LDPC decoding method based on node reliability dynamic correction criterion

Abstract

The invention provides an LDPC (Low Density Parity Check) decoding method based on a node reliability dynamic correction criterion. The LDPC decoding method comprises the following steps: step 1, initializing channel information; step 2, judging whether the maximum number of iterations is reached or not, if so, outputting a result, otherwise, entering step 3; 3, calculating extrinsic information from the check node to the variable node; 4, calculating extrinsic information from the variable node to the check node; 5, adjusting the correction factor according to the matching condition of the variable node symbol and the last iteration symbol; 6, if the correction factor exceeds the threshold value, limiting the correction factor to be the threshold value, otherwise, continuing the next step; 7, multiplying the correction factor by the variable node information, and adjusting the weight of the node information; 8, calculating full information of the variable nodes and performing hard decision; step 9, checking whether the judgment result meets the condition that the checksum is zero or not, if so, outputting a decoding result, otherwise, continuing iteration; and step 10, if the condition is not met, entering a next round of iteration, and updating check node information.

Description

Technical Field

[0001] The invention relates to the technical field of LDPC decoding methods for error correction codes, and in particular to an LDPC decoding method based on a node reliability dynamic correction criterion. Background Art

[0002] With the increasing complexity of communication service scenarios and application types and the improvement of communication performance indicators, people have put forward higher requirements for the transmission security and transmission rate of mobile communication technology. Channel coding technology is an important part of modern mobile communications. Low-density parity check code LDPC is a coding scheme close to the Shannon limit. Its coding method is simple and clear, and the decoding complexity is relatively low. At the same time, it has the ability of parallel operation, which is very suitable for hardware implementation in practical applications. The more classic LDPC decoding algorithms mainly include SPA and BP algorithms. This decoding algorithm uses soft information for decoding iteration, including probability information and log-likelihood ratio LLR information. Through continuous iterative operations, decoding performance close to the Shannon limit can be achieved. However, since the soft decision algorithm uses a large number of multiplication operations when calculating probability information, it requires high computational complexity and computing resources, and hardware implementation is difficult.

[0003] Using log-likelihood ratio information to represent probability information can convert a large number of multiplication operations into addition operations, greatly reducing the computational complexity. Since then, scholars have proposed various improved algorithms such as the minimum sum MS algorithm and the BP-based approximation algorithm to reduce the decoding complexity, but this will also lead to a loss in decoding accuracy. Therefore, a binary LDPC decoding algorithm based on reliability is proposed. This type of algorithm will extract reliability information based on the bit information before and after the decision to help decoding, while improving the decoding complexity to a certain extent and improving the decoding performance. The most typical of these is the weighted bit flipping WBF algorithm and its improved algorithms, such as the MWBF (Modified WBF) algorithm and the IMWBF algorithm. These two algorithms reduce the computational complexity while reducing the loss of decoding performance. However, because their weighting coefficients are fixed, it will lead to the appearance of pseudo-local maximum phenomena, and there is an error flat layer in some multivariate LDPC codes. In addition, it also has oscillation phenomena, and may cause problems such as the lower limit of the error. Summary of the invention

[0004] The purpose of the present invention is to provide an LDPC decoding method based on a node reliability dynamic correction criterion, aiming to solve the technical problems that when the existing LDPC decoding method reduces the algorithm complexity, the decoding performance is reduced and there is non-convergence of variable node oscillation.

[0005] To achieve the above object, the present invention provides an LDPC decoding method based on a node reliability dynamic correction criterion, comprising the following steps:

[0006] Step 1: Initialize the incoming channel information value to serve as the initialization information of the variable node;

[0007] Step 2: Determine whether the current number of iterations has reached the preset maximum number of iterations. If so, exit the loop and output the decoded codeword. Otherwise, proceed to step 3.

[0008] Step 3: Pass the variable node information in and calculate the external information passed by the check node to the variable node;

[0009] Step 4: Pass the check node information in and calculate the external information passed by the variable node to the check node;

[0010] Step 5: Determine the sign of the information value of each variable node at present. If the sign is the same as that in the previous iteration, the value of the correction factor is enlarged. If it is different, the value of the correction factor is reduced.

[0011] Step 6: Determine whether the value of the correction factor has exceeded the set threshold. If it exceeds the maximum threshold, the value of the correction factor and the value of the corresponding variable node are reset to zero; if it exceeds the minimum threshold, the correction factor is equal to the minimum threshold;

[0012] Step 7: Multiply the variable node by the correction factor, the information value of the variable node with high reliability is amplified, and the information value of the variable node with low reliability is reduced;

[0013] Step 8: Calculation of full information of variable nodes and hard decision;

[0014] Step 9: Check the judgment result. If the checksum is zero, the decoding is completed and the decoding result is output. If the condition is not met, go to step S10;

[0015] Step 10: Return to the decoding iteration process of step S2 and enter the next round of iteration. At the same time, the information of the previous iteration is passed to the check nodes that need to be updated for allocation of the check nodes.

[0016] Optionally, for the initialization processing of the channel information, the initialization information value of the check node is set to 0, and the initial information value of the variable node comes from the channel initial message

[0017]

[0018] L(P i ) represents the information value initially defined by the channel, i.e., the external information transmitted from the variable node to the check node in the first iteration; P i (b) indicates that the receiving end receives y i After that, the corresponding codeword c i = the posterior probability of b, b = 0, 1; v iis the i-th variable node.

[0019] Optionally, the update rule of the check node includes the following calculation formula:

[0020] a. BP algorithm, the calculation formula for information transmission from the check node to the variable node is as follows:

[0021]

[0022] b.MS algorithm, the calculation formula for information transmission from the check node to the variable node is as follows:

[0023] r ji =Π i′∈V(j)\i sgn(q i′j )·min i′∈V(j)\i (|q i′j |)

[0024] c. NMS algorithm, the calculation formula for information transmission from the check node to the variable node is as follows:

[0025] r ji =α·Π i′∈V(j)\i sgn(q i′j )·min i′∈V(j)\i (q i′j )

[0026] d. OMS algorithm, the calculation formula for information transmission from the check node to the variable node is as follows:

[0027] r ji =max(min i′∈V(j)\i (|q i′j |)-β,0)·Π i′∈V(j)\i sgn(q i′j )

[0028] in, Indicates the external information passed from check node j to variable node i in the lth iteration, b = 0, 1; represents the external information transmitted by variable node i to check node j in the lth iteration, b = 0, 1; V(j)\i represents the set of other variable nodes connected to the jth check node except the i-th variable node, V(j)\i = {k:h kj =1, k≠i}; α is the scaling factor of the NMS algorithm, and β is the offset factor of the OMS algorithm.

[0029] Optionally, the external information passed by the variable node to the check node, the calculation formula of V2C is as follows:

[0030]

[0031] q ij =qij *α ij

[0032] α ij =1 / num(i,j)

[0033] Where C(i)\j represents the set of other check nodes connected to the i-th variable node except the j-th check node, C(i)\j={k:h ik =1,k≠j};α ij Represents the dynamic correction factor of each variable node; num(i,j) is used to measure the reliability of the external information transmitted by variable node i to check node j.

[0034] Optionally, α ij It is the dynamic offset correction factor of the information passed from the variable node to the check node. In each new iteration, each variable node will have a one-to-one corresponding correction factor α ij The value of the correction factor is determined by the reliability of the node. If the reliability of the node is high, the value of the correction factor is enlarged, otherwise it is reduced. The reliability of the variable node is determined by whether the sign of the information transmitted to the check node by the variable node in the two iterations before and after is consistent. If the signs are the same, it is a reliable node, and if the signs are different, it is an unstable node. The formula is as follows:

[0035]

[0036] Among them, Vij (l) (i, idx(j)) represents the external information passed to the check node by the current variable node, Vij (l-1) (i,idx(j)) represents the external information passed from the variable node to the check node in the previous iteration.

[0037] Optionally, the dynamic offset correction factor α of the information outside the check node is passed to the variable node ij The size of has a specified threshold, under which the size of the external information transmitted by the variable node to the check node changes within a stable range.

[0038] Optionally, the calculation of the full information of the variable node and the hard decision formula are as follows:

[0039]

[0040] When ij >0, that is, when the posterior probability is greater than 0, the i-th variable node v ij The judgment is 0, otherwise the judgment is 1.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] (1) The present invention improves the accuracy and convergence speed of decoding by dynamically adjusting the node reliability correction factor.

[0043] (2) The present invention sets a correction factor threshold to avoid over-adjustment, ensure decoding stability and improve anti-noise capability.

[0044] (3) The present invention can quickly exit iteration by intelligently judging the checksum, thereby reducing decoding delay and improving real-time performance.

[0045] (4) The present invention adaptively adjusts the decoding strategy under different channel conditions to improve the robustness and fault tolerance of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0047] Figure 1 The present invention is a flow chart of an LDPC decoding method based on a node reliability dynamic correction criterion.

[0048] Figure 2 It is the BER data of various commonly used LDPC decoding algorithms.

[0049] Figure 3 It is the BER data of various LDPC decoding algorithms to which the present invention is added. DETAILED DESCRIPTION

[0050] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.

[0051] See also Figure 1 The present invention provides an LDPC decoding method based on a node reliability dynamic correction criterion, comprising the following steps:

[0052] S1: Initialize the incoming channel information value and serve as the initialization information of the variable node;

[0053] S2: Determine whether the current number of iterations has reached the preset maximum number of iterations. If so, exit the loop and output the decoded codeword. Otherwise, proceed to step 3.

[0054] S3: Pass the variable node information in and calculate the external information passed by the check node to the variable node;

[0055] S4: pass the check node information in and calculate the external information passed by the variable node to the check node;

[0056] S5: Determine the sign of the information value of each variable node at present. If the sign is the same as that in the previous iteration, the value of the correction factor is enlarged. If the sign is different, the value of the correction factor is reduced.

[0057] S6: Determine whether the value of the correction factor exceeds the set threshold. If it exceeds the maximum threshold, the value of the correction factor and the value of the corresponding variable node are reset to zero; if it exceeds the minimum threshold, the correction factor is equal to the minimum threshold;

[0058] S7: multiply the variable node by the correction factor, so that the information value of the variable node with high reliability is amplified, and the information value of the variable node with low reliability is reduced;

[0059] S8: Calculation of full information of variable nodes and hard decision;

[0060] S9: Check the judgment result. If the checksum is zero, the decoding is completed and the decoding result is output. If the condition is not met, go to step S10;

[0061] S10: Return to the decoding iteration process of step S2 and enter the next round of iteration. At the same time, the information of the previous iteration is passed to the check nodes that need to be updated for allocation of the check nodes; the number of iterations +1.

[0062] Specifically, in the study of the Low Density Parity Check (LDPC) decoding algorithm, it was found that the reliability of its variable nodes is different. Nodes with poor reliability will reduce the decoding performance, and make the decoding process oscillate, unable to converge quickly, and will also cause the problem of error floor that cannot be ignored. In order to improve the above defects, an LDPC decoding method based on the dynamic correction criterion of node reliability is proposed. The algorithm reasonably judges the reliability of the variable node by the difference in the information value sign before and after the two iterations of the variable node, and then judges the degree of reliability of the variable node by the threshold. By obtaining the node reliability information, the correction factor corresponding to the variable node is adaptively obtained. The correction factor can improve the reliability and stability of the variable node by multiplying it with the variable node, thereby improving the decoding performance.

[0063] Further, the following is an explanation in combination with specific implementation steps:

[0064] Step S1: During information transmission, the original binary codeword c=(c1, c2, …, c n) is binary phase shift keying (BPSK, Binary Phase Shift Keying, BPSK mapping rule is x i =2c i -1) modulation mode After modulation, the received signal sequence y=(y1,y2,…,y n )for

[0065] y i =x i +n i i=1,2,…,n

[0066] Among them, n i It has a mean of 0 and a variance of σ 2 =N0 / 2 Gaussian white noise.

[0067] Redefine the following symbols:

[0068] v i represents the i-th variable node; c j represents the jth check node; Indicates the external information passed from check node j to variable node i in the lth iteration, b = 0, 1; represents the external information transmitted from variable node i to variable node j in the lth iteration, b = 0, 1; C(i) represents the set of all check nodes connected to the i-th variable node, C(i) = {i:h ij =1}; V(j) represents the set of all variable nodes connected to the jth check node, V(j) = {j:h ij =1}; C(i)\j represents the set of other check nodes connected to the i-th variable node except the j-th check node, C(i)\j={k:h ik =1, k≠j}; V(j)\i represents the set of other variable nodes connected to the jth check node except the i-th variable node, V(j)\i={k:h kj =1,k≠i};P i (b) indicates that the receiving end receives y i After that, the corresponding codeword c i = the posterior probability of b, b = 0, 1; Represents the posterior probability information of the i-th variable node in the l-th iteration, b = 0, 1;

[0069] Channel initialization processing: Initialize the information value of the check node to 0, and the variable node is initialized from the channel information as follows:

[0070]

[0071] At the same time, set the maximum number of iterations of the entire algorithm I max .

[0072] Step S3, the variable node information is passed in, and the external information passed to the variable node by the check node is calculated. The present invention proposes an LDPC decoding method based on a node reliability dynamic correction criterion. The method is suitable for optimizing a variety of decoding algorithms, and demonstrates the check node update rules of the commonly used BP, MS, NMS and OMS algorithms.

[0073] The update rules of the check nodes include the following calculation formula:

[0074] a. BP algorithm, the calculation formula for information transmission from the check node to the variable node is as follows:

[0075]

[0076] b.MS algorithm, the calculation formula for information transmission from the check node to the variable node is as follows:

[0077] r ji =Π i′∈V(j)\i sgn(q i′j )·min i′∈V(j)\i (q i′j )

[0078] c. NMS algorithm, the calculation formula for information transmission from the check node to the variable node is as follows:

[0079] r ji =α·Π i′∈V(j)\i sgn(q i′j )·min i′∈V(j)\i (|q i′j |)

[0080] d. OMS algorithm, the calculation formula for information transmission from the check node to the variable node is as follows:

[0081] r ji =max(min i′∈V(j)\i (|q i′j |)-β,0)·Π i′∈V(j)\i sgn(q i′j )

[0082] Step S4: pass the check node information in and calculate the external information passed by the variable node to the check node. The process of passing the check node information in involves the operation of calculating the external information according to the constraints. External information refers to the information calculated by the check node based on the received variable node information and constraints, which reflects the check node's expectations and constraint information on the variable node. This external information will be passed back to the variable node to update the variable node information, and further check node information will be passed in and calculated in the next round of iteration. The calculation formula is as follows:

[0083]

[0084] Step S5, during the decoding process, some variable nodes are reliable, while others are not. During the decoding process, this paper proposes an operation based on the dynamic correction criterion of node reliability. Specifically, the operation involves determining the sign of the current information value of each variable node, and adjusting the value of the correction factor based on whether the sign is the same as that of the previous iteration. If the signs are the same, the value of the correction factor will be amplified; if the signs are different, the value of the correction factor will be reduced. The specific logic for determining node reliability is as follows:

[0085]

[0086] α ij =1 / num(i,j)

[0087] Among them, Vij (l) (i, idx(j)) represents the external information passed by the current variable node to the check node, Vij (l-1) (i,idx(j)) represents the external information passed from the variable node to the check node in the previous iteration, α ij Represents the dynamic correction factor of each variable node; num(i,j) is used to measure the reliability of the external information transmitted by variable node i to check node j.

[0088] The purpose of this dynamic correction criterion is to adaptively adjust the size of the correction factor according to the sign change of the variable node information value. By enlarging or reducing the value of the correction factor, the information values ​​of different variable nodes can be adjusted more finely during the decoding process. The advantage of this is that if the current symbol is the same as the symbol in the previous iteration, it means that the information value of the variable node is consistent, so the value of the correction factor can be increased to enhance the credibility and influence of the information value. On the contrary, if the current symbol is different from the symbol in the previous iteration, it means that the information value of the variable node has changed, and there may be a certain error or uncertainty, so it is necessary to reduce the value of the correction factor to reduce the degree of influence on the information value.

[0089] By comparison Figure 2 and Figure 3 It can be clearly seen from the data that the LDPC decoding method based on the dynamic correction criterion of node reliability (referred to as the NRDC method) proposed in this paper shows excellent performance advantages among various soft decision decoding algorithms. Figure 3 The results clearly show that the LDPC decoding algorithm improved by the proposed method has achieved significant performance improvement in bit error rate. This innovative technology not only performs well in various soft decision decoding algorithms, but also greatly improves their performance level. The introduction of this method has made significant contributions to improving the reliability and efficiency of communication systems, and has opened up new prospects for the development of future communication technologies.

Claims

1. An LDPC decoding method based on a node reliability dynamic correction criterion, characterized in that: The following steps are involved: Step 1: Initialize the incoming channel information value to serve as the initialization information of the variable node; Step 2: Determine whether the current number of iterations has reached the preset maximum number of iterations. If so, exit the loop and output the decoded codeword. Otherwise, proceed to step 3. Step 3: Pass the variable node information in and calculate the external information passed by the check node to the variable node; Step 4: Pass the check node information in and calculate the external information passed by the variable node to the check node; Step 5: Determine the sign of the information value of each variable node at present. If the sign is the same as that in the previous iteration, the value of the correction factor is enlarged. If it is different, the value of the correction factor is reduced. Step 6: Determine whether the value of the correction factor has exceeded the set threshold. If it is less than the minimum threshold, the value of the correction factor and the value of the corresponding variable node are reset to zero. If the maximum threshold is exceeded, the correction factor is made equal to the maximum threshold; Step 7: Multiply the variable node by the correction factor, the information value of the variable node with high reliability is amplified, and the information value of the variable node with low reliability is reduced; Step 8: Calculation of full information of variable nodes and hard decision; Step 9: Check the judgment result. If the checksum is zero, the decoding is completed and the decoding result is output. If the condition is not met, go to step S10; Step 10: Return to the decoding iteration process of step S2 and enter the next round of iteration. At the same time, pass the information of the previous iteration to the check nodes that need to be updated for allocation of the check nodes.

2. The LDPC decoding method based on the node reliability dynamic correction criterion according to claim 1, characterized in that: For the initialization of channel information, the initialization information value of the check node is set to 0, and the initial information value of the variable node comes from the initial message of the channel. L(P i ) represents the information value initially defined by the channel, i.e., the external information transmitted from the variable node to the check node in the first iteration; P i (b) indicates that the receiving end receives y i After that, the corresponding codeword c i = the posterior probability of b, b = 0, 1; v i is the i-th variable node.

3. The LDPC decoding method based on the node reliability dynamic correction criterion as claimed in claim 2, characterized in that: The update rules of the check nodes include the following calculation formula: a. BP algorithm, the calculation formula for information transmission from the check node to the variable node is as follows: b.MS algorithm, the calculation formula for information transmission from the check node to the variable node is as follows: r ji =Π i′∈V(j)\i sgn(q i′j )·min i′∈V(j)\i (|q i′j |) c. NMS algorithm, the calculation formula for information transmission from the check node to the variable node is as follows: r ji =a·P i′∈V(j)\i sgn(q i′j )·min i′∈V(j)\i (q| i′j |) d. OMS algorithm, the calculation formula for information transmission from the check node to the variable node is as follows: r ji =max(min i′∈V(j)\i (|q i′j |)-β,0)·Π i′∈V(j)\i sgn(q i′j ) in, Indicates the external information passed from the check node j to the variable node i in the first iteration, b = 0, 1; represents the external information passed from variable node i to check node j in the first iteration, b = 0, 1; V(j)\i represents the set of other variable nodes connected to the j-th check node except the i-th variable node, V(j)\i = {k:h kj =1, k≠i}; α is the scaling factor of the NMS algorithm, and β is the offset factor of the OMS algorithm.

4. The LDPC decoding method based on the node reliability dynamic correction criterion as claimed in claim 3, characterized in that: The external information passed by the variable node to the check node, the calculation formula of V2C is as follows: q ij =q ij *α ij a ij =1 / num(i,j) Where C(i)\j represents the set of other check nodes connected to the i-th variable node except the j-th check node, C(i)\j={k:h ik =1,k≠j};α ij Represents the dynamic correction factor of each variable node; num(i,j) is used to measure the reliability of the external information transmitted by variable node i to check node j.

5. The LDPC decoding method based on the node reliability dynamic correction criterion as claimed in claim 4, characterized in that: α ij It is the dynamic offset correction factor of the information passed from the variable node to the check node. In each new iteration, each variable node will have a one-to-one corresponding correction factor α ij The value of the correction factor is determined by the reliability of the node. If the reliability of the node is high, the value of the correction factor is enlarged, otherwise it is reduced. The reliability of the variable node is determined by whether the sign of the information transmitted to the check node by the variable node in the two iterations before and after is consistent. If the signs are the same, it is a reliable node, and if the signs are different, it is an unstable node. The formula is as follows: Among them, Vij (l) (i, idx(j)) represents the external information passed to the check node by the current variable node, Vij (l-1) (i,idx(j)) represents the external information passed from the variable node to the check node in the previous iteration.

6. The LDPC decoding method based on the node reliability dynamic correction criterion as claimed in claim 5, characterized in that: The dynamic offset correction factor α of the information outside the check node transmitted by the variable node i The size of j has a specified threshold value, under which the size of the external information transmitted by the variable node to the check node changes within a stable range.

7. The LDPC decoding method based on the node reliability dynamic correction criterion as claimed in claim 6, characterized in that: The calculation of the full information of the variable node and the hard decision formula are as follows: When ij >0, that is, when the posterior probability is greater than 0, the i-th variable node v ij The judgment is 0, otherwise the judgment is 1.

Images