5 results about "Tanner graph" patented technology

A method for constructing a permutation matrix or a check matrix, a processing device, a storage medium and a coding method. The method for constructing the permutation matrix comprises: obtaining a base matrix for the permutation matrix; and performing lifting on the base matrix to obtain the permutation matrix. The lifting on the base matrix to obtain the permutation matrix comprises: obtaining a modulus graph of the base matrix, obtaining each macro ring in the modulus graph, and for each macro ring in the modulus graph, determining the size of a short ring corresponding to the macro ring in a Tanner graph of a check matrix corresponding to the permutation matrix from an equivalent cycle value ECS of the macro ring, and determining whether at least one cycle value in the macro ring needs to be set according to the size of the short ring. The method can more efficiently obtain a required permutation matrix.

The invention relates to a short code length QC-LDPC (Quasi-Cyclic Low-Density Parity-Check) code construction method based on a Stanley sequence and a genetic algorithm. According to the method, a Stanley sequence is used for forming first two rows of an index matrix meeting the condition that no four rings exist, then a genetic algorithm is used for searching a third row element which enables the index matrix to meet the requirements that the ring length is at least 8 and the number of trap sets is small, and therefore the index matrix with the girth length being at least 8 is obtained, the genetic algorithm can guarantee that the number of the trap sets is minimum, and the number of trap sets is small. And finally, the index matrix is expanded to obtain a parity check matrix of the index matrix, so that four rings and six rings do not exist in a Tanner graph of the check matrix, the number of trap sets is small, the storage space required by code words is small, and hardware implementation is easy. Simulation shows that compared with an HS-QC-LDPC code, an MT-QC-LDPC code and an SES-QC-LDPC code which have the same code rate and code length, the SS-QC-LDPC code constructed by the method has the advantages that the net coding gain is improved to a certain extent, and the constructed SS-QC-LDPC code has better waterfall region performance in a high signal-to-noise ratio region and has no obvious error leveling phenomenon.

Various aspects of the present disclosure relate to enhanced encoder designs (and associated signaling), such as enhanced encoder structures for quasi-cyclic low-density parity-check (QC-LDPC) encoders. The new encoders can achieve a larger girth (e.g., length of the shortest cycle of a Tanner graph) by using permutation matrices without increasing the complexity of hardware implementations of the encoders. Thus, the enhanced encoders can facilitate use of short or limited length codes for wireless communication data channels, among other benefits.

The application provides an LDPC optimization method based on dynamic parameter adaptive coding and a Tanner graph neural network. The method comprises the following steps: collecting a data set through a channel sensing module; the data set comprises channel state information, bit error rate statistics and service QoS requirement information; preprocessing the collected data set; dividing the preprocessed data set according to a preset ratio; constructing an LSTM prediction model; training and optimizing the LSTM prediction model based on a loss function and using a training set, and saving an optimal model; inputting a test set into the optimal model to predict the coding parameters of channel transmission, and obtaining a prediction result of the adaptive coding parameters of channel transmission; building an LDPC encoder and a Tanner-GNN decoder; building a joint optimization controller; the joint optimization controller is used for dynamically adjusting the coding parameters through the LSTM prediction model and receiving the bit error rate BER counted by the Tanner-GNN decoder. The application breaks through the technical limitation of traditional coding and decoding separation optimization, and significantly improves the performance of the LDPC code in a complex channel environment.

This invention relates to a Normalized Minimum Offset Sum (NOMS) decoding method for LDPC codes based on optimized grouped neural networks, belonging to the field of wireless communication channel coding technology. This method utilizes the topological relationship between the LDPC code base graph and the Tanner graph to transform the Tanner graph into a non-fully connected grouped neural network. It groups edges based on the base graph type and achieves parameter sharing using edge bundles. A greedy hierarchical training algorithm is used to optimize the normalization factor and offset factor round by round, obtaining dynamically adaptive parameters that adapt to each iteration stage. The optimized factors are substituted into the decoding update formula to replace the traditional fixed factors and complete the normalized minimum offset sum decoding. This invention can suppress error amplification during iteration, avoid error leveling and decoding deadlock, improve decoding accuracy while reducing computational complexity and parameter size, exhibit strong generalization ability, and can adapt to LDPC codes with different code lengths and rates under the same base graph. It also features low latency, high robustness, and engineering practicality.