An LDPC code sequential decoding method and device based on an unstable variable node maximum remaining degree silent mechanism

Abstract

The present invention provides an LDPC code timing decoding method and apparatus based on an unstable variable node maximum residual degree silent mechanism. The method includes: initializing messages and residual degrees, locating unstable variable nodes, selecting the maximum residual degree edges of unstable variable nodes, silencing the information update of these edges, transmitting the input information of the silent edges, transmitting the output information of the silent edges, updating the corresponding residual degrees, updating the set of unstable variable nodes, etc. Compared with the traditional IVC-RBP decoding method, the INP-VCRBP decoding method significantly improves the decoding performance while maintaining a low complexity, especially showing more prominent performance under low signal-to-noise ratio and low bit error rate requirements. Experimental results show that this decoding method can maintain good performance under different code lengths, has good practical application value, and can promote the application and development of LDPC codes in the fields of next-generation wireless communication, future geosynchronous satellite communication, etc.

Description

Technical Field

[0001] The present invention belongs to the fields of electronics, communication, and information engineering, and particularly relates to a timing decoding method and device for LDPC codes based on the maximum residual degree silent mechanism of unstable variable nodes. Background Art

[0002] The IVC-RBP (Improved Variable Node Classification Residual Belief Propagation) algorithm, an improved variable node classification residual belief propagation algorithm, is a dynamic decoding algorithm further improved on the basis of the RBP algorithm. It not only considers the residual of the message but also introduces the concept of unstable variable nodes, thereby optimizing the message update strategy. In the decoding process of LDPC codes, the message passing between variable nodes and check nodes is crucial. The IVC-RBP algorithm classifies variable nodes into two categories: unstable variable nodes and stable nodes. Unstable variable nodes refer to those nodes with large message changes during the iteration process, while stable nodes are those with small message changes. This classification method enables the algorithm to more accurately identify which nodes' message updates are more critical to the decoding process. When selecting the maximum residual edge, the IVC-RBP algorithm preferentially updates the edge related to the unstable variable node. If the residuals of all edges related to unstable variable nodes are small, the algorithm will consider the edges related to stable nodes. This strategy not only improves the decoding efficiency to a certain extent but also further enhances the error correction ability of the algorithm. However, the decoding complexity of the IVC-RBP algorithm is relatively high, and its performance needs to be further improved when dealing with complex communication channels and strong noise environments. Summary of the Invention

[0003] Object of the Invention: The technical problem to be solved by the present invention is to provide a timing decoding method and device for LDPC codes based on the maximum residual degree silent mechanism of unstable variable nodes (INP-VCRBP) in view of the deficiencies of the existing IVC-RBP technology.

[0004] The method includes the following steps:

[0005] Step 1, at the receiving end, receive the codeword sequence of the LDPC code, and then perform initialization and residual calculation: initialize the messages from all check nodes to variable nodes to 0, and initialize the messages from all variable nodes to check nodes to the received log-likelihood ratio (LLR) value; meanwhile, calculate the initial residual degree of the message from each variable node to the check node.

[0006] Step 2, Locate unstable variable nodes: Compare the decision log-likelihood ratio (LLR) values of each variable node before and after message update to detect the stability of the variable node; an unstable variable node refers to a node whose sign of the decision log-likelihood ratio (LLR) value changes before and after message update; the method of the present invention preferentially processes unstable variable nodes, which is beneficial to improving the convergence speed and accuracy of decoding.

[0007] Step 3, Silence the edge with the maximum residual degree of the unstable variable node: By calculating and sorting the residual degrees of all edge information of the unstable variable node, locate the edge with the maximum residual degree of the unstable variable node and silence the state of the edge with the maximum residual degree; the unstable variable node is called a silent node, and the edge where the maximum residual degree is located is called a silent edge; the setting of the silent node reduces the number of information transmissions, which is beneficial to improving the convergence speed of the decoding method and reducing the decoding complexity.

[0008] Step 4, Generate and transmit the input message of the silent edge: For the check node connected to the silent edge, generate and transmit all input information connected to the check node, but not including the information originating from the variable node connected to the silent edge.

[0009] Step 5, Generate and transmit the output message of the silent edge: For the variable node connected to the silent edge, generate and transmit all output information connected to the variable node, but not including the information originating from the check node connected to the silent edge.

[0010] Step 6, Update the residual degree value and the set of unstable variable nodes: Update all changes in the residual degree caused by Steps 4 and 5, and re-statistically update the set of unstable variable nodes.

[0011] Step 7, Check the stop condition: If all variable nodes satisfy the parity-check equation or reach the preset maximum number of iterations, stop decoding; otherwise, return to Step 2 to detect unstable variable nodes and continue iteration.

[0012] In Step 1, the information bit sequence to be transmitted is encoded by an LDPC encoder, and the information bit sequence to be transmitted is extended into a codeword sequence of an LDPC code and then sent. The codeword sequence of the LDPC code is transmitted to the receiving end through a noisy channel. The information bit sequence to be transmitted comes from various data sources at the sending end, such as binary forms of text, images, audio, or video, etc. The decoding method and device of the present invention start error correction decoding of the received sequence at the receiving end.

[0013] Step 1 further includes:

[0014] Initialize messages: , ;

[0015] Initialize residual degrees: ;

[0016] wherein represents the message from the check node a to the variable node v;

[0017] represents the message from the variable node v to the check node a;

[0018] represents the log-likelihood ratio (LLR) value received by the variable node v;

[0019] represents the message redundancy;

[0020] represents the absolute value of the received log-likelihood ratio (LLR) value.

[0021] In step 2, the following method is used to locate the unstable variable nodes: If sign( , then the variable node v is unstable, wherein represents the message from the variable node v to the check node a before update, represents the message from the variable node v to the check node a after update, and sign is a sign function used to determine the sign of a number (positive, negative, or zero).

[0022] In step 3, the search range for the edge with the maximum redundancy is limited to the unstable variable nodes, greatly reducing the search range; if there are no unstable variable nodes, then search among all nodes;

[0023] The following method is used to select the edge with the maximum redundancy:

[0024] For the non-empty set of unstable variable nodes, ;

[0025] Otherwise, ;

[0026] wherein represents the maximum redundancy;

[0027] represents selecting the maximum redundancy among the unstable variable nodes;

[0028] represents selecting the maximum redundancy among all the edges of the LDPC code;

[0029] represents the set of unstable nodes;

[0030] represents all the edges.

[0031] In step 3, the state of the silent maximum residue edge means that: the information value of the edge where the maximum residue of the unstable variable node is located is not updated and not transmitted, that is, no processing is performed on the edge information, reducing the complexity of the decoding device.

[0032] In step 4, the messages are updated using the following formula:

[0033] Update the check node message: , where represents the updated message value;

[0034] The updated message value is calculated using the following formula :

[0035] = ),

[0036] where u represents a node in the set N(a) of variable nodes connected to the check node a, but does not include the specified variable node v; represents the message from node u to check node a;

[0037] where represents the set of all variable nodes connected to check node a;

[0038] represents the set after removing variable node v;

[0039] σ is a function, usually the hyperbolic tangent function, used to perform a non-linear transformation on the message;

[0040] Set the residue to 0: = 0.

[0041] Step 5 includes:

[0042] Update the variable node message: ;

[0043] Stability judgment: If sign( , calculate the residue:

[0044] = ,

[0045] If sign( , it means that the confidence of the variable node has not changed significantly, so no further processing is required. Specifically, the following measures can be taken:

[0046] Do not update the residue: Since the confidence of the variable node has not changed, there is no need to calculate a new residue value; the residue value remains unchanged;

[0047] Skip variable node: In subsequent iteration processes, the update of this variable node can be skipped, and other variable nodes can be directly processed; this can reduce unnecessary calculations and improve the decoding efficiency.

[0048] Continue iteration: Although the confidence of the current variable node has not changed, other variable nodes may still need to be updated; therefore, continue the iteration until the stop condition is met.

[0049] Step 6 includes: If all variable nodes satisfy: = 0, or the maximum number of iterations is reached: t ≥ , then stop decoding; where N(v) represents the set of check nodes connected to variable node v; t is the current number of iterations; is the maximum number of iterations.

[0050] The present invention also provides an LDPC code timing decoding device based on the maximum remaining degree silent mechanism of unstable variable nodes implemented according to the above method, which is characterized by including an initialization module, an unstable variable detection module, a maximum remaining degree selection module, a check node message update module, a variable node message propagation module, a stop condition check module, and a control module;

[0051] The initialization module is used for:

[0052] Message initialization: Initialize the messages from all check nodes to variable nodes to 0;

[0053] LLR initialization: Initialize the messages from variable nodes to check nodes to the received LLR values;

[0054] Calculate the remaining degree: Calculate the initial remaining degree of the messages from each variable node to check nodes, and sort the remaining degrees in descending order;

[0055] The unstable variable detection module is used for:

[0056] Symbol comparison: Compare the symbols before and after the update of the variable node messages;

[0057] Mark unstable variable nodes: Detect and mark unstable nodes;

[0058] The maximum remaining degree selection module is used for:

[0059] Remaining degree comparison: Select the edge with the maximum remaining degree among unstable variable nodes;

[0060] Global selection: If there are no unstable variable nodes, select the edge with the maximum remaining degree in the entire system;

[0061] The check node message update module is used for:

[0062] Check node message update: Update the check node message corresponding to the selected edge;

[0063] Residual degree reset to zero: Set the residual degree of the selected edge to 0;

[0064] The variable node message propagation module is used for:

[0065] Variable node message update: Update the variable node message corresponding to the selected edge;

[0066] Stability judgment: Judge the stability of the variable node;

[0067] Residual degree update: Calculate the new residual degree value;

[0068] The stop condition check module is used for:

[0069] Check equation check: Check whether all variable nodes satisfy the check equation;

[0070] Iteration number check: Check whether the maximum iteration number is reached;

[0071] The control module is used for:

[0072] Process control: Control the entire decoding process, including module coordination and iteration control;

[0073] State management: Manage the states and conditions during the decoding process.

[0074] The present invention relates to the timing design of LDPC code decoding, aiming to improve the decoding performance of LDPC codes and reduce the decoding complexity. The LDPC code timing decoding method (INP-VCRBP) based on the maximum residual degree silence mechanism of unstable variable nodes significantly improves the decoding efficiency and error correction ability by optimizing the message update strategy and reducing unnecessary calculations.

[0075] Advantageous effects: The present invention proposes an LDPC code timing decoding method (INP-VCRBP) based on the maximum residual degree silence mechanism of unstable variable nodes, which has significant advantageous effects compared with traditional decoding methods.

[0076] First of all, the present invention significantly improves the decoding performance. By optimizing the message update strategy, the INP-VCRBP algorithm's approach of silencing the edges with the maximum residual degree of unstable variable nodes can more effectively reduce the number of unstable variable nodes. This strategy enables the algorithm to perform excellently under low signal-to-noise ratio and low bit error rate requirements, converge to the correct solution faster, and reduce the bit error rate and frame error rate. Experimental results show that whether the code length is 155 or 576, the INP-VCRBP algorithm is superior to the IVC-RBP algorithm in terms of BER (bit error rate) and FER (frame error rate), especially under high signal-to-noise ratio conditions, where the performance advantage is more obvious.

[0077] Secondly, the present invention effectively reduces the decoding complexity. By silencing the edges with the maximum residual degree of unstable variable nodes, unnecessary message updates are reduced. At the same time, the INP-VCRBP algorithm avoids repeated updates of the initially selected edges, reducing the number of message recalculations. This mechanism significantly reduces the computational complexity and improves the decoding efficiency. Meanwhile, the algorithm introduces a dual decision-making mechanism, combining the concepts of residual degree and unstable variable nodes, to more effectively locate the edges that need to be updated preferentially, further optimizing the decoding process. This enables the INP-VCRBP algorithm to significantly improve the decoding performance while maintaining a low complexity, enhancing the robustness of the algorithm.

[0078] In addition, the present invention has important practical application value. The INP-VCRBP algorithm can perform excellently under different signal-to-noise ratio conditions, especially under low signal-to-noise ratio and low bit error rate requirements. This makes the algorithm have broad application prospects in fields such as wireless communication and satellite communication, and can promote the further application and development of LDPC codes in these fields. Future research can further explore the application of this algorithm in LDPC code decoding to further improve the performance and reliability of the coding scheme.

[0079] In summary, through innovative decoding methods and device designs, the present invention improves the decoding performance of LDPC codes while reducing the decoding complexity, and has important practical application value. Brief Description of the Drawings

[0080] Figure 1 It is a schematic diagram of the technical solution of the INP-VCRBP decoder proposed by the present invention.

[0081] Figure 2 It is a schematic diagram of the relationship between the bit error rate and frame error rate of each decoding scheme and the signal-to-noise ratio of Code A in the Gaussian white noise channel environment.

[0082] Figure 3 It is a schematic diagram of the relationship between the bit error rate and frame error rate of each decoding scheme and the signal-to-noise ratio of Code B in the Gaussian white noise channel environment.

[0083] Figure 4 This is the flowchart of the method of the present invention. Specific embodiments

[0084] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments, and the above and / or other advantages of the present invention will become clearer.

[0085] As Figure 4 shown, the embodiment of the present invention provides an LDPC code timing decoding method based on the maximum residual degree silent mechanism of unstable variable nodes. For the convenience of understanding the decoding process, the following symbols are defined in the present invention:

[0086] represents the variable node to the check node message residual degree;

[0087] σ is a function, usually the hyperbolic tangent function, used to perform a non-linear transformation on the message;

[0088] : represents the information passed from the i-th check node to the j-th variable node ;

[0089] : represents the information passed from the j-th variable node to the i-th check node ;

[0090] N( )={ ∣ =1} represents a set of variable nodes connected to the i-th check node in the Tanner graph;

[0091] N( )={ ∣ =1} represents the set of all check nodes connected to the j-th variable node in the graph;

[0092] N( )\ represents the subset generated after removing the j-th variable node ;

[0093] N( )\ represents the subset obtained after removing the i-th check node ;

[0094] L( ) is a real value representing the confidence of the received signal for the j-th variable node .

[0095] The IVC-RBP algorithm can identify unstable variable nodes as an indicator of algorithm convergence. Based on the fact that the greater the redundancy, the faster the convergence speed, IVC-RBP accelerates convergence by first updating and propagating the messages of the unstable variable nodes with the largest V2C redundancy.

[0096] The following is a simple modification to the IVC-RBP algorithm: Without loss of generality, assume that the variable node is unstable and has the largest redundancy. The first step is to update and propagate the information , ; instead of updating itself to make the node more reliable. The second step is to set , update and propagate the message , . The last step is to determine whether is still unstable, calculate the new redundancy , , , which will be updated simultaneously.

[0097] The detailed calculation steps of the INP-VCRBP decoding algorithm include:

[0098] Step 1, initialization:

[0099] ,

[0100] Step 2, check whether there are unstable variable nodes. If there are unstable variable nodes, execute Step 3; otherwise, execute Step 5.

[0101] If is unstable;

[0102] Step 3, search for the maximum redundancy of the unstable variable nodes :

[0103] ,

[0104] where unstable variable node represents an unstable variable node;

[0105] Step 4, if there are no unstable variable nodes, find the maximum redundancy in the entire system :

[0106] ,

[0107] Step 5, for each , generate and transmit a message :

[0108] ,

[0109] Step 6, set the remaining degree of the selected edge to 0, that is ;

[0110] Step 7, for each , generate and transmit a message , and perform a stability check on the variable node . For each , calculate the new remaining degree , and update the message :

[0111] ,

[0112] ,

[0113] Step 8, if the stop condition is not met, return to Step 2 to continue the iteration; otherwise, end the decoding process.

[0114] If , , stop.

[0115] Code A for simulation: A quasi-cyclic LDPC code with a code rate of 0.4 and a code length of 155.

[0116] Code B for simulation: A quasi-cyclic LDPC code with a code rate of 0.5 and a code length of 576.

[0117] In the embodiments of the present invention, the computer used has a configuration of 16 GB of memory, a CPU of a desktop computer with 12th Gen Intel(R) Core(TM) i5-1240P 1.70 GHz, and the code used is developed in Matlab2024b language. There are two types of LDPC codes, A and B, for testing the inventive solution: Code A is a quasi-cyclic LDPC code with a code rate of 0.4 and a code length of 155; Code B is a quasi-cyclic LDPC code with a code rate of 0.5 and a code length of 576. After analyzing the performance of each decoding scheme, the following conclusions are obtained:

[0118] Figure 1 Shows the core technical schematic diagram of the INP-VCRBP decoder proposed by the present invention. Circular nodes: labeled , , , , , , , representing a check node. Triangular node: Marked as , , , representing a variable node. And assume that the variable node is an unstable variable node, and has the largest degree of redundancy. Subsequently, update and propagate the message , set the degree of redundancy r( ) = 0, update and propagate the message . Subsequently, judge whether it is still unstable, and calculate the new degree of redundancy r( ).

[0119] Figure 2 shows the comparison of the decoding performance between the INP-VCRBP algorithm and the IVC-RBP algorithm when the code length is 155. It can be seen from the figure that the INP-VCRBP algorithm is superior to the IVC-RBP algorithm in both BER and FER. Specifically:

[0120] In the low signal-to-noise ratio region, both the bit error rate and the frame error rate of the INP-VCRBP algorithm are lower than those of the IVC-RBP algorithm, which indicates that the INP-VCRBP algorithm has better robustness under low signal-to-noise ratio conditions. As the signal-to-noise ratio increases, the performance gap between the two algorithms gradually increases. In the high signal-to-noise ratio region, the performance advantage of the INP-VCRBP algorithm is more obvious. At a bit error rate of 10 -4 , the INP-VCRBP algorithm has a gain of 0.15 dB compared to the IVC-RBP algorithm. This means that under the same bit error rate requirement, the INP-VCRBP algorithm can be achieved at a lower signal-to-noise ratio, thus saving transmission power or improving spectral efficiency.

[0121] Figure 3 shows the comparison of the decoding performance between the INP-VCRBP algorithm and the IVC-RBP algorithm when the code length is 576. It can be seen from the figure that the INP-VCRBP algorithm is also superior to the IVC-RBP algorithm in both BER and FER. Specifically: In the low signal-to-noise ratio region, both the bit error rate and the frame error rate of the INP-VCRBP algorithm are lower than those of the IVC-RBP algorithm, which indicates that the INP-VCRBP algorithm also has better robustness under low signal-to-noise ratio conditions. As the signal-to-noise ratio increases, the performance gap between the two algorithms gradually increases. In the high signal-to-noise ratio region, the performance advantage of the INP-VCRBP algorithm is more obvious. Despite the increase in the code length to 576, the performance advantage of the INP-VCRBP algorithm still exists, which indicates that the INP-VCRBP algorithm can maintain good performance under different code lengths.

[0122] In summary, throughFigure 2 and Figure 3 From the detailed analysis of Figure 2 and Figure 3 , it can be seen that the INP-VCRBP algorithm significantly improves the decoding performance while maintaining a low decoding complexity. Whether the code length is 155 or 576, the INP-VCRBP algorithm is superior to the IVC-RBP algorithm in terms of BER and FER, especially under low SNR and low BER requirements. In addition, the INP-VCRBP algorithm requires fewer average iteration times and can complete the decoding process in a shorter time. This makes the INP-VCRBP algorithm of great value in practical applications and can promote the further application and development of LDPC codes in the field of wireless communication. Future research can further explore the application of this algorithm in LDPC code decoding to further improve the performance and reliability of the coding scheme.

[0123] Although the IVC-RBP algorithm has excellent error correction performance, it also has some deficiencies. The implementation complexity of the IVC-RBP algorithm is relatively high, mainly because it needs to classify variable nodes and dynamically select unstable variable nodes for priority update in each iteration. Although this dynamic scheduling mechanism can improve the decoding efficiency, it also increases the difficulty of algorithm design and implementation. Although the IVC-RBP algorithm improves the decoding efficiency by preferentially processing unstable variable nodes, in some cases, this strategy may cause the algorithm to fall into a local optimal solution. This is because the algorithm relies too much on residual information when selecting update nodes and ignores other factors that may affect the decoding performance. For example, when dealing with certain specific error patterns, the algorithm may not be able to effectively correct errors, thus affecting the overall decoding performance. Under certain channel conditions, the performance improvement of the IVC-RBP algorithm may not be as significant as expected. For example, in a low-noise channel, the performance improvement of the algorithm may not be as obvious as in a high-noise channel. This means that the performance of the algorithm may vary greatly under different channel conditions, which needs to be particularly noted in practical applications. The purpose of the INP-VCRBP (Informed and Non-Propagated variable-to-check Residual Belief Propagation) algorithm is to improve the original IVC-RBP algorithm to improve the decoding performance of low-density parity-check codes while reducing the decoding complexity. The INP-VCRBP algorithm significantly reduces the computational complexity, improves the decoding performance, and enhances the robustness of the algorithm by optimizing the dynamic scheduling strategy and reducing unnecessary message updates. The INP-VCRBP algorithm avoids repeated updates of the initially selected edges, reduces the number of message recalculations, thus accelerating the convergence speed and reducing the bit error rate and frame error rate. In addition, the INP-VCRBP algorithm introduces a dual decision-making mechanism, combining the concepts of residual degree and unstable variable nodes, to more effectively locate the edges that need to be updated preferentially, further optimizing the decoding process. This improvement enables the INP-VCRBP to perform well under different signal-to-noise ratio conditions, especially at high signal-to-noise ratios, where its performance improvement is particularly significant. In summary, the INP-VCRBP comprehensively improves the IVC-RBP in terms of computational complexity, decoding performance, and robustness, making it more advantageous in practical applications.

[0124] The present invention provides a timing decoding method and device for LDPC codes based on an unstable variable node maximum remaining degree silent mechanism. There are many methods and ways to specifically implement this technical solution. The above is only the preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art of this technology, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention. Each component not clearly defined in this embodiment can be implemented using existing technologies.

Claims

1. A method for sequential decoding of LDPC codes based on a maximum residual silence mechanism of unstable variable nodes, characterized in that: The following steps are involved: Step 1: Receive the codeword sequence of the LDPC code at the receiving end, and then perform initialization and residual calculation: initialize all messages from the check node to the variable node to 0, and initialize all messages from the variable node to the check node to the received log-likelihood ratio LLR value; at the same time, calculate the initial residual degree of each variable node to the check node message; Step 2, locate unstable variable nodes: compare the decision log-likelihood ratio (LLR) values ​​of each variable node before and after update to detect the stability of the variable node; Unstable variable nodes refer to nodes whose LLR value signs change before and after the message update; Step 3, silencing the maximum remaining degree edge of the unstable variable node: by calculating and sorting the remaining degrees of all edge information of the unstable variable node, locating the maximum remaining degree edge of the unstable variable node, and silencing the state of the maximum remaining degree edge; Unstable variable nodes are called silent nodes, and the edge with the maximum residual degree is called silent edge; Step 4, generating and transmitting input messages of silent edges: for a check node connected to a silent edge, generating and transmitting all input messages connected to the check node, but excluding input messages from variable nodes connected to the silent edge; Step 5, generating and transmitting output messages of silent edges: for variable nodes connected to silent edges, generating and transmitting all output messages connected to the variable nodes, but excluding those from check nodes connected to the silent edges; Step 6, update the remaining degree value and the unstable variable node set: update all the remaining degree changes caused by steps 4 and 5, and re-count and update the unstable variable node set; Step 7, check the stop condition: if all variable nodes satisfy the verification equation or reach the preset maximum number of iterations, stop decoding; otherwise, return to step 2 to detect unstable variable nodes and continue iteration.

2. The method according to claim 1, characterized in that In step 1, the information bit sequence to be transmitted is encoded by an LDPC encoder, and the information bit sequence to be transmitted is expanded into a codeword sequence of an LDPC code and then sent. The codeword sequence of the LDPC code is transmitted to the receiving end via a noisy channel.

3. The method according to claim 2, characterized in that Step 1 also includes: Initialization message: m a→v =0,m v→a =C v ; Initialize the residual degree: r(m v→a )=|C v |; Where m a→v Represents the message from check node a to variable node v; m v→a Represents the message from variable node v to check node a; C v Represents the log-likelihood ratio LLR value received by the variable node v; r(m v→a ) indicates message m a→v The residual degree; |C v | represents the absolute value of the received log-likelihood ratio LLR value.

4. The method according to claim 3, characterized in that In step 2, the following method is used to locate the unstable variable node: If Then the variable node v is unstable, where Represents the message from variable node v to check node a before update, Represents the message from the updated variable node v to the check node a. sign is an information function used to determine whether the value of the message is positive, negative, or zero.

5. The method according to claim 4, characterized in that In step 3, the search range of the maximum residual degree edge is limited to unstable variable nodes; if there is no unstable variable node, it is searched in all nodes; The maximum residual degree edge is selected as follows: For a non-empty set of unstable variable nodes, r max =max v∈instability r(m v→a ); Otherwise, r max =max alledges r(m v→a ); where r max represents the maximum residual degree; max v∈instability r(m v→a ) indicates selecting the maximum remaining degree among the unstable variable nodes; max alledges r(m v→a ) indicates selecting the maximum residual degree among all edges of the LDPC code; instability represents a set of unstable nodes; alledges means all edges.

6. The method according to claim 5, characterized in that In step 3, the state of the silent maximum residual degree edge means that the information value of the edge where the maximum residual degree of the unstable variable node is located is not updated or transmitted.

7. The method according to claim 6, characterized in that In step 4, the message is updated using the following formula: Update check node message: m a→v =X1, where X1 represents the updated message value; The updated message value X1 is calculated using the following formula: X1=∏ u∈N(a)\v σ(m u→a ), Where u represents a node in the variable node set N(a) connected to the check node a, but does not include the specified variable node v; m u→a Represents the message from node u to check node a; Where N(a) represents the set of all variable nodes connected to the check node a; N(a)\v represents the set after removing variable node v; σ is a function used to transform the message nonlinearly; Set the remainder to 0: r(m v→a )=0.

8. The method according to claim 7, characterized in that Step 5 includes: Update variable node message: m v→a =X1; Stability judgment: If Calculate the remainder: if This means that the confidence of the variable node has not changed significantly.

9. The method according to claim 8, characterized in that Step 6 includes: If all variable nodes satisfy: ∑ a∈N(v) m a→v = 0, or the maximum number of iterations is reached: t ≥ T max , then stop decoding; where N(v) represents the set of check nodes connected to the variable node v; t is the current iteration number; T max is the maximum number of iterations.

10. An LDPC code sequential decoding device based on the unstable variable node maximum residual silence mechanism implemented according to the method according to any one of claims 1 to 9, characterized in that: It includes an initialization module, an unstable variable detection module, a maximum surplus degree selection module, a check node message update module, a variable node message propagation module, a stop condition check module and a control module; The initialization module is used to: Message initialization: Initialize all messages from check nodes to variable nodes to 0; LLR initialization: Initialize the message from the variable node to the check node to the received LLR value; Calculate the remaining degree: calculate the initial remaining degree of each variable node to check node message, and sort the remaining degrees in descending order; The unstable variable detection module is used for: Symbol comparison: compare the symbols before and after the variable node message is updated; Unstable variable node marking: detect unstable nodes and mark them; The maximum remainder selection module is used for: Residual degree comparison: select the edge with the largest residual degree among the unstable variable nodes; Global selection: If there is no unstable variable node, select the edge with the largest remaining degree in the entire system; The check node message update module is used for: Check node message update: Update the check node message corresponding to the selected edge; Clear the remaining degree: set the remaining degree of the selected edge to 0; The variable node message propagation module is used for: Variable node message update: update the variable node message corresponding to the selected edge; Stability judgment: judge the stability of variable nodes; Residue update: calculate the new residual value; The stop condition checking module is used for: Verification equation check: check whether all variable nodes satisfy the verification equation; Iteration number check: check whether the maximum number of iterations has been reached; The control module is used for: Process control: control the entire decoding process, including module coordination and iterative control; state management: manage the state and conditions during the decoding process.

Images