Fast list decoding of special nodes of polar-adjusted convolutional codes

Abstract

The application provides a fast list decoding method for a special node of a polar adjusted convolution code, comprising the following steps: determining a special node in a decoding tree of list decoding of the polar adjusted convolution code; wherein the special node comprises a source node and a sequence of code rate zero or repetition code sub-nodes, and the special node is identified by a sequence set; calculating a log-likelihood ratio of the source node corresponding to the special node and determining a path number; calculating a metric of each path based on the log-likelihood ratio and the sequence set corresponding to the special node, reserving a path with the minimum metric, and the number of the reserved paths is a set list number; taking the reserved path as an input to decode the source node, obtaining an estimated code word of each path of the source node; and calculating an estimated output bit and an estimated information bit of the special node through the estimated code word. The special node is introduced into the successive cancellation list decoding of the polar adjusted convolution code, and the decoding algorithm constructed can reduce the decoding delay without changing the decoding performance.

Description

Technical Field

[0001] This invention belongs to the field of wireless communication technology. Background Technology

[0002] Polar codes are a channel coding scheme theoretically proven to achieve Shannon capacity and have been adopted by fifth-generation mobile communication systems (5G). In his Shannon talk at ISIT 2019, Arikan proposed a new polar coding scheme called polarization-adjusted convolutional (PAC) codes. This represents a significant breakthrough in polar coding, further improving its performance. For shorter code lengths, PAC codes outperform polar codes without CRC concatenation; while for longer code lengths, CRC-assisted PAC codes outperform CRC-assisted polar codes. The mainstream decoding algorithms for PAC codes are Fano sequential decoding and list decoding. Notably, under sequential decoding, the performance of PAC codes is very close to the dispersion bound approximation of binary additive white Gaussian noise (BIAWGN). In contrast, Successive Cancellation List (SCL) decoding is a non-backtracking tree search method with fixed complexity. Therefore, it has lower complexity in the worst case and advantages under low signal-to-noise ratio (SNR) conditions. List decoding, by maintaining a list to preserve multiple most reliable codeword sequences, greatly improves the error correction performance of canceled list SC decoding. By concatenating polar codes with Cyclic Redundancy Check (CRC), CRC-Aided SCL (CA-SCL) decoders can provide decoding performance close to that of maximum-likelihood (ML) decoders. However, the bit-by-bit sequential nature of SC decoding causes high decoding latency and low throughput, hindering its application in low-latency communication scenarios such as Ultra-Reliable Low-Latency Communication (URLLC). To achieve efficient decoding, a common approach is to parallelize the decoding process, i.e., decoding at the intermediate node level of the decoding tree rather than the leaf node level, thereby avoiding traversal of the entire code tree and improving decoding speed. To ensure that decoding performance does not degrade, node-level decoding needs to consider the constraints imposed by the distribution characteristics of information and frozen bits within the node, so that the decoding result is valid. Therefore, the higher the level of the decoding tree where the node is located, the higher the parallelism, the lower the decoding latency, and the higher the throughput. However, at the same time, the more bits a node contains, the more complex the constraints become, which makes it difficult to implement a lossless decoding algorithm. Summary of the Invention

[0003] The present invention aims to at least partially solve one of the technical problems in the related art.

[0004] Therefore, the purpose of this invention is to propose a fast list decoding method for special nodes of polarization-adjusted convolutional codes, which reduces time complexity without sacrificing decoding performance.

[0005] To achieve the above objectives, a first aspect of the present invention proposes a fast list decoding method for special nodes of polarization-adjusted convolutional codes, comprising:

[0006] Identify special nodes in the decoding tree of polarization-adjusted convolutional code list decoding; wherein, the special node includes a source node and a sequence of zero-rate or repeating code child nodes, and the special node is identified by a sequence set;

[0007] Calculate the log-likelihood ratio of the source nodes corresponding to the special nodes to determine the number of paths;

[0008] Based on the log-likelihood ratio and the sequence set corresponding to the special node, calculate the metric for each path, retain the path with the smallest metric, and retain the number of paths as set in the list.

[0009] Using the preserved path as input, the source node is decoded to obtain the estimated codeword for each path of the source node;

[0010] The estimated output bits and estimated information bits of the special node are calculated using the estimated codeword.

[0011] In addition, the fast list decoding method for special nodes of polarization-adjusted convolutional codes according to the above embodiments of the present invention may also have the following additional technical features:

[0012] Furthermore, in one embodiment of the present invention, the special node includes a polarization-adjusted convolutional code subcode consisting of a source node and a sequence of zero-rate or repeating code sub-nodes, located in the middle layer of the decoding tree of continuous cancellation decoding of polarization-adjusted convolutional codes, wherein the source node is located at the rightmost node of the special node and has arbitrary code rate and structure, and the length of the source node is N. r Satisfying N r ≥p-1, where p is the length of the convolution polynomial of the polarization-adjusted convolution code.

[0013] Furthermore, in one embodiment of the present invention, the sequence set includes:

[0014] For the m-th sequence set set up For functions that perform recursive computation, If the result is after a calculation, then Where s and r are the levels of the special node and the source node in the decoding tree, respectively, and F m The calculation is as follows:

[0015]

[0016] in, Denotes the Kronecker product, x s-a Let x be the first (p-1) bits of the a-th child node, for a child node x located at level t in the decoding tree. t The calculation can be performed as follows:

[0017]

[0018] Where q = 2t - p + 1, 0 q It is a zero vector of length q. It is a vector of length q and

[0019]

[0020] Furthermore, in one embodiment of the present invention, calculating the log-likelihood ratio of the source node corresponding to the special node includes:

[0021] Log-likelihood ratio based on special nodes The log-likelihood ratio of the source node to the m-th sequence set. The calculation is as follows:

[0022]

[0023] Furthermore, in one embodiment of the present invention, calculating the metric for each path includes:

[0024]

[0025]

[0026] in, and They are respectively and λ s,l A firm verdict.

[0027] Furthermore, in one embodiment of the present invention, the decoding of the source node includes:

[0028] If the source node has a special structure, the estimated codeword sequence of the source node can be calculated by the fast list decoding algorithm corresponding to its node;

[0029] If the source node has only a general structure, the estimated codeword sequence of the source node can be calculated by the traditional serial cancellation decoding list algorithm.

[0030] Further, in one embodiment of the present invention, the step of calculating the estimated output bits and estimated information bits of the special node using the estimated codeword includes:

[0031] Calculate the output bits:

[0032]

[0033] Calculate the estimated information bits:

[0034]

[0035] in, The estimated bits for a special node are calculated as follows:

[0036]

[0037] in, For polarization generating matrix and F is a polarized nucleus and SR node estimated information bits The calculation is as follows:

[0038]

[0039] To achieve the above objectives, a second aspect of the present invention provides a fast list decoding apparatus for special nodes of polarization-adjusted convolutional codes, comprising the following modules:

[0040] A filtering module is used to determine special nodes in the decoding tree of polarization-adjusted convolutional code list decoding; wherein, the special node includes a source node and a sequence of zero-rate or repeating code child nodes, and the special node is identified by a sequence set;

[0041] The first calculation module is used to calculate the log-likelihood ratio of the source node corresponding to the special node and determine the number of paths;

[0042] The second calculation module is used to calculate the metric of each path based on the log-likelihood ratio and the sequence set corresponding to the special node, and retain the path with the smallest metric. The number of retained paths is the set list number.

[0043] The decoding module is used to decode the source node with the reserved path as input to obtain the estimated codeword for each path of the source node;

[0044] The output module is used to calculate the estimated output bits and estimated information bits of the special node using the estimated codeword.

[0045] To achieve the above objectives, a third aspect of the present invention provides a computer device, characterized in that it includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements a fast list decoding method for special nodes of polarization-adjusted convolutional codes as described above.

[0046] To achieve the above objectives, a fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it implements a fast list decoding method for special nodes of a polarization-adjusted convolutional code as described above.

[0047] The fast list decoding method for special nodes of polar-adjusted convolutional codes proposed in this invention introduces a special node and makes full use of the properties of special nodes in polar-adjusted convolutional codes. A given special node can be determined by a sequence set, and the sequence set can be generated offline. In addition, the list decoding method improves the parallelism of decoding by considering the special node, and significantly reduces the time complexity of decoding without sacrificing decoding performance. Attached Figure Description

[0048] 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:

[0049] Figure 1 This is a flowchart illustrating a fast list decoding method for special nodes of polarization-adjusted convolutional codes provided in an embodiment of the present invention.

[0050] Figure 2 This is a schematic diagram of a special node binary tree structure provided in an embodiment of the present invention.

[0051] Figure 3 This diagram illustrates a comparison of the theoretical decoding performance of a fast list decoding algorithm for special nodes of polarization-adjusted convolutional codes provided in an embodiment of the present invention with that of a traditional list decoding algorithm.

[0052] Figure 4 This diagram illustrates a comparison of the theoretical decoding delays of a fast list decoding algorithm for special nodes of polarization-adjusted convolutional codes, provided in an embodiment of the present invention, and a traditional list decoding algorithm.

[0053] Figure 5 This is a flowchart illustrating a fast list decoding apparatus for special nodes of polarization-adjusted convolutional codes provided in an embodiment of the present invention. Detailed Implementation

[0054] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0055] The following describes a fast list decoding method for special nodes of polarization-adjusted convolutional codes according to embodiments of the present invention, with reference to the accompanying drawings.

[0056] Example 1

[0057] Figure 1 This is a flowchart illustrating a fast list decoding method for special nodes of polarization-adjusted convolutional codes provided in an embodiment of the present invention.

[0058] like Figure 1 As shown, the fast list decoding method for special nodes of polarization-adjusted convolutional codes includes the following steps:

[0059] S101: Determine the special nodes in the decoding tree of the polarization-adjusted convolutional code list decoding; wherein, the special node includes a source node and a sequence of zero-rate or repeating code child nodes, and the special node is identified by the sequence set;

[0060] Specifically, the continuous cancellation decoding process in S101 is as follows: The fast list decoding algorithm is based on fast continuous cancellation decoding, which can be viewed as a depth-first search process from the root node to the leaf node of the decoding tree. Two types of information are passed between nodes in the decoding tree: the first is the log-likelihood ratio sequence passed from the upper-level node to the lower-level node (specifically, at the root node, the decoding calculates the channel log-likelihood ratio sequence provided by the previous module); the second is the estimated codeword sequence passed from the lower-level node to the upper-level node. During the decoding process, the log-likelihood ratio is calculated level by level and passed to the next level node in turn. When decoding reaches the level of a special node, the estimated codeword sequence is directly calculated based on the log-likelihood ratio sequence of that special node. The estimated codeword sequence is calculated level by level and passed to the upper level node in turn until the root node. The above process is repeated until the decoding of the last leaf node is completed. Based on fast continuous cancellation decoding, list decoding splits the decoding path into parallel decoding branches and retains the most likely correct decoding path of the list number according to the path metric.

[0061] Furthermore, in one embodiment of the present invention, the special node includes a polarization-adjusted convolutional code subcode consisting of a source node and a sequence of zero-rate or repeating code sub-nodes, located in the middle layer of the decoding tree of continuous cancellation decoding of polarization-adjusted convolutional codes, wherein the source node is located at the rightmost node of the special node and has arbitrary code rate and structure, and the length of the source node is N. r Satisfying N r≥p-1, where p is the length of the convolution polynomial of the polarization-adjusted convolution code.

[0062] Furthermore, in one embodiment of the present invention, the sequence set includes:

[0063] For the m-th sequence set set up For functions that perform recursive computation, If the result is after a calculation, then Where s and r are the levels of the special node and the source node in the decoding tree, respectively, and F m The calculation is as follows:

[0064]

[0065] in, Denotes the Kronecker product, x s-a Let x be the first (p-1) bits of the a-th child node, for a child node x located at level t in the decoding tree. t The calculation can be performed as follows:

[0066]

[0067] Where q = 2t - p + 1, 0 q It is a zero vector of length q. It is a vector of length q and

[0068]

[0069] Once the specific nodes are determined, the sequence set is also determined. Furthermore, by enumerating the possible values ​​of the first p bits, the sequence set can be generated offline.

[0070] S102: Calculate the log-likelihood ratio of the source nodes corresponding to special nodes to determine the number of paths;

[0071] Further, in one embodiment of the present invention, calculating the log-likelihood ratio of the source node corresponding to the special node includes:

[0072] Log-likelihood ratio based on special nodes The log-likelihood ratio of the source node to the m-th sequence set. The calculation is as follows:

[0073]

[0074] S103: Based on the log-likelihood ratio and the sequence set corresponding to special nodes, calculate the metric for each path, retain the path with the smallest metric, and retain the number of paths as set in the list;

[0075] Furthermore, in one embodiment of the invention, calculating the metric for each path includes:

[0076]

[0077]

[0078] in, and They are respectively and λ s,l A firm verdict.

[0079] After obtaining the path metric for each path, select the path with the smallest number of paths.

[0080] S104: Using the preserved path as input, decode the source node to obtain the estimated codeword for each path of the source node;

[0081] Furthermore, in one embodiment of the present invention, decoding the source node includes:

[0082] If the source node has a special structure, the estimated codeword sequence of the source node can be calculated by the fast list decoding algorithm corresponding to its node;

[0083] If the source node has only a general structure, the estimated codeword sequence of the source node can be calculated using the traditional serial cancellation decoder list algorithm.

[0084] S105: Calculate the estimated output bits and estimated information bits of the special node by estimating the codeword.

[0085] Furthermore, in one embodiment of the present invention, calculating the estimated output bits and estimated information bits of a special node by estimating the codeword includes:

[0086] Calculate the output bits:

[0087]

[0088] Calculate the estimated information bits:

[0089]

[0090] in, The estimated bits for a special node are calculated as follows:

[0091]

[0092] in, For polarization generating matrix and F is a polarized nucleus and SR node estimated information bits The calculation is as follows:

[0093]

[0094] Embodiment 2

[0095] Figure 2 is a schematic diagram of a special node binary tree structure. For the node at the s-th layer in the successive cancellation decoding tree, λ is defined s,l [1:2 s and β s,l [1:2 s are respectively defined as the Log-likelihood Ratio (LLR) sequence and the codeword estimation sequence of the l-th path of the node. A type of special node located at the i-th node of the p-th layer in the decoding tree is defined, which consists of a source node and a sequence of Rate-0 (R0) or Repetition (REP) child nodes, where the source node is located at the rightmost end of the r-th layer, and the left node of each layer where r≤t<s is an R0 or REP child node, such that the special node is called a Sequence Repetition (SR) node.

[0096] Consider transmitting signals over an additive white Gaussian noise channel, with the modulation mode being Binary Phase Shift Keying (BPSK), and the code pattern being a polar-adjusted convolutional code of (128,64) constructed by RM construction, with the convolutional polynomial being c=(1,0,0,1,1) and the number of paths L=4. A fast list decoding method for special nodes of the proposed polar-adjusted convolutional code for this system includes the following steps:

[0097] S201: Construct the sequence set offline. For a given decoding binary tree that contains three of the special nodes. Taking a special node that contains two Rate-0 and one repetition code child node and has a source node length of 8 as an example, there are a total of 2 W =2 sequence sets. For the child node x at level t of the decoding tree t it can be calculated as follows:

[0098]

[0099] By enumerating the possible values of the first p bits of the SR node, x can be obtained s-1 ; while for other left child nodes, x t is as follows:

[0100]

[0101] Where η[t] is the value of the rightmost bit of the left sub-byte at level t. Therefore, F m The calculation is as follows:

[0102]

[0103] S202: Log-likelihood ratio λ based on special nodes s,l For each path and each sequence set, the log-likelihood ratio of the source node to the m-th (m∈{1,2}) sequence set. The calculation is as follows:

[0104]

[0105] At this point, there are a total of L·m = 8 paths.

[0106] S203: Calculate the path metrics for each of the 8 paths:

[0107]

[0108]

[0109] Select the four paths with the smallest path metrics.

[0110] S204: Decode the source node using the four paths selected in S203 as input. If the source node has a special structure, the estimated output codeword of the source node is calculated using the fast list decoding algorithm corresponding to its node; if the source node only has a general structure, the estimated output codeword of the source node is calculated using the traditional serial cancellation decoding list algorithm.

[0111] S205: Estimated output bits of the source node based on S204 The estimated output bits for each path of the m-th sequence set are calculated as follows:

[0112]

[0113] Output bits are estimated based on special nodes, and bits are estimated by SR nodes. The calculation is as follows:

[0114]

[0115] SR node estimated information bits for each path The calculation is as follows:

[0116]

[0117] This completes the fast list decoding of SR nodes.

[0118] Figure 3 This figure shows a comparison of the theoretical decoding performance of the fast list decoding algorithm based on special nodes of polarization-adjusted convolutional codes in this embodiment with that of the traditional list decoding algorithm. As can be seen from the figure, the fast list decoding algorithm based on SR nodes does not suffer any loss in error control performance compared to the traditional list decoding algorithm.

[0119] Figure 4 This figure shows a comparison of the theoretical decoding delays between the fast list decoding algorithm based on special nodes of polarization-adjusted convolutional codes and the traditional list decoding algorithm. As can be seen from the figure, the fast list decoding algorithm based on SR nodes significantly reduces decoding latency compared to the traditional list decoding algorithm, especially at low code rates.

[0120] The fast list decoding method for special nodes of polarization-adjusted convolutional codes proposed in this invention introduces a special node and makes full use of the properties of special nodes in polarization-adjusted convolutional codes. A given special node can be determined by a sequence set, and the sequence set can be generated offline. In addition, the list decoding method improves the parallelism of decoding by considering the special node, and significantly reduces the time complexity of decoding without sacrificing decoding performance.

[0121] To achieve the above embodiments, the present invention also proposes a fast list decoding device for special nodes of polarization-adjusted convolutional codes.

[0122] Figure 5 This is a schematic diagram of a fast list decoding device for special nodes of polarization-adjusted convolutional codes provided in an embodiment of the present invention.

[0123] like Figure 5 As shown, the fast list decoding device for special nodes of polarization-adjusted convolutional codes includes: a filtering module 100, a first calculation module 200, a second calculation module 300, a decoding module 400, and an output module 500, wherein...

[0124] The filtering module is used to identify special nodes in the decoding tree of polarization-adjusted convolutional code list decoding; wherein, a special node includes a source node and a sequence of zero-rate or repeating code child nodes, and the special node is identified by a set of sequences;

[0125] The first calculation module is used to calculate the log-likelihood ratio of the source nodes corresponding to special nodes and determine the number of paths;

[0126] The second calculation module is used to calculate the metric of each path based on the log-likelihood ratio and the sequence set corresponding to the special node, and retain the path with the smallest metric. The number of paths retained is the set list number.

[0127] The decoding module is used to decode the source node with the preserved path as input to obtain the estimated codeword for each path of the source node;

[0128] The output module is used to calculate the estimated output bits and estimated information bits of special nodes using the estimated codewords.

[0129] To achieve the above objectives, a third aspect of the present invention provides a computer device, characterized in that it includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the fast list decoding method for special nodes of polarization-adjusted convolutional codes as described above.

[0130] To achieve the above objectives, a fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it implements the fast list decoding method for special nodes of polarization-adjusted convolutional codes as described above.

[0131] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is 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.

[0132] 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.

[0133] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A fast list decoding method for special nodes of polarization-adjusted convolutional codes, characterized in that, Includes the following steps: Identify special nodes in the decoding tree for polar-adjusted convolutional code list decoding; wherein, the special node comprises a polar-adjusted convolutional code sub-code consisting of a source node and a sequence of zero-rate or repeating code child nodes, located in the middle layer of the decoding tree for continuous cancellation decoding of polar-adjusted convolutional codes, wherein, the source node is located at the rightmost node of the special node and has arbitrary code rate and structure, and the source node has a length of satisfy ,in, The length of the convolution polynomial in the convolutional code is adjusted for polarization; the special nodes are identified by a sequence set; the sequence set includes: For the a set of sequences ,set up For functions that perform recursive computation, For the process The result after the calculation is then ,in and These represent the levels of the special node and the source node in the decoding tree, respectively. The calculation is as follows: ， in, Represents the Kronecker product. For the first The first child node Each bit value is located in the decoding tree. child nodes of level The calculation can be performed as follows: in, , It is a length of A vector of all zeros, It is a length of vector and ; Calculate the log-likelihood ratio of the source nodes corresponding to the special nodes to determine the number of paths; Based on the log-likelihood ratio and the sequence set corresponding to the special node, calculate the metric for each path, retain the path with the smallest metric, and retain the number of paths as set in the list. Using the preserved path as input, the source node is decoded to obtain the estimated codeword for each path of the source node; The estimated output bits and estimated information bits of the special node are calculated using the estimated codeword.

2. The method according to claim 1, characterized in that, The calculation of the log-likelihood ratio of the source node corresponding to the special node includes: Log-likelihood ratio based on special nodes and the Given a set of sequences, the log-likelihood ratio of the source node. The calculation is as follows: 。 3. The method according to claim 1, characterized in that, The calculation of the metric for each path includes: ， ， in, and They are respectively and A firm verdict.

4. The method according to claim 1, characterized in that, The decoding of the source node includes: If the source node has a special structure, the estimated codeword sequence of the source node can be calculated by the fast list decoding algorithm corresponding to its node; If the source node has only a general structure, the estimated codeword sequence of the source node can be calculated by the traditional serial cancellation decoding list algorithm.

5. The method according to claim 1, characterized in that, The calculation of the estimated output bits and estimated information bits of the special node using the estimated codeword includes: Calculate the output bits: ， Calculate the estimated information bits: ， in, The estimated bits for a special node are calculated as follows: ， in, For polarization generating matrix and , polarized nucleus and SR node estimated information bits The calculation is as follows: 。 6. A fast list decoding device for special nodes of polarization-adjusted convolutional codes, characterized in that, Includes the following modules: A filtering module is used to determine special nodes in the decoding tree of polar-adjusted convolutional code list decoding; wherein, the special node includes a polar-adjusted convolutional code sub-code consisting of a source node and a sequence of zero-rate or repeating code sub-nodes, located in the middle layer of the decoding tree of continuous cancellation decoding of polar-adjusted convolutional codes, wherein, the source node is located at the rightmost node of the special node and has arbitrary code rate and structure, and the length of the source node is... satisfy ,in, The length of the convolution polynomial in the convolutional code is adjusted for polarization; the special nodes are identified by a sequence set; the sequence set includes: For the a set of sequences ,set up For functions that perform recursive computation, For the process The result after the calculation is then ,in and These represent the levels of the special node and the source node in the decoding tree, respectively. The calculation is as follows: ， in, Represents the Kronecker product. For the first The first child node Each bit value is located in the decoding tree. child nodes of level The calculation can be performed as follows: in, , It is a length of A vector of all zeros, It is a length of vector and ; The first calculation module is used to calculate the log-likelihood ratio of the source node corresponding to the special node and determine the number of paths; The second calculation module is used to calculate the metric of each path based on the log-likelihood ratio and the sequence set corresponding to the special node, and retain the path with the smallest metric. The number of retained paths is the set list number. The decoding module is used to decode the source node with the reserved path as input to obtain the estimated codeword for each path of the source node; The output module is used to calculate the estimated output bits and estimated information bits of the special node using the estimated codeword.

7. A computer device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the fast list decoding method for special nodes of polarization-adjusted convolutional codes as described in any one of claims 1-5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the fast list decoding method for special nodes of polarization-adjusted convolutional codes as described in any one of claims 1-5.

Images