A method and device for optimizing LDPC code structure for an interference environment

Abstract

The application discloses an LDPC code structure optimization method and device for an interference environment, and belongs to the technical field of channel coding and anti-interference communication technology. A check matrix is generated by constructing a base matrix and cyclic shift extension; the check matrix is subjected to initial optimization of ring structure, the minimum ring length is increased, and the number of small external edge degree rings is reduced; based on the optimized structure, a stopping set of each candidate cyclic shift value is searched, different sizes of stopping sets are subjected to weighted processing, and a normalized stopping set evaluation index is calculated; under the constraint of non-degradation of ring structure performance, the cyclic shift value is optimized and selected according to the index, iterative execution is performed until a termination condition is satisfied, and the optimized LDPC code structure is output. The application solves the problem of decoding performance decline caused by the stopping set in the interference environment, and realizes anti-interference optimization design.

Description

Technical Field

[0001] This invention relates to the field of channel coding and anti-interference communication technology, and in particular to a method and apparatus for optimizing LDPC code structure in interference environments. Background Technology

[0002] Low-density parity-check (LDPC) codes are widely used in modern communication systems due to their error-correcting performance, which is close to the Shannon limit. However, the decoding performance of LDPC codes degrades significantly in partial band interference (PBI) channels, narrowband interference, or other non-uniform interference environments.

[0003] In PBI channels, interference energy is concentrated within a limited frequency band, causing the log-likelihood ratio (LLR) information at the corresponding positions to approach zero, resulting in a severe reduction in the reliability of the information of the relevant bit nodes. In the decoding process based on the belief propagation (BP) algorithm, if the interfered positions form a stopping set, the variable nodes in this part cannot obtain valid update information from the check node, thus causing the decoding process to terminate prematurely.

[0004] Existing LDPC code construction methods mainly focus on optimizing parameters such as ring length and external edge degree (EMD) in the graph structure to reduce the performance degradation caused by short rings and weakly connected structures. However, the above methods do not fully consider the impact of stop set structure on decoding performance, especially in PBI channel environments, where stop sets often become a key factor leading to decoding failure.

[0005] Therefore, there is an urgent need for an LDPC code construction method that can simultaneously optimize the ring structure, EMD value, and stop set distribution to improve its robustness in interference environments. Summary of the Invention

[0006] The present invention aims to at least partially solve one of the technical problems in the related art.

[0007] To address this issue, this invention proposes an LDPC code structure optimization method for interference environments. The method involves constructing the base matrix of the LDPC code and generating a parity-check matrix through cyclic shift expansion. Initial optimization is performed on the ring structure of the parity-check matrix to increase the minimum ring length and reduce the number of small outer-degree rings. Based on the optimized structure, a stopping set corresponding to each candidate cyclic shift value is searched. The stopping sets are weighted, and a normalized evaluation index is calculated. Under the constraint of ensuring that the ring structure performance does not deteriorate, the cyclic shift value is optimized and selected according to the index. This process is iteratively executed until a preset termination condition is met, outputting the optimized LDPC code structure. This invention solves the problem of decoding performance degradation caused by stopping sets under interference environments and achieves anti-interference optimization design of the code structure.

[0008] Another objective of this invention is to propose an LDPC code structure optimization device for interference environments.

[0009] To achieve the above objectives, this invention proposes a method for optimizing LDPC code structure in interference environments, comprising:

[0010] Construct the basis matrix of the LDPC code and generate the corresponding parity check matrix through cyclic shift expansion. The parity check matrix corresponds to the Tanner graph. The ring structure of the parity check matrix is ​​initially optimized by increasing the minimum ring length and reducing the number of small outer edge degree rings to obtain the initially optimized LDPC code structure. Based on the initially optimized LDPC code structure, the stopping sets corresponding to each candidate cyclic shift value are searched, and the stopping sets of different sizes are weighted to calculate the normalized stopping set evaluation index. Under the constraint of ensuring that the performance of the ring structure does not deteriorate, the cyclic shift value is optimized according to the normalized stopping set evaluation index, and the stopping set search and optimization process is iteratively executed until the preset termination condition is met, and the optimized LDPC code structure is output.

[0011] The LDPC code structure optimization method for interference environments according to an embodiment of the present invention may also have the following additional technical features: In one embodiment of the present invention, the construction of the basis matrix of the LDPC code and the generation of the corresponding parity-check matrix through cyclic shift expansion, wherein the parity-check matrix corresponds to a Tanner graph, includes: Construct a sparse basis matrix B, where each element of the basis matrix B represents the cyclic shift of the corresponding submatrix; Select a boosting factor Z, and generate a corresponding cyclic shift identity matrix for each non-zero element in the basis matrix B; By concatenating all the submatrices, we obtain the complete parity matrix H.

[0012] In one embodiment of the present invention, the initial optimization of the ring structure of the parity-check matrix, increasing the minimum ring length and reducing the number of small outer edge degree rings to obtain the initially optimized LDPC code structure, includes: Search for ring structures in the Tanner graph and count the minimum ring length; Calculate the external edge degree of each ring to measure the degree of connection between the ring and external nodes; The simulated annealing algorithm is used to optimize the cyclic shift value so that the minimum ring length is not lower than the preset threshold, while reducing the number of small outer edge rings, thus obtaining the initial optimized LDPC code structure.

[0013] In one embodiment of the present invention, searching for the stopping set corresponding to each candidate cyclic shift value includes: For each position to be optimized in the verification matrix, enumerate all possible cyclic shift values ​​for the corresponding position to be optimized, with the value range from 0 to Z-1; Construct a corresponding local Tanner plot for each candidate cyclic shift value; Search the stopping sets formed in the local Tanner graph, and record the number of stopping sets and the size of each stopping set, where the size is the number of variable nodes.

[0014] In one embodiment of the present invention, weighting of stopping sets of different sizes includes: Determine the minimum size of the stopping set corresponding to all candidate cyclic shift values, denoted as . ; For a scale of The stopping set is assigned a weight coefficient of 1; for sets larger than 1, the weight coefficient is 1. The stopping set is determined using a monotonically decreasing weight function:

[0015] Where τ is the maximum range of the stopping set search, and s is the size of the stopping set; The normalized stopping set evaluation index is obtained by weighting and summing the number of stopping sets corresponding to each cyclic shift value based on the weighting coefficients.

[0016] In one embodiment of the present invention, optimizing the selection of the cyclic shift value based on the normalized stopping set evaluation index includes: If a candidate cyclic shift value reduces the stopping set evaluation index without increasing the number of cycles or decreasing the minimum cycle length, then the corresponding value shall be selected first. If there exists a candidate value that makes the number of cycles zero and the stopping set evaluation index optimal, then the corresponding value is directly selected and subsequent optimization is skipped.

[0017] In one embodiment of the present invention, it further includes: Construct an energy function that comprehensively reflects the changes in the number or structure of loops, as well as the evaluation index of stopping sets; A new solution is randomly selected from the candidate cyclic shift values. If the new solution is better than the current solution, it is likely to be accepted directly. If the new solution is worse than the current solution, it is less likely to be accepted. The probability of accepting a poor solution is gradually reduced as the number of iterations increases, and when there are candidate values ​​that do not introduce a new cycle, the corresponding candidate set is selected first.

[0018] In one embodiment of the present invention, the preset termination condition includes at least one of the following: The LDPC code structure remains unchanged throughout the iterative process, across multiple iterations. The number of iterations has reached the preset maximum number of iterations; Stop the evaluation index when it reaches the preset threshold.

[0019] In one embodiment of the present invention, the iterative execution of the stopping set search and optimization process includes: In each iteration, the position to be optimized is selected randomly or sequentially, the cyclic shift value of the corresponding position to be optimized is updated, and the cyclic information and stopping set distribution of the global structure are updated synchronously. The updated LDPC code structure will be used as the starting point for the next iteration until the termination condition is met.

[0020] To achieve the above objectives, another aspect of the present invention proposes an LDPC code structure optimization device for interference environments, comprising: The parity check matrix generation module is used to construct the base matrix of the LDPC code and generate the corresponding parity check matrix through cyclic shift expansion. The parity check matrix corresponds to the Tanner diagram. The ring structure optimization module is used to perform initial optimization on the ring structure of the parity check matrix, increase the minimum ring length and reduce the number of small outer edge degree rings, and obtain the initially optimized LDPC code structure. The stopping set evaluation module is used to search for the stopping set corresponding to each candidate cyclic shift value based on the initially optimized LDPC code structure, and to perform weighted processing on stopping sets of different sizes to calculate the normalized stopping set evaluation index. The iterative optimization output module is used to optimize the selection of cyclic shift values ​​based on the normalized stopping set evaluation index under the constraint of ensuring that the performance of the ring structure does not deteriorate. It iteratively executes the stopping set search and optimization process until the preset termination condition is met, and outputs the optimized LDPC code structure.

[0021] This invention discloses an LDPC code structure optimization method and apparatus for interference environments. It initially optimizes the ring structure of the parity-check matrix, increases the minimum ring length, and reduces the number of small outer-degree rings. Based on this, it searches for the stopping sets corresponding to each candidate cyclic shift value, weights different sizes of stopping sets, and calculates a normalized evaluation index. Under the constraint of ensuring no degradation in ring structure performance, iteratively optimizes the cyclic shift value, effectively solving the decoding performance degradation problem caused by neglecting the influence of stopping sets in existing technologies. It achieves an integrated solution for the entire process from ring structure optimization and stopping set evaluation to cyclic shift value selection, significantly improving the decoding reliability and robustness of LDPC codes in interference environments and enhancing the engineering applicability of the optimization method under actual channel conditions.

[0022] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0023] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a flowchart of an LDPC code structure optimization method for interference environments according to an embodiment of the present invention; Figure 2 This is a flowchart of the LDPC code stop set search algorithm according to an embodiment of the present invention; Figure 3 This is a flowchart of the QC-LDPC code construction algorithm based on the simulated annealing algorithm according to an embodiment of the present invention. Figure 4 This is a performance comparison chart of the QC-LDPC code construction algorithm based on the method of the present invention according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of an LDPC code structure optimization device for interference environments according to an embodiment of the present invention. Detailed Implementation

[0024] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0025] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0026] The following description, with reference to the accompanying drawings, describes an LDPC code structure optimization method and apparatus for interference environments, based on an embodiment of the present invention.

[0027] The core idea of ​​this invention is to address the performance degradation of LDPC codes due to stop sets under interference environments. It constructs a code structure optimization method covering the entire process of "base matrix construction—ring structure optimization—stop set search—cyclic shift value optimization," thus solving the performance deviation caused by traditional designs that only focus on ring length while ignoring the impact of stop sets. First, a base matrix is ​​constructed and a parity check matrix is ​​generated through cyclic shift expansion. Based on this, the ring structure is initially optimized by increasing the minimum ring length and reducing the number of small outer-degree rings. For the optimized structure, the stop sets corresponding to each candidate cyclic shift value are searched, and a normalized evaluation index is obtained by weighting stop sets of different sizes. Under the constraint of ensuring that the ring structure performance does not deteriorate, the cyclic shift value is iteratively optimized according to this index until the termination condition is met. Through the above technical path that integrates "ring length constraint—stop set quantification—iterative optimization," this invention transforms the traditional empirical design based on a single ring length criterion into an intelligent optimization system that can adapt to interference environments, fully exploit the structural features of the Tanner graph, and achieve synergistic enhancement of anti-stop set capabilities, significantly improving the decoding reliability and engineering applicability of LDPC codes under interference conditions.

[0028] The following description, with reference to the accompanying drawings, describes an LDPC code structure optimization method and apparatus for interference environments, based on an embodiment of the present invention.

[0029] like Figure 1 As shown, the LDPC code structure optimization method for interference environments according to the present invention includes the following steps: S1, construct the basis matrix of the LDPC code, and generate the corresponding parity check matrix through cyclic shift expansion, wherein the parity check matrix corresponds to the Tanner graph; S2, perform initial optimization on the ring structure of the parity check matrix, increase the minimum ring length and reduce the number of small outer edge degree rings to obtain the initially optimized LDPC code structure; S3. Based on the initially optimized LDPC code structure, search for the stopping set corresponding to each candidate cyclic shift value, and perform weighted processing on stopping sets of different sizes to calculate the normalized stopping set evaluation index. S4, under the constraint of ensuring that the performance of the ring structure does not deteriorate, optimize the selection of the cyclic shift value according to the normalized stopping set evaluation index, iteratively execute the stopping set search and optimization process until the preset termination condition is met, and output the optimized LDPC code structure.

[0030] In general, existing LDPC code construction methods mainly improve decoding performance by optimizing the ring structure in the Tanner graph (such as increasing the minimum ring length) and improving the external edge degree (EMD). However, in some frequency band interference, narrow band interference or other non-uniform interference environments, the log-likelihood ratio (LLR) corresponding to some bits tends to zero, which easily forms a stopping set structure. This causes the belief propagation (BP) decoding to fail to update information effectively, resulting in decoding failure.

[0031] The core of this invention lies in constructing an evaluation index to characterize the impact of the stop set on decoding performance, and introducing this evaluation index into the LDPC code construction process, thereby suppressing the adverse effects of the stop set while optimizing the ring structure.

[0032] During the optimization process, a multi-objective optimization mechanism is introduced to improve the graph structure characteristics while suppressing the impact of unfavorable structures on decoding performance. The optimization process can be implemented using various heuristic or stochastic optimization strategies, including but not limited to greedy search, random perturbation methods, and heuristic optimization algorithms, thereby obtaining an LDPC code structure with superior overall performance.

[0033] Therefore, this invention provides an LDPC code structure optimization method for interference environments, which optimizes the stop set distribution while ensuring ring structure performance (such as minimum ring length); improves the decoding success rate and anti-interference capability of LDPC codes under partial frequency band interference, narrowband interference, or other non-uniform interference environments; and constructs a multi-objective optimization framework that takes into account both structural performance and decoding performance. To achieve the above objectives, the specific steps of the method adopted in this invention are as follows: Step 1: Initial LDPC code construction.

[0034] Construct the LDPC code base matrix B, select a lifting factor, and obtain the corresponding parity-check matrix H through cyclic shifting. The Tanner graph corresponding to the parity-check matrix includes a set of variable nodes and a set of parity nodes.

[0035] Step 2: Initial optimization of the ring structure and EMD.

[0036] Without considering the stopping set, the LDPC code structure undergoes initial optimization, including: searching for ring structures in the graph and calculating the minimum ring length (girth); calculating the external margin degree (EMD) corresponding to each ring; and using a heuristic optimization algorithm to optimize the cyclic shift values, maximizing the minimum ring length and minimizing the number of small EMD rings. This yields the initially optimized LDPC code structure.

[0037] Step 3: Stop the set search.

[0038] Based on the optimized LDPC code, a stopping set search algorithm is used to perform stopping set search on the Tanner graph, including: for each position to be optimized, traversing all possible cyclic shift values; for each candidate value, searching for the stopping set it participates in; and counting the number and size of different stopping sets (i.e., the number of variable nodes).

[0039] Step 4: Stop set weighting and normalization.

[0040] Assume the maximum range of the stopping set search is To reflect the different impacts of different stop sets on decoding performance, the stop sets are weighted: the minimum size of all stop sets is determined. For a scale of The stopping set is assigned the highest weight; for sets larger than [a certain size], [the following is also true]. The stopping set, whose weight decreases as the size increases, has the following possible implementation: (1) Determine the minimum size of the stopping set corresponding to all candidate cyclic shift values, denoted as . ; (2) For a scale of The stopping set is assigned a weight coefficient of 1; (3) For those with a scale greater than The stopping set, whose weight function is a monotonically decreasing function, for example:

[0041] (4) Based on the weighting coefficients, the number of stopping sets corresponding to each cyclic shift value is weighted and summed to obtain the normalized stopping set evaluation index; (5) The normalized stopping set evaluation index is used to optimize different cyclic shift values.

[0042] Since a smaller stop set has a greater impact on LDPC code decoding, the first step is to find all candidate cyclic shift values ​​at that position. The smallest stopping set in the stopping set that participates in the formation. In one embodiment, it is... ,Will The size of the stop set weight is reset to 1, ( +1) The weight coefficients of the size stopping sets are , ( +2) The weight coefficients of the large and small stopping sets are Calculated in this way The number of normalized stopping sets for each value.

[0043] The number of stopping sets corresponding to each candidate cyclic shift value is weighted and summed to obtain a normalized stopping set evaluation index. This index is used to measure the potential impact of different structures on decoding performance.

[0044] Step 5: Greedy optimization strategy.

[0045] Under the constraint of ensuring that the ring structure does not deteriorate, local optimization is performed: if a certain cycle shift value reduces the stopping set evaluation index without increasing the number of rings or reducing the minimum ring length, then the cycle shift value is selected first. If there is a candidate value that makes both the number of rings and the stopping set evaluation index optimal, then the value is directly adopted and the position is fixed.

[0046] In step 2, the ring and EMD value distributions of the LDPC code have already been optimized. Therefore, in subsequent optimization processes, it is important to ensure that the ring distribution is not disrupted, while also optimizing the stopping set distribution. First, a greedy algorithm is performed; if a value exists... If the number of cycles is less than or equal to the number of stopping sets before optimization, and the number of stopping sets is also less than the number of stopping sets before optimization, then the cyclic shift value at that position is replaced with n. If the number of cycles and stopping sets formed by the value n is both 0, then the value is directly saved, and no further heuristic optimization algorithm is performed on that position.

[0047] Step 6: Joint optimization using heuristic optimization algorithms.

[0048] To avoid greedy algorithms getting trapped in local optima, a heuristic optimization algorithm is introduced: an energy function is constructed based on the stopping set evaluation index and the change of the ring structure, and a poor solution is accepted with a certain probability, thereby expanding the search space.

[0049] One possible instance of this heuristic optimization algorithm is simulated annealing: Simulated annealing gradually decreases the temperature parameter as it iterates, causing the optimization process to gradually converge. When candidate values ​​exist that do not introduce new cycles, selection is prioritized from this subset to ensure that the minimum cycle length does not decrease. The specific process is as follows: if Among the candidate cyclic shift values, the number of cycles formed by the cyclic shift values ​​is 0, and the probability of each candidate cyclic shift value is:

[0050] otherwise

[0051] Step 7: Iterative optimization and termination.

[0052] Repeat steps 3 through 6, performing random or traversal optimization at different positions until one of the following conditions is met: the LDPC code structure remains unchanged across multiple iterations; or the preset maximum number of iterations is reached. The final output is the optimized LDPC code cyclic shift matrix.

[0053] Furthermore, considering signal transmission in environments with partial frequency band interference, narrowband interference, or other non-uniform interference, the detailed process of the method in this embodiment is as follows: (1) Construct the basis matrix B of the LDPC code and select the lifting factor Z to generate the parity check matrix H by cyclic shift expansion.

[0054] Specifically: the basis matrix B is a sparse matrix, and its elements represent the cyclic shift amount of the corresponding submatrix; for each non-zero element, a corresponding cyclic shift identity matrix is ​​generated according to its value; all submatrices are concatenated to obtain the complete parity check matrix H. The Tanner graph corresponding to the parity check matrix consists of a set of variable nodes and a set of parity check nodes.

[0055] (2) Based on the initial LDPC code, the graph structure is optimized to improve the performance of the ring structure. Specifically, this includes: Search for ring structures in the Tanner graph and count the minimum ring length; calculate the external edge degree (EMD) of each ring to measure the connectivity between the ring and external nodes; use simulated annealing to optimize the cycle shift values ​​so that: the minimum ring length is not lower than a preset threshold; and the number of rings with small EMDs is minimized. After optimization, the initially optimized LDPC code structure is obtained.

[0056] (3) Based on the optimized LDPC code described above, a search is performed on the stopping set. For example... Figure 2 As shown, it specifically includes: For each position to be optimized in the check matrix: enumerate all possible cyclic shift values ​​for that position (ranging from 0 to Z). 1) For each candidate cyclic shift value, construct the corresponding local Tanner graph. Search for the stopping sets formed in this structure and record the number of stopping sets and the size of each stopping set (number of variable nodes).

[0057] Taking matrix H as an example: let its lifting factor be Z=6:

[0058] A random position to be optimized is selected, in one embodiment the first position of the first row. This position may take the value (0, 1, 2, 3, 4, 5). The number of stopping sets and the number of cycles are calculated for each possible value of this position. As shown in Tables 1 and 2.

[0059] Table 1: Number of Stop Sets

[0060] Table 2: Number of rings

[0061] (4) To reflect the impact of different stop sets on decoding performance, the stop sets are weighted. One possible weighting method is:

[0062] Taking the results in Table 1 and Table 2 as examples: First, determine the minimum stopping set size in the current candidate set. For a scale of The stopping set is assigned the highest weight. Its size is greater than... The stopping set has a weight that decreases as the size increases. For each candidate cyclic shift value, the corresponding stopping set number is multiplied by the weight and summed to obtain a normalized stopping set evaluation index. This index is used to measure the merits of different candidate values.

[0063] Taking the result of step (3) as an example: the smallest stopping set among all values ​​is 24, therefore The weight of a 24-size stopping set is 1. The weight coefficient of a 25-size stopping set is... And so on for the rest. For The total number of normalized stopping sets is calculated for each value after multiplying by its weight. See Table 3.

[0064] Table 3: Number of Normalized Stop Sets

[0065] (5) Select the cyclic shift value while ensuring that the performance of the ring structure is not reduced.

[0066] The specific selection rules are as follows: if a candidate cyclic shift value reduces the stopping set evaluation index without increasing the number of cycles or reducing the minimum cycle length, then the value is selected first; if there is a candidate value that simultaneously satisfies "minimum number of cycles and minimum stopping set evaluation index", then the position is directly selected and fixed; if there is a candidate value that makes the number of cycles zero and the stopping set evaluation index optimal, then the value is directly selected and subsequent optimization is skipped.

[0067] Taking the results in Table 3 as an example: the optimized forward shift value is 4, after the search... The number of cycles is less than or equal to 4, and the number of stopping sets is also less, so a greedy approach is used to directly replace the position with 1.

[0068] (6) Based on greedy optimization, a heuristic optimization algorithm is introduced for joint optimization to avoid getting trapped in local optima. A possible example of this heuristic optimization algorithm is simulated annealing, such as... Figure 3 As shown: An energy function is constructed that comprehensively considers: the number of rings or changes in ring structure, and the stopping set evaluation index. A new solution is randomly selected from the candidate cyclic shift values. If the new solution is better than the current solution, it is accepted directly; if the new solution is worse than the current solution, it is accepted with a certain probability. The acceptance probability gradually decreases with the number of iterations. When there are candidate values ​​that do not introduce new rings, selection is prioritized from this candidate set.

[0069] Taking the results in Table 3 as an example: Since the number of rings 0, 1, 2, 4, and 5 is all 0, the following annealing probabilities are used:

[0070] As shown in Table 4: Table 4: Simulated Annealing Probabilities

[0071] (7) Iterative optimization: Repeat steps 3 to 6 to iteratively optimize different positions in the verification matrix. Specifically: In each round, a position to be optimized is selected randomly or sequentially, and then the cyclic shift value of that position is updated, along with the cyclic information and stopping set distribution of the global structure.

[0072] (8) Termination condition: The optimization process terminates when any of the following conditions are met: The LDPC code structure remained unchanged throughout multiple iterations, reaching the preset maximum number of iterations, and the stopping set evaluation index reached the preset threshold. The final output is the optimized LDPC code parity-check matrix.

[0073] Figure 4 The QC-LDPC code constructed using this algorithm and the 5G-LDPC code with the same code length and code rate achieve different interference probabilities in some frequency band interference channels. The minimum signal-to-noise ratio required to achieve a bit error rate. At a bit error rate of... At the same time, compared with 5G-LDPC codes, the method of this invention can reduce the signal-to-noise ratio requirement under the same conditions, thereby improving the system's anti-interference performance. It can achieve better anti-interference capabilities in some frequency bands.

[0074] This invention proposes a QC-LDPC code construction algorithm based on stop set search and optimization, targeting partial-band interference, narrowband interference, and other non-uniform interference environments. This method optimizes the cyclic shift matrix of the QC-LDPC code by jointly considering the ring structure characteristics of the LDPC code, the distribution characteristics of the stop sets, and the impact of stop sets on error performance during decoding. This effectively reduces the number of small-scale stop sets and improves the decoding performance and anti-interference capability of the LDPC code under complex interference environments. Compared with traditional construction methods that optimize only based on minimum ring length or EMD value, this invention further improves the performance of the LDPC code in the error plateau region and enhances its reliability and robustness under non-uniform channel conditions.

[0075] In addition, this invention is not only applicable to some frequency band interference channels, but can also be extended to narrowband interference, multi-tone interference, burst interference, frequency selective fading channels and other communication scenarios with local high bit error rate characteristics. It can also be applied to the design of structured LDPC codes in fields such as 5G communication, satellite communication, deep space communication, storage systems and high-speed data transmission.

[0076] To achieve the above invention, such as Figure 5 As shown, this embodiment also provides an LDPC code structure optimization device 10 for interference environments, the device 10 including: The parity check matrix generation module 100 is used to construct the base matrix of the LDPC code and generate the corresponding parity check matrix through cyclic shift expansion. The parity check matrix corresponds to the Tanner diagram.

[0077] The ring structure optimization module 200 is used to perform initial optimization on the ring structure of the parity check matrix, increase the minimum ring length and reduce the number of small outer edge degree rings, and obtain the initially optimized LDPC code structure.

[0078] The stopping set evaluation module 300 is used to search for the stopping set corresponding to each candidate cyclic shift value based on the initially optimized LDPC code structure, and to perform weighted processing on stopping sets of different sizes to calculate the normalized stopping set evaluation index.

[0079] The iterative optimization output module 400 is used to optimize the selection of cyclic shift values ​​based on the normalized stopping set evaluation index under the constraint of ensuring that the performance of the ring structure does not deteriorate. It iteratively executes the stopping set search and optimization process until the preset termination condition is met, and outputs the optimized LDPC code structure.

[0080] The device of this invention effectively solves the problem of degraded decoding performance caused by neglecting the influence of the stopping set in existing designs. It realizes multi-objective collaborative optimization of ring length constraint and stopping set suppression, significantly improves the decoding reliability and robustness of LDPC codes in interference environments, and enhances the adaptability and engineering practical value of the optimization device under complex channel conditions.

[0081] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0082] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

Claims

1. A method for optimizing LDPC code structure in interference environments, characterized in that, include: Construct the basis matrix of the LDPC code and generate the corresponding parity check matrix through cyclic shift expansion. The parity check matrix corresponds to the Tanner graph. The ring structure of the parity check matrix is ​​initially optimized by increasing the minimum ring length and reducing the number of small outer edge degree rings to obtain the initially optimized LDPC code structure. Based on the initially optimized LDPC code structure, the stopping sets corresponding to each candidate cyclic shift value are searched, and the stopping sets of different sizes are weighted to calculate the normalized stopping set evaluation index. Under the constraint of ensuring that the performance of the ring structure does not deteriorate, the cyclic shift value is optimized according to the normalized stopping set evaluation index, and the stopping set search and optimization process is iteratively executed until the preset termination condition is met, and the optimized LDPC code structure is output.

2. The method as described in claim 1, characterized in that, The basis matrix for constructing the LDPC code is then used to generate the corresponding parity-check matrix through cyclic shift expansion. The parity-check matrix corresponds to a Tanner graph, including: Construct a sparse basis matrix B, where each element of the basis matrix B represents the cyclic shift of the corresponding submatrix; Select a boosting factor Z, and generate a corresponding cyclic shift identity matrix for each non-zero element in the basis matrix B; By concatenating all the submatrices, we obtain the complete parity matrix H.

3. The method as described in claim 1, characterized in that, The initial optimization of the ring structure of the parity-check matrix, by increasing the minimum ring length and reducing the number of small outer edge degree rings, yields the initially optimized LDPC code structure, including: Search for ring structures in the Tanner graph and count the minimum ring length; Calculate the external edge degree of each ring to measure the degree of connection between the ring and external nodes; The simulated annealing algorithm is used to optimize the cyclic shift value so that the minimum ring length is not lower than the preset threshold, while reducing the number of small outer edge rings, thus obtaining the initial optimized LDPC code structure.

4. The method as described in claim 1, characterized in that, Search for the stopping set corresponding to each candidate cyclic shift value, including: For each position to be optimized in the verification matrix, enumerate all possible cyclic shift values ​​for the corresponding position to be optimized, with the value range from 0 to Z-1; For each candidate cyclic shift value, construct a corresponding local Tanner plot; Search the stopping sets formed in the local Tanner graph, and record the number of stopping sets and the size of each stopping set, where the size is the number of variable nodes.

5. The method as described in claim 4, characterized in that, Weighted processing of stopping sets of different sizes includes: Determine the minimum size of the stopping set corresponding to all candidate cyclic shift values, denoted as . ; For a scale of The stopping set is assigned a weight coefficient of 1; for sets larger than 1, the weight coefficient is 1. The stopping set is determined using a monotonically decreasing weight function: Where τ is the maximum range of the stopping set search, and s is the size of the stopping set; The normalized stopping set evaluation index is obtained by weighting and summing the number of stopping sets corresponding to each cyclic shift value based on the weighting coefficients.

6. The method as described in claim 1, characterized in that, The cyclic shift value is optimized based on the normalized stopping set evaluation index, including: If a candidate cyclic shift value reduces the stopping set evaluation index without increasing the number of cycles or decreasing the minimum cycle length, then the corresponding value shall be selected first. If there exists a candidate value that makes the number of cycles zero and the stopping set evaluation index optimal, then the corresponding value is directly selected and subsequent optimization is skipped.

7. The method as described in claim 6, characterized in that, Also includes: Construct an energy function that comprehensively reflects the changes in the number or structure of loops, as well as the evaluation index of stopping sets; A new solution is randomly selected from the candidate cyclic shift values. If the new solution is better than the current solution, it is likely to be accepted directly. If the new solution is worse than the current solution, it is less likely to be accepted. The probability of accepting inferior solutions is gradually reduced as the number of iterations increases, and when there are candidate values ​​that do not increase the number of cycles, the corresponding candidate set is selected first.

8. The method according to claim 1, characterized in that, The preset termination conditions include at least one of the following: The LDPC code structure remains unchanged throughout the iterative process, across multiple iterations. The number of iterations has reached the preset maximum number of iterations; Stop the evaluation index when it reaches the preset threshold.

9. The method as described in claim 1, characterized in that, The iterative process of stopping set search and optimization includes: In each iteration, the position to be optimized is selected randomly or sequentially, the cyclic shift value of the corresponding position to be optimized is updated, and the cyclic information and stopping set distribution of the global structure are updated synchronously. The updated LDPC code structure will be used as the starting point for the next iteration until the termination condition is met.

10. An LDPC code structure optimization device for interference environments, characterized in that, include: The parity check matrix generation module is used to construct the base matrix of the LDPC code and generate the corresponding parity check matrix through cyclic shift expansion. The parity check matrix corresponds to the Tanner diagram. The ring structure optimization module is used to perform initial optimization on the ring structure of the parity check matrix, increase the minimum ring length and reduce the number of small outer edge degree rings, and obtain the initially optimized LDPC code structure. The stopping set evaluation module is used to search for the stopping set corresponding to each candidate cyclic shift value based on the initially optimized LDPC code structure, and to perform weighted processing on stopping sets of different sizes to calculate the normalized stopping set evaluation index. The iterative optimization output module is used to optimize the selection of cyclic shift values ​​based on the normalized stopping set evaluation index under the constraint of ensuring that the performance of the ring structure does not deteriorate. It iteratively executes the stopping set search and optimization process until the preset termination condition is met, and outputs the optimized LDPC code structure.

Images