A LDPC layered decoding method based on dual-path LLR updating

通过在LDPC分层译码器中采用双路LLR更新方法,利用重叠和非重叠子矩阵缓冲区并行计算LLR,解决了数据更新冲突问题,提高了译码器的吞吐率和性能,避免了计算 High complexity problem.

CN114499540BActive Publication Date: 2024-11-26SHANGHAI TIANYU OPTICAL COMM TECH CO LTD
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202111550108.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-12-17
Publication Date
2024-11-26
Estimated Expiration
2041-12-17

AI Technical Summary

Technical Problem

The existing LDPC hierarchical decoder has data update conflicts during the LLR update process in the pipelined form, resulting in a reduced throughput of the decoder or a high computational complexity, and it is impossible to effectively adjust the LLR update order to meet the needs of the next layer of submatrix.

Method used

Using a dual-channel LLR update method, the submatrix is ​​preprocessed and cached, and the LLR is updated separately using overlapping and non-overlapping submatrix buffers, and the two adders are computed in parallel. The update result of the overlapping submatrix is ​​directly bypassed. Enter the next layer to avoid data update conflicts.

Benefits of technology

Without increasing the pipeline stages, the number of data update conflicts is reduced, the throughput of the LDPC decoder is improved, and performance losses are avoided, ensuring efficient data processing.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114499540B_ABST
    Figure CN114499540B_ABST
Patent Text Reader

Abstract

The present invention discloses an LDPC layered decoding method based on dual-path LLR updating and a decoder that uses three buffers to cache variable node information transmitted to a check node operation unit by a variable node operation unit for updating LLR, and after calculating the check node information, uses a dual-path adder to update the LLR of overlapping sub-matrices and non-overlapping sub-matrices respectively. The LDPC layered decoding method based on dual-path LLR updating of the present invention can ensure that data update conflict problems do not occur when the number of pipeline stages is not greater than the number of non-overlapping sub-matrices of the layer, and when the number of pipeline stages is greater than the number of non-overlapping sub-matrices of the layer, the number of data update conflicts that occur is reduced, and the performance loss of the LDPC decoder can be ignored.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of communication coding and decoding, and in particular to an LDPC layered decoding method based on dual-path LLR updating. Background Art

[0002] As a forward error correction code, LDPC is widely used in wireless communication, wired communication, aerospace, data storage and other fields. The throughput of the LDPC decoder determines the upper limit of the data processing rate in the communication system. When the code length and code rate are determined and the layers are layered in units of sub-matrix order, the only way to improve the throughput of the LDPC decoder is to increase the clock operating frequency and reduce the pipeline time. Due to the layered decoding algorithm, each clock cycle processes the data volume of a sub-matrix, so the working clock cycle is fixed. Increasing the clock frequency can be achieved by inserting a pipeline on the critical path where the combinational logic delay is too long.

[0003] In the LDPC layered decoding process, it is common for non-zero sub-matrices to overlap between layers. In the layered decoding algorithm, the log-likelihood ratio (LLR) of a sub-matrix of a certain layer should be updated earlier than the reading of the LLR of the sub-matrix of the next layer. In order to increase the throughput of the decoder and improve the clock operating frequency of the decoder, a certain number of pipeline stages have to be inserted, resulting in data update conflicts. The existing method of further layering the layers in units of sub-matrix orders can reduce the number of data update conflicts, but reduces the throughput of the LDCP decoder. The existing method of adjusting the processing order of sub-matrices within a layer and between layers through optimization algorithms such as genetic algorithms and ant colony algorithms can reduce the number of data update conflicts, but the calculation complexity of optimizing the processing order of sub-matrices is high and time-consuming. In the existing method, there is only one data path for updating LLR, which makes LLR only updated according to the processing order of the sub-matrix of this layer, and the update order cannot be adjusted to meet the needs of the processing of the sub-matrix of the next layer. At the same time, LLR must be stored in the LLR memory before being read out, which increases the delay of LLR update. Summary of the invention

[0004] In view of the above-mentioned defects of the prior art, the technical problem to be solved by the present invention is that the existing LDPC layered decoder has data update conflicts in the process of updating LLR in a pipeline form, and in the process of reducing the number of data update conflicts, the decoder throughput is reduced or the calculation complexity is high, the time consumption is long, and other problems are caused. The present invention provides an LDPC layered decoding method based on dual-path LLR update, which can ensure that the data update conflict problem does not occur when the number of pipeline stages is not greater than the number of non-overlapping sub-matrices of the layer, and when the number of pipeline stages is greater than the number of non-overlapping sub-matrices of the layer, the number of data update conflicts is reduced, and the performance loss of the LDPC decoder can be ignored.

[0005] To achieve the above object, the present invention provides an LDPC layered decoding method based on dual-path LLR updating, comprising the following steps:

[0006] Before decoding, the processing order of the sub-matrices in the check matrix is ​​preprocessed;

[0007] The variable node information calculated by the variable node processing unit is divided into two categories according to whether both the previous layer and the next layer need to be processed. The variable node information corresponding to the submatrix that needs to be processed in both the current layer and the next layer is stored in the overlapping submatrix buffer or the adjustment order submatrix buffer, and the rest is stored in the non-overlapping submatrix buffer;

[0008] After the check node processing unit calculates the check node information in the current layer, it reads data sequentially from the buffers of the overlapping submatrix and the non-overlapping submatrix, updates them in parallel through two adders, and calculates the latest LLR;

[0009] The LLR updated by the non-overlapping sub-matrix is ​​stored in the LLR memory, and if there is no data update conflict in the overlapping sub-matrix, the updated LLR is directly bypassed and input into the barrel shifter to directly participate in the operation of the next layer; thus completing a round of decoding.

[0010] Furthermore, the processing order of the submatrices of the current layer is determined by the overlapping submatrices and non-overlapping submatrices of the previous layer. The non-overlapping submatrices are processed first, and then the overlapping submatrices. Both the non-overlapping submatrices and the overlapping submatrices are processed in ascending order according to the submatrix subscripts.

[0011] Furthermore, before decoding, the sub-matrix in the check matrix is ​​preprocessed, specifically including:

[0012] The processing order of the sub-matrices in the check matrix is ​​pre-sorted, the order of the sub-matrices is fixed, and each round of iterative decoding is also set to be continuous. During the first iteration, each layer of the sub-matrix is ​​processed in turn. After the last layer of the sub-matrix is ​​processed in the first iteration, the first layer of the sub-matrix of the second iteration is processed. The processing order of the first layer of the sub-matrix of the next iteration is determined by the last layer of the sub-matrix of the previous iteration, and the classification of overlapping sub-matrices and non-overlapping sub-matrices is also determined by the last layer of the sub-matrix.

[0013] Furthermore, after preprocessing, the data that needs to be processed in both the current layer and the next layer is stored in the overlapping sub-matrix buffer, and the rest is stored in the non-overlapping sub-matrix buffer, which also includes an adjustment order sub-matrix buffer, which is used to store the sub-matrix that is a non-overlapping sub-matrix in the previous layer and an overlapping sub-matrix in the current layer.

[0014] Furthermore, two adders run in parallel without interfering with each other, including a first adder and a second adder, wherein after the check node operation unit calculates the check node information, the non-overlapping sub-matrix uses the first adder to update the LLR separately, and the overlapping sub-matrix selects to read the sub-matrix information from the non-overlapping sub-matrix buffer and the adjusted order sub-matrix buffer, and uses the second adder to update the LLR separately.

[0015] Furthermore, during the decoding process, if there is no data update conflict in the overlapping sub-matrix, the overlapping sub-matrix is ​​buffered in the overlapping sub-matrix buffer or the adjusted order sub-matrix buffer. If there is a data update conflict, the overlapping sub-matrix with the data update conflict is not stored in any buffer, and the remaining overlapping sub-matrices are stored in the overlapping sub-matrix buffer or the adjusted order sub-matrix buffer according to the processing order.

[0016] Furthermore, when there is no data update conflict in the overlapping sub-matrices in the current layer, the LLR is updated through the adder, and the updated LLR is bypassed and input into the barrel shifter. If there is a data update conflict in an overlapping sub-matrix in the current layer, the LLR of the sub-matrix is ​​not updated, and the LLR is read from the LLR memory during the decoding process of the next layer.

[0017] Furthermore, after the LLR of the submatrix is ​​calculated by the barrel shifter and the variable node processing unit, variable node information is obtained.

[0018] Furthermore, before decoding, the number of pipeline stages to be inserted is determined, and a decoder is designed according to the number of pipeline stages to be inserted.

[0019] Another preferred embodiment of the present invention provides a decoder for implementing an LDPC layered decoding method based on dual-path LLR update, comprising a barrel shifter, a variable node processing unit, a non-overlapping submatrix buffer, an overlapping submatrix buffer, an adjusted order submatrix buffer, a check node operation unit, a first adder and a second adder. After the LLR of the submatrix is ​​calculated by the barrel shifter and the variable node processing unit, variable node information is obtained. After that, if the submatrix does not continue to participate in the operation in the next layer, it is cached in the non-overlapping submatrix buffer. If the submatrix continues to participate in the operation in the next layer and the submatrix index is greater than the subsequent submatrix index of the same type, it is cached in the adjusted order submatrix buffer. The remaining submatrices that need to participate in the operation processing in the next layer are cached in the overlapping submatrix buffer in sequence. After the check node operation unit calculates the minimum value and the second minimum value of the current layer, the overlapping submatrix and the non-overlapping submatrix are updated respectively by the two-path adders.

[0020] Technical Effects

[0021] The present invention proposes an LDPC layered decoding method and a decoder based on dual-path LLR updating, which adopts the design that sub-matrices involved in operation are divided into overlapping and non-overlapping and cached in two buffers, two-path adders are used to update LLR for overlapping and non-overlapping sub-matrices respectively, and the overlapping sub-matrices are bypassed and input to the calculation module of the next layer after being updated, thereby overcoming the data update conflict problem that frequently occurs when pursuing the high throughput of the LDPC layered decoder and inserting a pipeline for improving the timing, avoiding the reduction of the decoding performance of the LDPC decoder, and ensuring the high throughput of the LDPC layered decoder.

[0022] The concept, specific structure and technical effects of the present invention will be further described below in conjunction with the accompanying drawings to fully understand the purpose, characteristics and effects of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a schematic diagram of the structure of a traditional LDPC layered decoder;

[0024] Figure 2 It is a structural schematic diagram of an LDPC layered decoder based on dual-path LLR updating according to a preferred embodiment of the present invention;

[0025] Figure 3 It is a schematic diagram comparing the LDPC layered decoder based on dual-path adder update of the present invention and the traditional LDPC layered decoder. DETAILED DESCRIPTION

[0026] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0027] In the following description, specific details such as specific internal procedures and techniques are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present invention. However, it should be clear to those skilled in the art that the present invention may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present invention.

[0028] The structural diagram of the traditional LDPC layered decoder is as follows Figure 1 As shown. After the sub-matrix is ​​processed by the barrel shifter and the variable node calculation unit, the calculated variable node information is cached in the sub-matrix buffer. After waiting for the corresponding check node information to be calculated, the latest LLR value is updated by the adder and finally saved in the LLR memory. When improving the throughput of the LDPC decoder, a large number of pipeline stages will be inserted on the critical timing path. This structure will cause the LLR of the current layer sub-matrix to be not updated yet, and the same sub-matrix of the next layer needs to participate in decoding. This is called data update conflict, which causes the performance of the LDPC layered decoder to degrade.

[0029] like Figure 2As shown, the present invention provides a decoder for implementing an LDPC layered decoding method based on dual-path LLR update, including a barrel shifter, a variable node processing unit, a non-overlapping sub-matrix buffer, an overlapping sub-matrix buffer, an adjusted order sub-matrix buffer, a check node operation unit, a first-path adder and a second-path adder. After the LLR of the sub-matrix is ​​calculated by the barrel shifter and the variable node processing unit, variable node information is obtained. After that, if the sub-matrix does not continue to participate in the operation in the next layer, it is cached in the non-overlapping sub-matrix buffer. If the sub-matrix continues to participate in the operation in the next layer and the sub-matrix index is greater than the subsequent sub-matrix index of the same type, it is cached in the adjusted order sub-matrix buffer. The remaining sub-matrices that need to participate in the operation processing in the next layer are cached in the overlapping sub-matrix buffer in order. After the check node operation unit calculates the minimum value and the second minimum value of the current layer, the overlapping sub-matrix and the non-overlapping sub-matrix are respectively updated by the two-path adders. Different from the traditional LDPC decoder, the sub-matrices to be updated are cached into two queues according to whether they overlap or not. The corresponding check node information of the overlapping sub-matrices and non-overlapping sub-matrices is calculated respectively through two check node information calculation modules, and the LLR is updated through two adders, which shortens the cycle of updating the LLR in the sub-matrix, so that the overlapping sub-matrix can obtain the updated LLR before entering the next layer of update.

[0030] Another preferred embodiment of the present invention provides an LDPC layered decoding method based on dual-path LLR updating, comprising the following steps:

[0031] Step 100, before decoding, preprocess the processing order of the sub-matrices in the check matrix; specifically, pre-sort the processing order of the sub-matrices in the check matrix, the order of the sub-matrices is fixed, and each round of iterative decoding is also set to be continuous. During the first iteration, each layer of sub-matrix is ​​processed in turn. After the last layer of sub-matrix is ​​processed during the first iteration, the first layer of sub-matrix of the second iteration is processed. The processing order of the first layer of sub-matrix of the next iteration is determined by the last layer of sub-matrix of the previous iteration, and the classification of overlapping sub-matrices and non-overlapping sub-matrices is also determined by the last layer of sub-matrix.

[0032] Step 200, the variable node information calculated by the variable node processing unit is divided into two categories according to whether both the current layer and the next layer need to be processed. The variable node information corresponding to the submatrix that needs to be processed by both the current layer and the next layer is stored in the overlapping submatrix buffer or the adjustment order submatrix buffer, and the rest is stored in the non-overlapping submatrix buffer; in addition, it also includes an adjustment order buffer, which is used to store the submatrix that is a non-overlapping submatrix in the previous layer and an overlapping submatrix in the current layer. The LLR of the submatrix is ​​calculated by the barrel shifter and the variable node processing unit to obtain the variable node information. Among them, the current layer and the next layer are not limited to the same iteration. There is no pause time between the processing of the last layer in the current iteration and the first layer in the next iteration, so the overlap of the last layer in the current iteration and the first layer in the next iteration needs to be considered in decoding.

[0033] Step 300, after the check node processing unit calculates the check node information in the current layer, data is sequentially read from the buffers of the overlapping submatrix and the non-overlapping submatrix respectively, and updated in parallel by two adders to calculate the latest LLR; the two adders run in parallel without interfering with each other, and include a first adder and a second adder, wherein, after the check node operation unit calculates the check node information, the non-overlapping submatrix uses the first adder to update the LLR separately, and the overlapping submatrix selects to read submatrix information from the non-overlapping submatrix buffer and the adjusted order submatrix buffer, and uses the second adder to update the LLR separately.

[0034] Step 400, the LLR updated by the non-overlapping sub-matrix is ​​stored in the LLR memory, and if there is no data update conflict in the overlapping sub-matrix, the updated LLR is directly bypassed and input into the barrel shifter to directly participate in the operation of the next layer; thus completing a round of decoding.

[0035] Among them, the processing order of the submatrices of the current layer is determined by the overlapping submatrices and non-overlapping submatrices of the previous layer. The non-overlapping submatrices are processed first, and then the overlapping submatrices. Both the non-overlapping submatrices and the overlapping submatrices are processed in ascending order according to the submatrix subscripts.

[0036] During the decoding process, if there is no data update conflict in the overlapping sub-matrix, the overlapping sub-matrix is ​​buffered in the overlapping sub-matrix buffer or the adjustment order sub-matrix buffer. If there is a data update conflict, the overlapping sub-matrix with the data update conflict is not stored in any buffer, and the remaining overlapping sub-matrices are stored in the overlapping sub-matrix buffer or the adjustment order sub-matrix buffer according to the processing order.

[0037] When there is no data update conflict in the overlapping sub-matrix in the current layer, the LLR is updated through the adder, and the updated LLR is bypassed and input into the barrel shifter. If there is a data update conflict in an overlapping sub-matrix in the current layer, the LLR of the sub-matrix is ​​not updated, and the LLR is read from the LLR memory during the decoding process of the next layer.

[0038] Before decoding, it also includes determining the number of pipeline stages that need to be inserted, and designing the decoder based on the number of pipeline stages that need to be inserted.

[0039] The following is a specific example to illustrate the specific steps of an LDPC layered decoding method based on dual-path LLR updating provided by an embodiment of the present invention. Figure 3 shown.

[0040] The LDPC basic matrix in the example is 3 rows and 8 columns, as shown in formula (1)

[0041]

[0042] In the implementation of the LDPC layered decoding method based on dual-path LLR update, the pipeline stage is set to 2. Since the number of non-overlapping sub-matrices in each layer of the matrix is ​​not less than the pipeline stage, there is no update conflict problem.

[0043] Before decoding, the processing order of the submatrices needs to be preprocessed. According to the above regulations, the processing order of the 0th row is determined by whether the submatrices of the 0th row and the 2nd row overlap. Since the 0th and 4th submatrices overlap, the 1st, 5th, and 6th submatrices are processed first, and the processing order is 1, 5, 6, 0, 4. The processing order of the 1st row is determined by whether the submatrices of the 0th row and the 1st row overlap. Since the 1st, 4th, and 5th submatrices overlap, the 3rd and 7th submatrices are processed first, and the processing order is 3, 7, 1, 4, 5. The processing order of the 2nd row is determined by whether the submatrices of the 1st row and the 2nd row overlap. Since the 3rd and 4th submatrices overlap, the 0th, 2nd, and 8th submatrices are processed first, and the processing order is 0, 2, 8, 3, 4.

[0044] The variable node information buffering after the variable node processing unit is as follows: in the 0th layer, the 6th and 0th sub-matrices are buffered in sequence to the non-overlapping sub-matrix buffer, the variable node information corresponding to the 1st and 4th sub-matrices are buffered in sequence to the overlapping sub-matrix buffer, and the variable node information corresponding to the 5th sub-matrix is ​​buffered to the adjusted order sub-matrix buffer. In the 1st layer, the 7th, 1st, and 5th sub-matrices are buffered in sequence to the non-overlapping sub-matrix buffer, and the variable node information corresponding to the 3rd and 4th sub-matrices are buffered in sequence to the overlapping sub-matrix buffer. In the 2nd layer, the 2nd, 8th, and 3rd sub-matrices are buffered in sequence to the non-overlapping sub-matrix buffer, and the variable node information corresponding to the 0th and 4th sub-matrices are buffered in sequence to the overlapping sub-matrix buffer.

[0045] After calculating the check node information, the two-way LLR update of the overlapping submatrix and the non-overlapping submatrix is ​​as follows: in the 0th layer, the 6th and 0th submatrices read the corresponding variable node information from the non-overlapping submatrix buffer in turn, add the check node information, and then update to obtain the LLR, and store the LLR back in the LLR memory; the 1st, 4th, and 5th submatrices read the corresponding variable node information from the overlapping submatrix buffer and the adjustment order submatrix buffer in turn, add the check node information, and then update to obtain the LLR, which is bypassed and input into the barrel shifter. In the 1st layer, the 7th, 1st, and 5th submatrices read the corresponding variable node information from the non-overlapping submatrix buffer in turn, add the check node information, and then update to obtain the LLR, and store the LLR back in the LLR memory; the 3rd and 4th submatrices read the corresponding variable node information from the overlapping submatrix buffer in turn, add the check node information, and then update to obtain the LLR, which is bypassed and input into the barrel shifter. In the second layer, the 2nd, 8th, and 3rd sub-matrices read the corresponding variable node information from the non-overlapping sub-matrix buffer in turn, add the check node information, and then update the LLR, and store the LLR back in the LLR memory; the 0th and 4th sub-matrices read the corresponding variable node information from the overlapping sub-matrix buffer in turn, add the check node information, and then update the LLR, which is bypassed and input into the barrel shifter.

[0046] After the LLR update is completed, the overlapping sub-matrices are bypassed and input into the barrel shifter as follows: In the 0th layer, the 1st, 4th, and 5th sub-matrices also need to participate in decoding in the 1st layer, so their corresponding updated LLRs are bypassed and input into the barrel shifter and are no longer read from the LLR memory. In the 1st layer, the 3rd and 4th sub-matrices also need to participate in decoding in the 2nd layer, so their corresponding updated LLRs are bypassed and input into the barrel shifter and are no longer read from the LLR memory. In the 2nd layer, the 0th and 4th sub-matrices also need to participate in decoding in the 0th layer in the next iteration, so their corresponding updated LLRs are bypassed and input into the barrel shifter and are no longer read from the LLR memory.

[0047] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A LDPC layered decoding method based on dual-path LLR updating, characterized in that: The following steps are involved: Before decoding, the processing order of the sub-matrices in the check matrix is ​​preprocessed; The variable node information calculated by the variable node processing unit is divided into two categories according to whether both the current layer and the next layer need to be processed. The variable node information corresponding to the submatrix that needs to be processed in both the current layer and the next layer is stored in the overlapping submatrix buffer or the adjustment order submatrix buffer, and the rest is stored in the non-overlapping submatrix buffer; After the check node processing unit calculates the check node information in the current layer, it sequentially reads data from the buffers of the overlapping submatrix and the non-overlapping submatrix, updates them in parallel through two adders, and calculates the latest LLR; The LLR updated by the non-overlapping sub-matrix is ​​stored in the LLR memory, and if there is no data update conflict in the overlapping sub-matrix, the updated LLR is directly bypassed and input into the barrel shifter to directly participate in the operation of the next layer; thus completing a round of decoding.

2. The LDPC layered decoding method based on dual-path LLR updating according to claim 1, characterized in that: The processing order of the submatrices of the current layer is determined by the overlapping submatrices and non-overlapping submatrices of the previous layer. The non-overlapping submatrices are processed first, and then the overlapping submatrices. The non-overlapping submatrices and the overlapping submatrices are processed in ascending order according to the submatrix subscripts.

3. The LDPC layered decoding method based on dual-path LLR updating according to claim 1, characterized in that: Before decoding, the sub-matrix in the check matrix is ​​preprocessed, including: The processing order of the submatrices in the check matrix is ​​pre-sorted, the order of the submatrices is fixed, and each round of iterative decoding is also set to be continuous. During the first iteration, each layer of submatrix is ​​processed in turn. After the last layer of submatrix is ​​processed during the first iteration, the first layer of submatrix of the second iteration is processed. The processing order of the first layer of submatrix of the next iteration is determined by the last layer of submatrix of the previous iteration, and the classification of overlapping submatrices and non-overlapping submatrices is also determined by the last layer of submatrix.

4. The LDPC layered decoding method based on dual-path LLR updating according to claim 1, characterized in that: After preprocessing, the data that needs to be processed in both the current layer and the next layer is stored in the overlapping sub-matrix buffer, and the rest is stored in the non-overlapping sub-matrix buffer, which also includes an adjustment order sub-matrix buffer, which is used to store the sub-matrix of the previous layer that is a non-overlapping sub-matrix and the current layer that is an overlapping sub-matrix.

5. The LDPC layered decoding method based on dual-path LLR updating as claimed in claim 4, characterized in that: The two adders run in parallel without interfering with each other, and include a first adder and a second adder. After the check node operation unit calculates the check node information, the non-overlapping sub-matrix uses the first adder to update the LLR separately, and the overlapping sub-matrix selects to read the sub-matrix information from the non-overlapping sub-matrix buffer and the adjusted order sub-matrix buffer, and uses the second adder to update the LLR separately.

6. The LDPC layered decoding method based on dual-path LLR updating according to claim 1, characterized in that: During the decoding process, if there is no data update conflict in the overlapping sub-matrix, the overlapping sub-matrix is ​​buffered in the overlapping sub-matrix buffer or the adjusted order sub-matrix buffer. If there is a data update conflict, the overlapping sub-matrix with the data update conflict is not updated, so it is not stored in any buffer. The remaining overlapping sub-matrices are stored in the overlapping sub-matrix buffer or the adjusted order sub-matrix buffer according to the processing order.

7. The LDPC layered decoding method based on dual-path LLR updating as claimed in claim 6, characterized in that: When there is no data update conflict in the overlapping sub-matrix in the current layer, the LLR is updated through the adder, and the updated LLR is bypassed and input into the barrel shifter. If there is a data update conflict in an overlapping sub-matrix in the current layer, the LLR of the sub-matrix is ​​not updated, and the LLR is read from the LLR memory during the decoding process of the next layer.

8. The LDPC layered decoding method based on dual-path LLR updating as claimed in claim 1, characterized in that: After the LLR of the submatrix is ​​calculated by the barrel shifter and the variable node processing unit, the variable node information is obtained.

9. The LDPC layered decoding method based on dual-path LLR updating according to claim 1, characterized in that: Before decoding, it also includes determining the number of pipeline stages that need to be inserted, and designing the decoder based on the number of pipeline stages that need to be inserted.

10. A decoder for implementing the LDPC layered decoding method based on dual-path LLR updating as claimed in any one of claims 1 to 9, characterized in that: It includes a barrel shifter, a variable node processing unit, a non-overlapping sub-matrix buffer, an overlapping sub-matrix buffer, an adjusted order sub-matrix buffer, a check node operation unit, a first adder and a second adder. After the LLR of the sub-matrix is ​​calculated by the barrel shifter and the variable node processing unit, the variable node information is obtained. After that, if the sub-matrix does not continue to participate in the operation in the next layer, it is cached in the non-overlapping sub-matrix buffer. If the sub-matrix continues to participate in the operation in the next layer and the sub-matrix index is greater than the sub-matrix index of the following sub-matrix of the same type, it is cached in the adjusted order sub-matrix buffer. The remaining sub-matrices that need to participate in the operation processing in the next layer are cached in the overlapping sub-matrix buffer in order. After the check node operation unit calculates the minimum value and the second minimum value of the current layer, the overlapping sub-matrix and the non-overlapping sub-matrix are updated respectively through the two-way adders.

Citation Information

Patent Citations

  • LDPC (low density parity check) decoder and decoding method based on layer decoding processing

    CN102394661A

  • Quasi-cyclic low-density parity-check (QC-LDPC) decoder, hierarchical decoding method, storage equipment and communication module

    CN108649963A

  • Multi-mode LDPC decoder applied to CCSDS satellite deep space communication

    CN112187286A

  • Low Density Parity Check Layer Decoder For Codes With Overlapped Circulants

    US20140101510A1