbol and a preset initial a-posteriori probability vector corresponding to each received symbol;

inputting each target a-posteriori probability vector into the a-priori calculation decoder for decoding, and outputting an initial decoded bit sequence; and

determining a target decoded bit sequence according to the initial decoded bit sequence and a preset check matrix; and

the target extrinsic information-aided network comprises a convolutional neural network-long short-term memory block, a fully connected block, and a decision block; and the step of using the target extrinsic information-aided network to determine a target a-posteriori probability vector corresponding to each received symbol according to each received symbol and a preset initial a-posteriori probability vector corresponding to each received symbol comprises:

using the convolutional neural network-long short-term memory block to perform one-dimensional feature extraction on each received symbol, and generating a first feature vector corresponding to each received symbol;

using the fully connected block to perform vector feature extraction on the preset initial a-posteriori probability vector corresponding to each received symbol, and determining a second feature vector corresponding to each preset initial a-posteriori probability vector; and

inputting each first feature vector and each second feature vector into the decision block for feature concatenating separately, and outputting the target a-posteriori probability vector corresponding to each received symbol.