Decoding method and device based on LDPC (Low Density Parity Check) and electronic equipment

Abstract

The invention provides an LDPC (Low Density Parity Check)-based decoding method and device and electronic equipment. The method comprises the following steps: acquiring an optimization model corresponding to LDPC decoding based on a received signal; transmitting the optimization model to a quantum annealing device to execute multiple times of calculation to obtain a plurality of first decoding results; obtaining a second decoding result, wherein the second decoding result is screened out from the plurality of first decoding results according to a constraint condition; and performing maximum likelihood decoding on the second decoding result to obtain a final decoding result. The quantum annealing algorithm is used for replacing a belief propagation algorithm to decode the LDPC code, so that the method has the advantages of higher global optimization capability, higher convergence speed, difficulty in being influenced by a loop and more stable performance.

Description

Technical Field

[0001] The present invention belongs to the technical field of quantum computers, and particularly relates to a decoding method, device and electronic device based on LDPC. Background Art

[0002] LDPC decoding (Low-Density Parity-Check Code) is a class of linear block codes with a sparse parity-check matrix and is widely used in the communication field. LDPC decoding usually uses the belief propagation algorithm to perform multiple iterations to obtain the correct decoding result. However, the belief propagation algorithm has the characteristics of high computational complexity, slow convergence speed and easy to fall into a cycle, resulting in unstable performance, being easily affected by the initial conditions and the number of iterations, and may lead to decoding failure or a high error rate. Summary of the Invention

[0003] The object of the present invention is to propose a decoding method, device and electronic device based on LDPC, which solves the problems of decoding failure or high error rate by using the quantum annealing algorithm instead of the belief propagation algorithm.

[0004] To this end, in the first aspect, the present invention provides a decoding method based on LDPC, including the following steps:

[0005] Obtain an optimization model corresponding to LDPC decoding based on the received signal;

[0006] Transmit the optimization model to a quantum annealer to perform multiple calculations to obtain multiple first decoding results;

[0007] Obtain a second decoding result, where the second decoding result is selected from multiple first decoding results according to the constraint conditions;

[0008] Perform maximum likelihood decoding on the second decoding result to obtain the final decoding result.

[0009] Optionally, the method for obtaining an optimization model corresponding to LDPC decoding based on the received signal includes:

[0010] Calculate the distance metric value and the constraint satisfaction metric value in the received signal;

[0011] Assign corresponding Lagrangian weights to the distance metric value and the constraint satisfaction metric value;

[0012] Combine the distance metric value and the constraint satisfaction metric value assigned with Lagrangian weights, and its expression is shown in Equation (1):

[0013] F = W1δ + W2L (1)

[0014] Among them, F is the constructed objective function, δ is the distance metric value, L is the constraint satisfaction metric value, and W1 and W2 are Lagrangian weights.

[0015] Optionally, by dynamically adjusting the Lagrangian weight W1 or the Lagrangian weight W2, a plurality of optimization models are correspondingly obtained, and the plurality of optimization models are transmitted to a quantum annealer for calculation to obtain the plurality of first decoding results.

[0016] Optionally, the method for dynamically adjusting the Lagrangian weight W1 or the Lagrangian weight W2 includes: keeping the Lagrangian weight W2 constant at 1 and dynamically adjusting the Lagrangian weight W1, or keeping the Lagrangian weight W1 constant at 1 and dynamically adjusting the Lagrangian weight W2.

[0017] Optionally, the method for obtaining the distance metric value is as follows:

[0018] Obtain the expected value of the received signal, which is represented by the following formula (2):

[0019]

[0020] where the binary variable q i represents the i-th bit of the decoded codeword, r is the received signal, and the parameter σ is used to adjust the signal-to-noise ratio SNR in the transmission channel;

[0021] Make the decoded codeword close to the received signal, which is represented by the following formula (3):

[0022]

[0023] where the distance metric δ is the distance metric value, and the binary variable q i represents the i-th bit of the decoded codeword. The minimum value of δ is the estimated value of the transmitted codeword, which is calculated using Pr(q i = 1|r i ).

[0024] Optionally, the method for obtaining the constraint satisfaction metric value includes:

[0025] For any check node c i , define the LDPC constraint function, which is represented by the following formula (4):

[0026]

[0027] where h is the parity-check matrix, h ij is an element in the parity-check matrix, represents the sum of all q ij＝1 that satisfy h j , c i represents the check node, and L e (ci ) represents the decision information indicating whether the bit of the check node is 0 or 1, L e (c i ) needs to be implemented with additional auxiliary qubits;

[0028] Minimize the constraint function to obtain the constraint satisfaction metric L, as shown in the following formula (5);

[0029] L = ∑ i L sat (c i ) (5)

[0030] where c i represents the check node. Minimizing L can make the check node satisfy the LDPC constraint.

[0031] Optionally, the method for performing maximum likelihood decoding on the second decoding result to obtain the final decoding result includes:

[0032] Based on the probability of observing the received signal from the transmitted encoded information,

[0033] Obtain the prior information for maximum likelihood decoding, where the prior information is the annealing energy of the candidate codewords in the second decoding result or the frequency of occurrence of the candidate codewords;

[0034] Based on the prior information, obtain the prior probability term;

[0035] Based on the probability of observing the received signal and the prior probability term, obtain the final maximum likelihood decoding;

[0036] Based on the maximum likelihood decoding, obtain the final decoding result.

[0037] Optionally, the prior probability term is as shown in the following formula (5):

[0038]

[0039] where x is the encoded word, E(x) is the annealing energy of x, and T is the adjustment parameter that controls the influence of energy on the prior.

[0040] Optionally, the maximum likelihood decoding is represented by the following formula (6):

[0041]

[0042] where is the solution of the maximum likelihood decoding.

[0043] Optionally, the final decoding result is represented by the following formula (7):

[0044]

[0045] Among them, is the final decoding result, argmax x p(y|x) is the maximum likelihood decoding, and P(x) is the prior probability.

[0046] In a second aspect, a decoding device based on LDPC is provided, including:

[0047] An optimization module for obtaining an optimization model for LDPC decoding based on the received signal;

[0048] A quantum annealer that transfers the optimization model to the quantum annealer to perform multiple calculations to obtain multiple first decoding results;

[0049] A screening module for obtaining a second decoding result, where the second decoding result is screened from multiple first decoding results according to constraint conditions;

[0050] A maximum likelihood optimization module for performing maximum likelihood decoding on the second decoding result to obtain a final decoding result.

[0051] In a third aspect, an electronic device is provided, including a memory and a processor;

[0052] The memory stores computer execution instructions;

[0053] The processor executes the computer execution instructions stored in the memory, so that the processor executes the described LDPC-based decoding method.

[0054] Beneficial effects:

[0055] (1) The present disclosure provides an LDPC-based decoding method, device, and electronic device. Using the quantum annealing algorithm instead of the belief propagation algorithm for LDPC code decoding has stronger global optimization ability, faster convergence speed, is not easily affected by loops, and has more stable performance.

[0056] (2) The present disclosure performs quadratic unconstrained binary optimization on the signal to obtain an optimization model for LDPC decoding, thereby converting the signal decoding problem into a problem of solving quadratic unconstrained binary, which is convenient for solving through the quantum annealing algorithm.

[0057] (3) The traditional quantum annealing algorithm usually selects the codeword with the lowest energy as the decoding result. Although it can avoid problems such as high computational complexity and slow convergence speed of the traditional belief propagation algorithm, the bit error rate of the decoding result is higher than that of the belief propagation algorithm. The present disclosure processes the second decoding result by performing maximum likelihood decoding on the second decoding result to reduce the bit error rate.

[0058] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present invention, nor is it used to limit the scope of the present invention. Other features of the present invention will become easily understandable through the following description. Description of the Drawings

[0059] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0060] Figure 1 It is a flowchart of a method for an embodiment of a decoding method based on LDPC in the present disclosure;

[0061] Figure 2 It is a flowchart of a process for an embodiment of a decoding method based on LDPC in the present disclosure;

[0062] Figure 3 It is a flowchart of a method included in obtaining a final decoding result for an embodiment of a decoding method based on LDPC in the present disclosure;

[0063] Figure 4 It is a schematic structural diagram of an embodiment of a quantum annealing decoding device in the present disclosure;

[0064] Figure 5 It is a schematic structural diagram of an embodiment of an electronic device in the present disclosure;

[0065] In the figure, 101 - signal acquisition module, 102 - optimization module, 103 - quantum annealer, 104 - screening module, 105 - maximum likelihood optimization module. Detailed Embodiments

[0066] To make the objectives, technical solutions, and advantages of the present application clearer, the following will clearly and completely describe the technical solutions in the present application with reference to the drawings in the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present application.

[0067] In the description, claims and the above-mentioned drawings of this application, terms such as "first", "second", "third", "fourth", etc. are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances. For example, without departing from the scope of this article, the first information can also be referred to as the second information, and similarly, the second information can also be referred to as the first information.

[0068] Depending on the context, as used herein, the word "if" can be interpreted as "when" or "while" or "in response to a determination".

[0069] Furthermore, as used herein, the singular forms "a", "an" and "the" are also intended to include the plural forms, unless the context indicates otherwise.

[0070] It should be further understood that the terms "comprising", "including" indicate the presence of features, steps, operations, elements, components, items, types, and / or groups, but do not exclude the presence, occurrence or addition of one or more other features, steps, operations, elements, components, items, types, and / or groups.

[0071] The term "or" and "and / or" used herein are interpreted inclusively, or means any one or any combination. Thus, "A, B or C" or "A, B and / or C" means "any one of the following: A; B; C; A and B; A and C; B and C; A, B and C". An exception to this definition occurs only when the combination of elements, functions, steps or operations is inherently mutually exclusive in some way.

[0072] LDPC decoding (Low-Density Parity-Check Code) is a class of linear block codes with sparse parity-check matrices and is widely used in the field of communication. LDPC decoding usually uses the belief propagation algorithm to perform multiple iterations to obtain the correct decoding result. However, the belief propagation algorithm has the following defects: (1) high computational complexity, especially for long codewords, which requires a large amount of computing and communication resources; (2) slow convergence speed, and multiple iterations may be required to approach the correct decoding result; (3) prone to cycles, especially in graph structures with loops, resulting in incorrect convergence or cycling in local minima. Its performance is unstable and is easily affected by initial conditions and the number of iterations, which may lead to decoding failure or a high error rate.

[0073] For this reason, the present disclosure decodes LDPC codes by replacing the belief propagation algorithm with a quantum annealing algorithm, which has stronger global optimization ability, faster convergence speed, is not easily affected by loops, and has more stable performance.

[0074] In a first aspect, the present disclosure provides asFigure 1-2 A decoding method based on LDPC is shown as follows, including the following steps:

[0075] S1. Obtain a signal;

[0076] Among them, the signal is to convert original information such as text, images, sounds, etc. into a signal form suitable for transmission in a specific channel or processing in a specific system according to certain coding rules.

[0077] Its core principle is to use different characteristics of the signal to represent different information elements. For example, in digital communication, binary coding is often used, and different combinations of "0" and "1" are used to represent various information. By sampling, quantifying, and coding the original information, it is converted into a series of discrete digital signals, which can be transmitted and stored in communication lines or storage media.

[0078] In one embodiment, the steps of encoding LDPC include that the core of LDPC encoding is to calculate the generation matrix G according to the parity-check matrix H, multiply the original information m by the generation matrix G, and perform modulo-2 operation to obtain the encoded codeword c corresponding to the original information m:

[0079] mG = c

[0080] During the signal transmission process, the encoded codeword will have a certain amount of noise, so the obtained signal r is:

[0081] r = c + n

[0082] Among them, n is the noise during the transmission process.

[0083] S2. Obtain an optimized model for LDPC decoding based on the received signal;

[0084] Perform quadratic unconstrained binary optimization on the signal to obtain an optimized model for LDPC decoding, thereby converting the signal decoding problem into a problem of solving quadratic unconstrained binary, which is convenient for solving through the quantum annealing algorithm.

[0085] It includes calculating the distance metric value and the constraint satisfaction metric value in the received signal;

[0086] Assign corresponding Lagrangian weights to the distance metric value and the constraint satisfaction metric value;

[0087] Combine the distance metric value and the constraint satisfaction metric value assigned with Lagrangian weights, and its expression is shown in Equation (1):

[0088] F = W1δ + W2L (1)

[0089] Among them, F is the constructed objective function, δ is the distance metric value, L is the constraint satisfaction metric value, and W1 and W2 are Lagrangian weights.

[0090] Among them, the method for obtaining the distance metric value is as follows:

[0091] Obtain the expected value of the received signal, which is represented by the following formula (2):

[0092]

[0093] Among them, the binary variable q t represents the i-th bit of the decoded codeword, r is the received signal, and the parameter σ is used to adjust the signal-to-noise ratio SNR in the transmission channel;

[0094] Make the decoded codeword close to the received signal, which is represented by the following formula (3):

[0095]

[0096] Among them, the distance metric δ is the distance metric value, and the binary variable q i represents the i-th bit of the decoded codeword. The minimum value of δ is the estimated value of the transmitted codeword, which is calculated using Pr(q t = 1|Γ t ). The transmitted codeword is the encoded codeword of the signal before transmission, and the received codeword is the encoded codeword obtained after the signal is transmitted. The received codeword has certain noise, and the smaller the degree of closeness between the transmitted codeword and the received codeword, the closer the received codeword is to the transmitted codeword.

[0097] Among them, the method for obtaining the constraint satisfaction metric value includes:

[0098] For any check node c i , an LDPC constraint function can be defined, which is represented by the following formula (4):

[0099]

[0100] Among them, h is the parity-check matrix, and h ij is an element in the parity-check matrix. denotes the sum of all q ij＝1 that satisfy h j , c i denotes the check node, and L e (c i ) represents the decision information of whether the bit of the check node is 0 or 1. L e (c i ) needs to be implemented using additional auxiliary qubits;

[0101] Minimize the constraint function to obtain the constraint satisfaction metric L, which is represented by the following formula (5);

[0102] L = ∑ i L sat (c i ) (5)

[0103] where c i represents a check node, and minimizing L can make the check node satisfy the LDPC constraint.

[0104] By minimizing the LDPC constraint function, the sum at this check node can be forced to be even, that is, the modulo-2 sum at this node is forced to be zero. The modulo-2 sum is an operation based on binary, which performs addition without considering carry.

[0105] Minimizing the constraint satisfaction metric value L can make this check node satisfy the LDPC constraint.

[0106] S3. Transmit the optimization model to a quantum annealer to perform multiple calculations to obtain multiple first decoding results;

[0107] By dynamically adjusting the Lagrangian weight W1 or the Lagrangian weight W2, multiple optimization models are correspondingly obtained. Transmit the multiple optimization models to the quantum annealer to perform calculations to obtain the multiple first decoding results.

[0108] The method for dynamically adjusting the Lagrangian weight W1 or the Lagrangian weight W2 includes: keeping the Lagrangian weight W2 constant at 1 and dynamically adjusting the Lagrangian weight W1, or keeping the Lagrangian weight W1 constant at 1 and dynamically adjusting the Lagrangian weight W2.

[0109] In one embodiment, by keeping the Lagrangian weight W2 constant at 1.0 and dynamically adjusting the Lagrangian weight W1, different objective functions F are obtained. Perform calculations on different objective functions F using the quantum annealing algorithm to obtain multiple first decoding results.

[0110] The Quantum Annealing Algorithm (QAA) is a heuristic algorithm for solving optimization problems. It draws on principles such as the annealing process and quantum tunneling effect in quantum mechanics. Using the quantum annealing algorithm instead of the belief propagation algorithm for LDPC code decoding has stronger global optimization capabilities. It can utilize the quantum tunneling effect to jump out of local minima and increase the probability of finding the global optimal solution. At the same time, the convergence speed is faster, it can approach the optimal solution in a shorter time, and the decoding delay is reduced. The quantum annealing algorithm can also handle graph structures with complex loops, is not easily affected by loops, and the decoding performance is more stable, not easily affected by initial conditions and the number of iterations. In addition, quantum annealing has natural parallel computing capabilities, can simultaneously process the evolution processes of a large number of quantum states, and improves the decoding efficiency.

[0111] S4. Obtain a second decoding result, which is selected from multiple first decoding results according to a constraint condition;

[0112] Among them, the constraint condition is a valid result that satisfies the LDPC constraint in the first decoding result.

[0113] S5. Perform maximum likelihood decoding on the second decoding result to obtain a final decoding result.

[0114] The quantum annealing algorithm usually selects the codeword with the lowest energy as the decoding result. Although this method can avoid problems such as high computational complexity and slow convergence speed of the traditional belief propagation algorithm, the decoding result, that is, the bit error rate, is higher than that of the belief propagation algorithm. In order to reduce the bit error rate, post-processing of performing maximum likelihood decoding is added after the quantum annealing algorithm is processed.

[0115] Among them, as Figure 3 shown, the methods included in obtaining the final decoding result include:

[0116] S51. Based on the probability of observing the received signal for the transmitted encoded information,

[0117] In an AWGN channel, the noise is usually modeled as a Gaussian distribution with a mean of zero and a variance of σ 2 . For a given received signal y, the likelihood function p(y|x) represents the probability of observing the received signal y when the encoded information x is transmitted, as follows:

[0118]

[0119] Among them, x is the encoded information, y is the received signal, and σ 2 is the variance.

[0120] S52. Obtain the prior information for maximum likelihood decoding, where the prior information is the annealing energy of candidate codewords in the second decoding result or the frequency of occurrence of candidate codewords;

[0121] Among them, the prior information is the energy value of the second decoding result or the frequency of occurrence of the same second decoding result. Among the second decoding results obtained after screening through multiple executions of the quantum annealing algorithm, each valid decoded codeword has a related energy value. In one embodiment, in order to effectively utilize the energy value, the energy value is used as the prior information in the framework of maximum likelihood. In another embodiment, there may be the same codewords among multiple second decoding results, and the frequency of occurrence of these same codewords can also be used as the prior information for maximum likelihood decoding.

[0122] S53. Based on the prior information, obtain a prior probability term;

[0123] The prior probability is P(x), which represents the energy value of the second decoding result or the frequency of occurrence of the same codeword in the second decoding result.

[0124]

[0125] Where x is the encoded word, E(x) is the annealing energy of x, and T is the adjustment parameter that controls the influence of energy on the prior.

[0126] S54. Obtain the final maximum likelihood decoding based on the probability of the observed received signal and the prior probability term;

[0127] Maximum likelihood decoding is a method of decoding the received codeword by selecting the encoded information with the highest probability of the received signal under a given channel model. This method relies on the principle of maximum likelihood estimation, that is, selecting the model parameters with the highest probability of the observed data.

[0128] Where the maximum likelihood decoding is represented by the following formula (6):

[0129]

[0130] Where is the solution of the maximum likelihood.

[0131] S55. Obtain the final decoding result based on the maximum likelihood decoding.

[0132] The final decoding result is represented by the following formula (7):

[0133]

[0134] Where is the final decoding result, argmax x p(y|x) is the maximum likelihood decoding, and P(x) is the prior probability.

[0135] By combining the candidate codeword set of the quantum annealing algorithm with the corresponding energy or frequency information, the original encoded codeword can be more accurately recovered within the framework of maximum likelihood estimation.

[0136] In a second aspect, as Figure 4 shown, a quantum annealing decoding device is provided, including:

[0137] A signal acquisition module 101 for acquiring a signal;

[0138] An optimization module 102 for obtaining an optimization model for LDPC decoding based on the received signal;

[0139] A quantum annealer 103 is used to transfer the optimization model to the quantum annealer to perform multiple calculations and obtain multiple first decoding results. Among them, the quantum annealer 103 is a device that uses the quantum annealing algorithm to solve complex optimization problems. In one embodiment, the quantum annealer 103 adopts a D-Wave annealer.

[0140] A screening module 104 is configured to obtain a second decoding result, where the second decoding result is screened from multiple first decoding results according to constraint conditions.

[0141] A maximum likelihood optimization module 105 is configured to perform maximum likelihood decoding on the second decoding result to obtain a final decoding result.

[0142] The specific implementation manner of the quantum annealing decoding device is the same as the foregoing quantum annealing decoding method, and will not be elaborated here.

[0143] In a third aspect, as Figure 5 shown, an electronic device is provided, which is characterized in that it includes: a memory, a processor;

[0144] The memory stores computer-executable instructions;

[0145] The processor executes the computer-executable instructions stored in the memory, so that the processor executes the above method.

[0146] In one embodiment, the electronic device 200 includes: at least one processor 201 and a memory 202. Optionally, the electronic device 200 further includes a communication component 203. Among them, the processor 201, the memory 202, and the communication component 203 are connected through a bus 204.

[0147] In a specific implementation process, at least one processor 201 executes the computer-executable instructions stored in the memory 202, so that at least one processor 201 executes the above method.

[0148] The specific implementation process of the processor 201 can refer to the above method embodiment, and its implementation principle and technical effect are similar, and will not be elaborated here in this embodiment.

[0149] In the above embodiments, it should be understood that the processor may be a central processing unit (CPU for short), or other general-purpose processors, digital signal processors (DSP for short), application specific integrated circuits (ASIC for short), etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The steps of the method disclosed in combination with the invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules in the processor.

[0150] The memory may include a random access memory (RAM), and may also include a non-volatile memory (NVM), such as at least one disk memory.

[0151] The bus may be an industry standard architecture (ISA) bus, a peripheral component interconnect (PCI) bus, an extended industry standard architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For the sake of convenience of representation, the bus in the drawings of this application is not limited to only one bus or one type of bus.

[0152] Finally, it should be noted that: After considering the specification and practicing the invention disclosed herein, those skilled in the art will easily think of other implementation manners of the present invention. The present invention aims to cover any variations, uses or adaptations of the present invention, which follow the general principles of the present invention and include the common general knowledge or conventional technical means in the technical field not disclosed in the present invention. It is not limited to the exact structures described above and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present invention is only limited by the appended claims.

Claims

1. A decoding method based on LDPC, characterized in that, It includes the following steps: Obtain an optimization model for LDPC decoding based on the received signal; Transmit the optimization model to a quantum annealer to perform multiple calculations to obtain multiple first decoding results; Obtain a second decoding result, where the second decoding result is selected from multiple first decoding results according to the constraint conditions; Perform maximum likelihood decoding on the second decoding result to obtain the final decoding result.

2. The decoding method based on LDPC according to claim 1, wherein The method for obtaining an optimization model for LDPC decoding based on the received signal includes: Calculate the distance metric value and the constraint satisfaction metric value in the received signal; Assign corresponding Lagrangian weights to the distance metric value and the constraint satisfaction metric value; Combine the distance metric value and the constraint satisfaction metric value with assigned Lagrangian weights, and its expression is as shown in formula (1): F = W1δ + W2L (1) Where, F is the constructed objective function, δ is the distance metric value, L is the constraint satisfaction metric value, and W1 and W2 are Lagrangian weights.

3. A decoding method based on LDPC according to claim 2, characterized in that, By dynamically adjusting the Lagrangian weight W1 or the Lagrangian weight W2, multiple optimization models are correspondingly obtained, and the multiple optimization models are transmitted to the quantum annealer to perform calculations to obtain the multiple first decoding results.

4. A decoding method based on LDPC according to claim 3, characterized in that, The method for dynamically adjusting the Lagrangian weight W1 or the Lagrangian weight W2 includes: keeping the Lagrangian weight W2 constant at 1 and dynamically adjusting the Lagrangian weight W1, or keeping the Lagrangian weight W1 constant at 1 and dynamically adjusting the Lagrangian weight W2.

5. A decoding method based on LDPC according to claim 2, characterized in that, The method for obtaining the distance metric value is as follows: Obtain the expected value of the received signal, which is represented by the following formula (2): where the binary variable q i represents the i-th bit of the decoded codeword, r is the received signal, and the parameter σ is used to adjust the signal-to-noise ratio SNR in the transmission channel; Make the decoded codeword close to the received signal, which is represented by the following formula (3): where the distance metric δ is a distance metric value, and the binary variable q i represents the i-th bit of the decoded codeword, and the minimum value of δ is the estimated value of the transmitted codeword, calculated using Pr(q i = 1|r i ).

6. The decoding method based on LDPC according to claim 2, characterized in that, The method for obtaining the constraint satisfaction metric value includes: For any check node c i , the LDPC constraint function is defined as shown in the following formula (4): Among them, h is the parity-check matrix, and h ij is an element in the parity-check matrix, denotes the sum of all q ij＝1 that satisfy h j , c i denotes the check node, and L e (c i ) represents the decision information of whether the bit of the check node is 0 or 1. L e (c i ) needs to be implemented with additional auxiliary qubits; Minimize the constraint function to obtain the constraint satisfaction metric L, which is represented by the following formula (5); L = ∑ i L sat (c i ) (5) Among them, c i represents a check node, and minimizing L can make the check node satisfy the LDPC constraint.

7. A decoding method based on LDPC according to claim 1, characterized in that, The method included in performing maximum likelihood decoding on the second decoding result to obtain the final decoding result includes: Based on the probability of observing the received signal from the transmitted encoded information, Obtain the prior information of the maximum likelihood decoding, where the prior information is the annealing energy of the candidate codewords in the second decoding result or the frequency of the occurrence of the candidate codewords; Based on the prior information, obtain the prior probability term; Based on the probability of observing the received signal and the prior probability term, obtain the final maximum likelihood decoding; Based on the maximum likelihood decoding, obtain the final decoding result.

8. A decoding method based on LDPC according to claim 7, characterized in that, Wherein, The prior probability term is as shown in the following formula (5): Where, x is the encoded word, E(x) is the annealing energy of x, and T is the adjustment parameter that controls the influence of energy on the prior.

9. A decoding method based on LDPC according to claim 8, characterized in that, The maximum likelihood decoding is represented by the following formula (6): Among them, is the maximum likelihood solution.

10. A decoding method based on LDPC according to claim 9, characterized in that, The final decoding result is represented by the following formula (7): Among them, is the final decoding result, argmax x p(y|x) is the maximum likelihood decoding, and P(x) is the prior probability.

11. An LDPC-based decoding device, characterized in that, It includes: An optimization module for obtaining an optimization model for LDPC decoding based on the received signal; A quantum annealer that transmits the optimization model to the quantum annealer to perform multiple calculations to obtain multiple first decoding results; A screening module for obtaining a second decoding result, where the second decoding result is selected from multiple first decoding results according to the constraint conditions; A maximum likelihood optimization module for performing maximum likelihood decoding on the second decoding result to obtain the final decoding result.

12. An electronic device, characterized in that, It includes a memory and a processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory, such that the processor executes a LDPC-based decoding method according to any one of claims 1-9.

Images