Method and apparatus for constructing multi-rate low density parity check codes

Abstract

A method and device for constructing a multi-rate low-density parity-check code, the embodiments of the present disclosure comprehensively consider the number of non-zero elements in the multi-rate check matrix and the decoding iteration number of each rate, introduce a global normalized throughput index, set the decoding iteration number to be less than a preset iteration number threshold, and set the column weight of the check matrix to be greater than a preset minimum column weight value, determine a base pattern matrix that has the maximum global normalized throughput index value and satisfies the first and second constraint conditions through a global search method, and construct a multi-rate low-density parity-check (LDPC) code based on the matrix, thereby reducing the hardware resource occupation and further improving the decoding throughput, ensuring that the generated multi-rate LDPC code has low hardware resource occupation, meeting the application requirements of resource-constrained scenarios, and improving the applicability of the LDPC code.

Description

Technical Field

[0001] This article relates to channel coding techniques, and more particularly to a method and apparatus for constructing a multi-rate low-density parity-check code. Background Technology

[0002] With the rapid development of emerging communication technologies such as high-speed satellite laser communication and millimeter-wave communication, modern communication systems are increasingly demanding high data throughput and low latency. In these high-speed communication applications, channel coding technology plays a crucial role, effectively ensuring reliable data transmission and protecting signals during transmission to address various channel interference and noise issues. Especially with the continuous increase in communication rates, achieving efficient and reliable data transmission in complex channel environments has become one of the core challenges in the development of communication technology.

[0003] Faced with the challenges of terminal miniaturization and limited hardware resources, existing multi-rate low-density parity-check (LDPC) coding schemes have certain limitations. Although multi-rate LDPC codes have been widely used in many communication standards due to their excellent decoding performance and flexible rate adjustment capabilities, these schemes often require significant hardware resources and multiple iterative decoding processes under the requirements of high data rates and low resource consumption. This makes it difficult to meet the demands of emerging communication scenarios for low resource consumption and high decoding throughput. Therefore, optimizing LDPC codewords under hardware constraints to achieve efficient data transmission has become a major research direction. Research in related technologies has found that the resource consumption of LDPC code decoder hardware implementation is approximately proportional to the number of non-zero elements in the parity-check matrix, while the decoding throughput is inversely proportional to the number of decoding iterations required. Based on this relationship, some studies have reduced the complexity of decoder hardware implementation by decreasing the number of non-zero elements in the LDPC parity-check matrix. However, this usually leads to an increase in the number of decoding iterations, resulting in a decrease in decoding throughput. On the other hand, some studies focus on optimizing the number of decoding iterations to improve decoding throughput, but often neglect the control of decoder hardware implementation complexity. Therefore, although research in related technologies has made progress in their respective optimization directions, it has failed to achieve comprehensive optimization of hardware resource consumption and decoding throughput, and most studies focus on the construction optimization of single-rate LDPC codewords, lacking comprehensive optimization of hardware implementation complexity and decoding throughput of multi-rate LDPC code decoders.

[0004] In summary, how to improve the encoding quality of multi-rate LDPC codes has become a problem that needs to be solved. Summary of the Invention

[0005] This application provides a method for constructing a multi-rate low-density parity-check code, including:

[0006] A global normalized throughput metric is constructed based on the number of non-zero elements in the parity check matrix and the number of decoding iterations for each code rate; wherein, the global normalized throughput metric is used to quantify the decoding throughput per unit of hardware resource consumption.

[0007] The fundamental graph matrix is ​​determined by a global search method when the global normalized throughput index is maximized and satisfies the first and second constraints. The first constraint is that the number of decoding iterations is less than a preset iteration threshold, and the second constraint is that the column weight of the parity check matrix is ​​greater than a preset minimum column weight.

[0008] Based on the determined fundamental schema matrix, a multi-rate low-density parity-check (LDPC) code is constructed. Furthermore, embodiments of this application also provide a computer storage medium storing a computer program, which, when executed by a processor, implements the aforementioned method for constructing multi-rate low-density parity-check codes.

[0009] Furthermore, embodiments of this application also provide a terminal, including: a memory and a processor, wherein the memory stores a computer program; wherein,

[0010] The processor is configured to execute computer programs in memory;

[0011] When the computer program is executed by the processor, it implements the construction method of the multi-rate low-density parity check code as described above.

[0012] Furthermore, embodiments of this application also provide a construction apparatus for a multi-rate low-density parity-check code, comprising: a construction unit, a search unit, and an encoding unit; wherein,

[0013] The construction unit is set as follows: a global normalized throughput index is constructed based on the number of non-zero elements in the parity check matrix and the number of decoding iterations for each code rate; wherein, the global normalized throughput index is used to quantify the decoding throughput under unit hardware resource consumption;

[0014] The search unit is set to: determine the fundamental graph matrix when the global normalized throughput index is maximized and satisfies the first and second constraints through a global search method; wherein, the first constraint is that the number of decoding iterations is less than a preset iteration number threshold, and the second constraint is that the column weight of the parity check matrix is ​​greater than a preset minimum column weight.

[0015] The encoding unit is set as follows: construct a multi-rate parity check LDPC code based on the determined fundamental model matrix.

[0016] This embodiment comprehensively considers the number of non-zero elements in the multi-rate check matrix and the number of decoding iterations for each rate. It introduces a global normalized throughput index, sets a first constraint that the number of decoding iterations is less than a pre-set threshold, and sets a second constraint that the column weight of the check matrix is ​​greater than a pre-set minimum column weight. A global search method is used to determine the fundamental graph matrix that maximizes the global normalized throughput index and satisfies the first and second constraints. Multi-rate check (LDPC) codes are constructed based on the determined fundamental graph matrix. This reduces hardware resource consumption while further improving decoding throughput, ensuring that the generated multi-rate LDPC codes have low hardware resource consumption. This meets the practical application requirements in resource-constrained scenarios and improves the applicability of the constructed multi-rate check matrix.

[0017] Other features and advantages of this application will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the application. Other advantages of this application can be realized and obtained by means of the solutions described in the description and the accompanying drawings. Attached Figure Description

[0018] The accompanying drawings are used to provide an understanding of the technical solutions of this application and constitute a part of the specification. They are used together with the embodiments of this application to explain the technical solutions of this application and do not constitute a limitation on the technical solutions of this application.

[0019] Figure 1 This is a flowchart illustrating a method for constructing a multi-rate low-density parity-check code according to an embodiment of this disclosure.

[0020] Figure 2 This is a structural block diagram of the apparatus for constructing a multi-rate low-density parity check code according to an embodiment of this disclosure. Detailed Implementation

[0021] This application describes several embodiments, but these descriptions are exemplary and not restrictive, and it will be apparent to those skilled in the art that many more embodiments and implementations are possible within the scope of the embodiments described herein. Although many possible combinations of features are shown in the drawings and discussed in the detailed description, many other combinations of the disclosed features are also possible. Unless specifically limited, any feature or element of any embodiment may be used in combination with, or may replace, any feature or element of any other embodiment.

[0022] This application includes and contemplates combinations of features and elements known to those skilled in the art. The embodiments, features, and elements disclosed in this application can also be combined with any conventional features or elements to form unique inventive solutions. Any feature or element of any embodiment can also be combined with features or elements from other inventive solutions to form another unique inventive solution. Therefore, it should be understood that any feature shown and / or discussed in this application can be implemented individually or in any suitable combination. Therefore, the embodiments are not limited except by the limitations imposed by the appended claims and their equivalents. Furthermore, various modifications and changes can be made within the scope of the appended claims.

[0023] Furthermore, in describing representative embodiments, the specification may have presented methods and / or processes as a specific sequence of steps. However, the method or process should not be limited to the specific order of steps described herein, to the extent that it does not depend on such a specific order. As will be understood by those skilled in the art, other sequences of steps are also possible. Therefore, the specific order of steps set forth in the specification should not be construed as a limitation of the claims. Moreover, the claims concerning the method and / or process should not be limited to the steps performed in the written order, and those skilled in the art will readily understand that these orders can be varied and still remain within the spirit and scope of the embodiments of this application.

[0024] The inventors of this application have found that multi-rate LDPC code decoders in related technologies typically face the problem of high hardware resource consumption. At the same time, their numerous decoding iterations significantly limit the decoding throughput, making it impossible to meet the requirements of low resource consumption and high throughput while ensuring decoding performance under limited hardware resources.

[0025] Figure 1 This is a flowchart of a method for constructing a multi-rate low-density parity-check code according to an embodiment of this disclosure, as follows: Figure 1 As shown, it includes:

[0026] Step 101: Construct a global normalized throughput index based on the number of non-zero elements in the parity check matrix and the number of decoding iterations for each code rate; wherein, the global normalized throughput index is used to quantify the decoding throughput under unit hardware resource consumption.

[0027] Step 102: Determine the fundamental graph matrix when the global normalized throughput index is maximized and satisfies the first and second constraints using a global search method; wherein the first constraint is that the number of decoding iterations is less than a preset iteration threshold, and the second constraint is that the column weight of the parity check matrix is ​​greater than a preset minimum column weight.

[0028] Step 103: Construct a multi-rate parity check (LDPC) code based on the determined fundamental graph matrix.

[0029] This embodiment comprehensively considers the number of non-zero elements in the multi-rate check matrix and the number of decoding iterations for each rate. It introduces a global normalized throughput index, sets a first constraint that the number of decoding iterations is less than a pre-set threshold, and sets a second constraint that the column weight of the check matrix is ​​greater than a pre-set minimum column weight. A global search method is used to determine the fundamental graph matrix that maximizes the global normalized throughput index and satisfies the first and second constraints. Multi-rate check (LDPC) codes are constructed based on the determined fundamental graph matrix. This reduces hardware resource consumption while further improving decoding throughput, ensuring that the generated multi-rate LDPC codes have low hardware resource consumption. This meets the practical application requirements in resource-constrained scenarios and improves the applicability of the constructed multi-rate check matrix.

[0030] In one exemplary instance, the globally normalized throughput metric in this disclosure embodiment is: The expression is:

[0031]

[0032] In the formula, ∑ i,j b i,j w represents the sum of the elements in the fundamental schema matrix; d The throughput weighting factor is set in advance; D represents the number of bitrates in the multi-rate LDPC code; K d Indicates bit rate r d The corresponding information bit length; I d Indicates bit rate r d Decoding threshold constraint value The number of decoding iterations required is as follows. Indicated by LDPC codes for the basis graph matrix in The minimum number of iterations required for successful decoding. For bitrate r d The corresponding fundamental schema matrix, Indicates bit rate r d The corresponding decoding threshold constraint value; Indicates throughput.

[0033] In one exemplary instance, this disclosure embodiment The ∑ in this embodiment can be calculated using LDPC codeword performance analysis methods such as the EXIT algorithm; i,j b i,j It can represent the number of non-zero vectors.

[0034] In one exemplary instance, the first constraint of this disclosure embodiment is: I d ≤I max ,d=1,…,D;The second constraint is: 1 T b u -d v,min ≥0, u=1,…,U;

[0035] Among them, I max The number of decoding iterations I for the pre-set code rate d The maximum limit value, 1 T Represents length and b u A column vector whose elements are all 1s, b u Let 1 be the column vector to be optimized in the fundamental model matrix. T b u This indicates the goal of finding the column vector b to be optimized. u The elements and d v,min This represents the minimum column weight of the column vector to be optimized.

[0036] In one exemplary instance, the second constraint of this disclosure embodiment can be set based on the empirical criteria when designing LDPC codewords, that is, the column weight of the parity check matrix needs to be no less than a constant (generally 3), thereby ensuring that the minimum code distance of the constructed LDPC code increases linearly with the code length, and thus has better decoding performance.

[0037] In one exemplary instance, the throughput weighting coefficient of this disclosure embodiment is set according to the usage scenario analysis; for example, it may include the frequency setting of using multi-rate LDPC codewords in different scenarios.

[0038] In one exemplary instance, the global search method of this disclosure includes:

[0039] A global search method based on differential evolution algorithm; or,

[0040] A global search method based on simulated annealing.

[0041] In one exemplary instance, the global search method based on the differential evolution algorithm in this disclosure generally includes: initialization, individual mutation, individual crossover, and next-generation individual selection, etc.

[0042] In one exemplary instance, when the global search method of this disclosure is a global search method based on differential evolution algorithm, the number of individuals in the fundamental schema matrix is ​​S, the number of evolutionary rounds of the differential evolution algorithm is G, and the crossover probability is p. c Based on a global search method, the fundamental schema matrix is ​​determined when the globally normalized throughput metric reaches its maximum value and satisfies both the first and second constraints. This includes:

[0043] Randomly generate S initial candidate basis mode graph matrices. Subscript 0 indicates the initialization phase; the Q variables of each fundamental schema matrix are randomly selected from the set of non-negative integers {x}. min ,x min +1,…,x max}, x min and x max These represent the minimum and maximum values ​​of the variable to be optimized, respectively.

[0044] In the g-th iteration of the G-round evolution, the following processing is performed:

[0045] Based on the S candidate fundamental mode graph matrices Generate S variant basis mode graph matrices The subscript g represents the number of evolutionary iterations; where g = 0, 1, ..., G-1. for The q-th variable, t1, t2, t3 are three distinct integers randomly selected from the set {1, 2, ..., S}. The function Φ(x) returns the set of non-negative integers {x}. min ,x min +1,…,x max The integer closest to x in};

[0046] Based on the S candidate fundamental mode graph matrices and S variant fundamental mode graph matrices Generate S cross-basic mode graph matrices in, for The q-th variable, With probability p c Values With probability 1-p c Values

[0047] Based on the S candidate fundamental mode graph matrices and S cross-fundamental model matrices Generate the S candidate fundamental schema matrix in the (g+1)th round. Among them, the s-th candidate fundamental mode graph matrix The generation method is as follows: when an individual The corresponding global normalized throughput value is higher than (superior to) ),and The number of inequalities satisfying the first and second constraints is greater than or equal to When the number of inequalities satisfying the first and second constraints is given, let... Otherwise, when individuals The corresponding global normalized throughput value is less than or equal to And / or, The number of inequalities satisfying the first and second constraints is less than When the number of inequalities satisfying the first and second constraints is given, let...

[0048] Through G rounds of evolution, S candidate basis graph matrices are obtained.

[0049] Select the matrix with the largest normalized throughput from the S candidate matrix bases, and satisfy the first and second constraints.

[0050] The first constraint condition of this embodiment is: I d ≤I max ,d=1,…,D, taking values ​​of d from 1 to D respectively, the number of inequalities satisfying the first constraint condition can be obtained; the second constraint condition in this embodiment is: 1 T b u -d v,min ≥0, u=1,…,U, by taking the values ​​of u from 1 to U, we can obtain the number of inequalities that satisfy the second constraint condition;

[0051] In each iteration, the present invention prioritizes the selection of a parity check matrix with higher global normalized throughput to ensure that the optimization process gradually converges to the optimal solution. At the same time, by strictly maintaining the first and second constraints, it ensures that the number of constraints satisfied does not decrease during the optimization process, thereby generating a parity check matrix with the largest global normalized throughput and good decoding performance.

[0052] In one exemplary instance, embodiments of this disclosure construct a multi-rate parity-check (LDPC) code based on a determined fundamental schema matrix, including:

[0053] Based on the selected fundamental parity graph matrix, the LDPC fundamental parity graph expansion algorithm is used for two expansions. The first expansion factor is L1, used to eliminate elements greater than 1 in the fundamental parity graph matrix, resulting in a basic parity check matrix consisting entirely of 0 and 1 elements. The second expansion factor is L2, used to perform quasi-cyclic expansion on the basic parity check matrix, ultimately obtaining a multi-rate-check matrix code of size (mL1L2)×(nL1L2). The shape of the multi-rate-check matrix code in this embodiment is M. d ×N d M d =m d L1L2, N d =n d L1L2.

[0054] This disclosure also provides a computer storage medium storing a computer program, which, when executed by a processor, implements the above-described method for constructing a multi-rate low-density parity-check code.

[0055] This disclosure also provides a terminal, including: a memory and a processor, wherein the memory stores a computer program; wherein,

[0056] The processor is configured to execute computer programs in memory;

[0057] When a computer program is executed by a processor, it implements the construction method of the multi-rate low-density parity check code as described above.

[0058] Figure 2 This is a structural block diagram of the apparatus for constructing a multi-rate low-density parity-check code according to an embodiment of this disclosure, as shown below. Figure 2 As shown, it includes: a construction unit, a search unit, and an encoding unit; wherein,

[0059] The construction unit is set as follows: based on the number of non-zero elements in the parity check matrix and the number of decoding iterations for each code rate, a global normalized throughput index is constructed; wherein, the global normalized throughput index is used to quantify the decoding throughput under unit hardware resource consumption;

[0060] The search unit is set to: determine the fundamental graph matrix when the global normalized throughput index is maximized and the first and second constraints are satisfied by a global search method; wherein the first constraint is that the number of decoding iterations is less than a preset iteration threshold, and the second constraint is that the column weight of the parity check matrix is ​​greater than a preset minimum column weight.

[0061] The encoding unit is set as follows: construct a multi-rate parity check LDPC code based on the determined fundamental model matrix.

[0062] In one exemplary instance, the globally normalized throughput metric of this disclosure embodiment is: The expression is:

[0063]

[0064] In the formula, ∑ i,j b i,j w represents the sum of the elements in the fundamental schema matrix; d The throughput weighting factor is set in advance; D represents the number of bitrates in the multi-rate LDPC code; K d Indicates bit rate r d The corresponding information bit length; I d Indicates bit rate r d Decoding threshold constraint value The number of decoding iterations required is as follows. Indicated by LDPC codes for the basis graph matrix in The minimum number of iterations required for successful decoding, B rd For bitrate r d The corresponding fundamental schema matrix, Indicates bit rate r d The corresponding decoding threshold constraint value; Indicates throughput.

[0065] In one exemplary instance, the first constraint in this disclosure embodiment is: I d ≤I max ,d=1,…,D, the second constraint is: 1 T b u -d v,min ≥0, u=1,…,U;

[0066] Among them, I max The number of decoding iterations I for the pre-set code rate d The maximum limit value, 1 T Represents length and b u A column vector whose elements are all 1s, b u Let 1 be the column vector to be optimized in the fundamental model matrix. T b u This indicates the goal of finding the column vector b to be optimized. u The elements and d v,min This represents the minimum column weight of the column vector to be optimized.

[0067] In one exemplary instance, the global search method in this disclosure includes:

[0068] A global search method based on differential evolution algorithm; or,

[0069] A global search method based on simulated annealing.

[0070] In one exemplary instance, when the global search method of this disclosure is a global search method based on differential evolution algorithm, the number of individuals in the fundamental schema matrix is ​​S, the number of evolutionary rounds of the differential evolution algorithm is G, and the crossover probability is p. c The search unit is set as follows:

[0071] Randomly generate S initial candidate basis mode graph matrices. The Q variables of each fundamental schema matrix are randomly selected from the set of non-negative integers {x}. min ,x min +1,…,x max}, x min and x max These represent the minimum and maximum values ​​of the variable to be optimized, respectively.

[0072] In the g-th iteration of the G-round evolution, the following processing is performed:

[0073] Based on the S candidate fundamental mode graph matrices Generate S variant basis mode graph matrices Where g = 0, 1, ..., G-1, for The q-th variable, t1, t2, t3 are three distinct integers randomly selected from the set {1, 2, ..., S}. The function Φ(x) returns the set of non-negative integers {x}. min ,x min +1,…,x max The integer closest to x in};

[0074] Based on the S candidate fundamental mode graph matrices and S variant fundamental mode graph matrices Generate S cross-basic mode graph matrices in, for The q-th variable, With probability p c Values With probability 1-p c Values

[0075] Based on the S candidate fundamental mode graph matrices and S cross-fundamental model matrices Generate the S candidate fundamental schema matrix in the (g+1)th round. Where the s-th candidate fundamental mode graph matrix The generation method is as follows: when an individual The corresponding global normalized throughput value is higher than (Right now Superior ),and The number of inequalities satisfying the first and second constraints is greater than or equal to When the number of inequalities satisfying the first and second constraints is given, let... Otherwise, when individuals The corresponding global normalized throughput value is less than or equal to And / or, The number of inequalities satisfying the first and second constraints is less than When the number of inequalities satisfying the first and second constraints is given, let...

[0076] Through G rounds of evolution, S candidate basis graph matrices are obtained.

[0077] Select the matrix with the largest normalized throughput from the S candidate matrix bases, and satisfy the first and second constraints.

[0078] The following application examples briefly illustrate the embodiments of this disclosure. These application examples are only used to illustrate the embodiments of this disclosure and are not intended to limit the scope of protection of the embodiments of this disclosure.

[0079] Application Examples

[0080] The embodiments disclosed herein are configured as follows:

[0081] Number of bitrates D = 4, bitrate includes

[0082] The fundamental schema matrix B to be optimized has a shape of 4×20; the fundamental schema matrix B corresponding to code rates of 1 / 2, 2 / 3, 3 / 4, and 4 / 5 is... 1 / 2 B 2 / 3 B 3 / 4 B 4 / 5 The first 8, 12, 16 and 20 columns of B are respectively, and the lengths of the encoded information bits k1, k2, k3 and k4 are 4, 8, 12 and 16 respectively.

[0083] The first 3 columns of the fundamental schema matrix B are fixed columns, and the last 17 columns are columns to be optimized, which can be represented as:

[0084]

[0085] Where, non-negative integer x i Let i = 1, 2, ..., 68 be the variables to be optimized, and let x be the minimum value. min =0, maximum value x max =3, that is, 0≤x i ≤3.

[0086] The minimum column weight of the column to be optimized in the fundamental schema matrix B is d. v,min =3;

[0087] The decoding threshold constraint values ​​corresponding to code rates of 1 / 2, 2 / 3, 3 / 4, and 4 / 5 are respectively and The throughput weighting coefficient is set to w1 = w2 = w3 = w4 = 1 / 4.

[0088] The global normalized throughput metric is defined as:

[0089]

[0090] In the global search method based on differential evolution, the number of individuals in the schema graph matrix is ​​S = 700, the number of differential evolution rounds is G = 200, and the crossover probability is p. c =0.3.

[0091] The parity check matrix H is obtained by expanding the fundamental graph matrix B using a two-step lifting method, with the first expansion factor L1 = 4 and the second expansion factor L2 = 256; the parity check matrices H corresponding to code rates 1 / 2, 2 / 3, 3 / 4, and 4 / 5 are shown. 1 / 2 H 2 / 3 H 3 / 4 and H 4 / 5 The shapes are 4096×8192, 4096×12288, 4096×16384 and 4096×20480 respectively, and the information bit lengths are 4096, 8192, 12288 and 16384 respectively.

[0092] Under the above settings, the processing of this disclosed embodiment includes:

[0093] Step 1: Define the constrained optimization problem:

[0094] The optimization objective is:

[0095]

[0096] The first and second constraints are:

[0097] I d ≤25, d=1,…,4;

[0098]

[0099] Wherein, the bit rate r d Decoding threshold constraint value The number of decoding iterations required is I. d Obtained by the EXIT algorithm, i.e.

[0100] Step 2: Solve the constrained optimization problem:

[0101] Solving constrained optimization problems using the differential evolution algorithm includes:

[0102] 2.1) Algorithm initialization: Randomly generate 700 initial candidate basis model matrices. Each of the 68 variables in the fundamental schema matrix is ​​randomly selected from the set of non-negative integers {0, 1, 2, 3}.

[0103] 2.2) Individual Variation: In the g-th iteration (g = 0, 1, ..., 199), a variational basis model matrix is ​​generated. in The qth variable Determined by the following formula: In the formula, t1, t2, t3 are three different integers randomly selected from the set {1,2,…,700}, and the function Φ(x) returns the integer closest to x in the set of non-negative integers {0,1,2,3}.

[0104] 2.3) Individual Crossover: In the g-th iteration (g = 0, 1, ..., 199), a crossover basis model matrix is ​​generated. in, The qth variable Taking a value with a probability of 0.3 Taking a value with a probability of 0.7

[0105] 2.4) Next-generation individual selection: Generate the candidate fundamental schema matrix for the (g+1)th round. Among them, the s-th candidate fundamental mode graph matrix The generation method is as follows: when an individual The corresponding global normalized throughput value is higher than (Right now Superior ),and The number of inequalities satisfying the first and second constraints is greater than or equal to When the number of inequalities satisfying the first and second constraints is given, let... Otherwise, when individuals The corresponding global normalized throughput value is less than or equal to And / or, The number of inequalities satisfying the first and second constraints is less than When the number of inequalities satisfying the first and second constraints is given, let...

[0106] After repeating the evolutionary iteration 200 times, the final matrix of 700 candidate schemas is obtained. The fundamental schema matrix B that has the maximum normalized throughput and satisfies all inequality constraints is selected as the final optimization result.

[0107] Step 3, Selection of the final verification matrix:

[0108] The fundamental schema matrix B selected in step 2 is:

[0109]

[0110] The basis graph matrices for each code rate can be represented as follows:

[0111]

[0112] In this embodiment, the final parity check matrix H is obtained by two-step boosting of the fundamental schema matrix B using the progressive edgegrow (PEG) algorithm. The first boost (boost factor L1 = 4) is used to eliminate elements greater than 1 in the fundamental schema matrix, resulting in a fundamental parity check matrix consisting entirely of 0s and 1s. The second boost (boost factor L2 = 256) is used to perform cyclic matrix expansion on the fundamental matrix to obtain the final parity check matrix.

[0113] To illustrate the effectiveness of the low decoding complexity and high throughput multi-rate LDPC code of the encoding method of this disclosure, AR4JA LDPC code and Delay-limited LDPC code with the same code rate and information bit length are compared.

[0114] 1) Comparison of hardware implementation complexity:

[0115] Referring to Table 1, the number of "1" elements in the three codeword parity-check matrices are as follows:

[0116]

[0117]

[0118] Table 1

[0119] Compared to AR4JA LDPC codes and Delay-limited LDPC codes, the E codewords proposed in this embodiment have different values. total The values ​​were reduced by 3.85% and 19.35% respectively, with the lowest decoder hardware implementation complexity.

[0120] 2) Comparison of the number of iterations required for decoding:

[0121] The decoding uses the LDPC BP decoding algorithm; the reference signal-to-noise ratios for code rates of 1 / 2, 2 / 3, 3 / 4, and 4 / 5 are respectively... Set E b The attenuation values ​​for / N0 are Δ = 0, 0.1, 0.2, and 0.3 dB. Let the codewords with code rates of 1 / 2, 2 / 3, 3 / 4, and 4 / 5 be at E... b The / N0 values ​​are respectively Make the bit error rate (BER) less than 10 -6 The required number of BP decoding iterations is I1, I2, I3, and I4.

[0122] Tables 2 to 5 correspond to the number of iterations required for decoding Δ = 0 dB, Δ = 0.1 dB, Δ = 0.2 dB, and Δ = 0.3 dB, respectively. Based on simulation results, different E... bThe number of decoding iterations required for the three codewords at level / N0 is:

[0123] Typing <![CDATA[I1]]> <![CDATA[I2]]> <![CDATA[I3]]> <![CDATA[I4]]> AR4JA code 29 32 39 49 Delay-limited code 20 20 20 20 The codewords provided in the embodiments of this disclosure 23 21 23 25

[0124] Table 2

[0125] Typing <![CDATA[I1]]> <![CDATA[I2]]> <![CDATA[I3]]> <![CDATA[I4]]> AR4JA code 24 25 28 27 Delay-limited code 18 19 18 18 The codewords provided in the embodiments of this disclosure 20 18 17 19

[0126] Table 3

[0127] Typing <![CDATA[I1]]> <![CDATA[I2]]> <![CDATA[I3]]> <![CDATA[I4]]> AR4JA code 23 22 22 23 Delay-limited code 17 16 15 15 The codewords provided in the embodiments of this disclosure 18 16 15 16

[0128] Table 4

[0129] Typing <![CDATA[I1]]> <![CDATA[I2]]> <![CDATA[I3]]> <![CDATA[I4]]> AR4JA code 20 19 19 19 Delay-limited code 15 15 14 13 The codewords provided in the embodiments of this disclosure 17 14 13 14

[0130] Table 5

[0131] As can be seen from the data in the table above, under the same decoding performance requirements, the number of decoding iterations required for the extracted codewords is significantly less than that for AR4JA LDPC codes, and slightly higher than that for Delay-limited LDPC codes.

[0132] Global normalized throughput comparison:

[0133] Table 5 shows a comparison of the global normalized throughput of each codeword, taking into account both hardware implementation complexity and decoding iteration count.

[0134]

[0135] Table 6

[0136] When Δ = 0, 0.1, 0.2, 0.3 dB: Compared with AR4JA LDPC code, the global normalized throughput of the proposed LDPC codeword is improved by 76.44%, 51.22%, 47.48%, and 42.62%, respectively; compared with Delay-limited LDPC code, the global normalized throughput of the proposed LDPC codeword is improved by 6.43%, 23.64%, 20.20%, and 23.34%, respectively.

[0137] The above analysis shows that the LDPC code obtained by using the encoding method of this embodiment not only has a lower decoder hardware implementation complexity, but also requires fewer iterations for decoding, thereby maximizing the global normalized throughput, which has important application value for high-throughput communication scenarios with limited resources.

[0138] It will be understood by those skilled in the art that all or some of the steps, systems, or apparatuses disclosed above, and their functional modules / units, can be implemented as software, firmware, hardware, or suitable combinations thereof. In hardware implementations, the division between functional modules / units mentioned above does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed collaboratively by several physical components. Some or all components may be implemented as software executed by a processor, such as a digital signal processor or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit (ASIC). Such software may be distributed on a computer-readable medium, which may include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term "computer storage medium" includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically contain computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

Claims

1. A method of constructing a multi-rate low density parity check code, characterized by, The method comprises the following steps: According to the number of non-zero elements in the check matrix and the decoding iteration number of each code rate, a global normalized throughput index is constructed, wherein the global normalized throughput index is used to quantify the decoding throughput under unit hardware resource consumption; By a global search method, a base matrix is determined when the global normalized throughput index reaches the maximum value and meets the first constraint condition and the second constraint condition, wherein the first constraint condition is that the decoding iteration number is less than a pre-set iteration number threshold, and the second constraint condition is that the column weight of the parity check matrix is greater than a pre-set column weight minimum value; According to the determined base matrix, a multi-code rate parity check LDPC code is constructed. The global normalized throughput index is , The expression of the global normalized throughput index is: ； wherein denotes the sum of elements in the base matrix; is a pre-set throughput weighting coefficient; D denotes the number of code rates of the multi-rate LDPC code; denotes code rate corresponding information bit length; denotes code rate decoding threshold constraint value required decoding iteration number under , denotes the minimum iteration number required for the LDPC code with base matrix to be successfully decoded under , code rate corresponding base matrix denotes code rate corresponding decoding threshold constraint value; denotes throughput.

2. The construction method according to claim 1, wherein: The first constraint condition is: ; The second constraint condition is: ; wherein, is a preset maximum value of the number of iterations of the code rate decoding, is a preset maximum value of the number of iterations of the code rate decoding, is a column vector with all elements being 1, is a column vector with all elements being 1, is a column vector to be optimized in the base module matrix, is a column vector to be optimized in the base module matrix, is a sum of elements of the column vector to be optimized, is a column weight minimum value of the column vector to be optimized.

3. The method of construction according to any one of claims 1 to 2, wherein, The global search method comprises: A global search method based on a differential evolution algorithm; or A global search method based on simulated annealing.

4. The method of construction according to claim 3, wherein, When the global search method is a global search method based on a differential evolution algorithm, the number of base model graph matrix individuals is , the number of evolution rounds of the differential evolution algorithm is , the crossover probability is , and the base model graph matrix when the global normalized throughput index has the maximum value and meets the first constraint condition and the second constraint condition is determined based on the global search method. randomly generated one initial candidate basis pattern matrix ; each basis pattern matrix has one variable randomly taken from the set of non-negative integers , and respectively represent the minimum and maximum values of the variable to be optimized; In the G wheel evolution, in the i-th wheel iteration, the following processing is performed: wheel iteration, the following processing is performed: According to one candidate basis matrix generate one variant basis matrix ; wherein , is the th variable ; are three different integers randomly selected from the set the function returns the integer closest to in the set of non-negative integers ; According to one candidate base mode matrix , and one variant base mode matrix , generate one cross base mode matrix ; wherein, is the th variable of , takes value with probability , and takes value with probability , and takes value with probability . According to one candidate basis mode pattern matrix , and one cross basis mode pattern matrix , generate the first round one candidate basis mode pattern matrix ; wherein the first one candidate basis mode pattern matrix is generated in the following manner: when the global normalized throughput value corresponding to the individual is higher than the global normalized throughput value corresponding to the individual , and the number of inequalities satisfying the first constraint condition and the second constraint condition is greater than or equal to the number of inequalities satisfying the first constraint condition and the second constraint condition, let ; otherwise, when the global normalized throughput value corresponding to the individual is less than or equal to the global normalized throughput value corresponding to the individual , and / or, the number of inequalities satisfying the first constraint condition and the second constraint condition is less than the number of inequalities satisfying the first constraint condition and the second constraint condition, let ; By G wheel evolution, get S candidate base model matrix ; From the S candidate base matrices, a base matrix is selected, which has the maximum normalized throughput value and meets the first constraint condition and the second constraint condition.

5. A computer storage medium, wherein a computer program is stored in the computer storage medium, and the computer program is executed by a processor to implement the construction method of the multi-code rate low-density parity check code according to any one of claims 1 to 4.

6. A terminal comprising: A memory and a processor, wherein the memory stores a computer program; The processor is configured to execute the computer program in the memory; The computer program is executed by the processor to implement the construction method of the multi-code rate low-density parity check code according to any one of claims 1 to 4.

7. An apparatus for constructing a multi-rate low density parity check code, comprising: A construction unit, a search unit and an encoding unit are provided, wherein: The construction unit is configured to construct a global normalized throughput index according to the number of non-zero elements in the check matrix and the decoding iteration number of each code rate, wherein the global normalized throughput index is used to quantify the decoding throughput under unit hardware resource consumption; The search unit is configured to determine a base matrix when the global normalized throughput index reaches the maximum value and meets the first constraint condition and the second constraint condition by a global search method, wherein the first constraint condition is that the decoding iteration number is less than a pre-set iteration number threshold, and the second constraint condition is that the column weight of the parity check matrix is greater than a pre-set column weight minimum value; The encoding unit is configured to construct a multi-code rate parity check LDPC code according to the determined base matrix. The global normalized throughput index is , The expression of the global normalized throughput index is: ； wherein denotes the sum of elements in the base matrix; is a pre-set throughput weighting coefficient; D denotes the number of code rates of the multi-rate LDPC code; denotes the code rate corresponding information bit length; denotes the code rate the decoding iteration number required under the decoding threshold constraint value , , denotes the minimum iteration number required for the LDPC code with the base matrix to be successfully decoded under , is the code rate corresponding base matrix, denotes the code rate corresponding decoding threshold constraint value; denotes the throughput.

8. The construction device according to claim 7, wherein: The first constraint condition is: ; The second constraint condition is: ; wherein, is a preset maximum value of the number of iterations of the code rate decoding, is a preset maximum value of the number of iterations of the code rate decoding, is a column vector whose elements are all 1s, is a column vector whose elements are all 1s, is a column vector to be optimized in the base module matrix, is a column vector to be optimized in the base module matrix, is a column vector to be optimized in the base module matrix, is a column vector to be optimized in the base module matrix.

Images