Soft input soft output decoding method and device suitable for polar code soscl decoder

Abstract

The application provides a soft input soft output decoding method and device suitable for a polar code SoSCL decoder, and the specific steps of the decoding method include: using the SoSCL decoder on a channel received LLR sequence to output an extrinsic information sequence; introducing a fading factor to weight the channel received LLR sequence and the extrinsic information sequence to calculate a posteriori information sequence; using the path metric value of the SoSCL decoder and a frozen bit set to calculate a correction factor; and using the posteriori information sequence, the correction factor and the channel received LLR sequence to calculate a bit-level soft output information sequence. The application provides a weighted bit-level soft output information sequence by introducing the fading factor and the correction factor, corrects the soft output of the SoSCL decoder, obtains a more accurate polar code soft output decoding method, and effectively improves the algorithm performance of the polar code soft input soft output decoding module.

Description

Technical Field

[0001] This invention belongs to the field of channel coding and decoding technology, specifically relating to a soft-input soft-output decoding method and apparatus suitable for SoSCL polar code decoders. Background Technology

[0002] Polar codes, as a coding method that can achieve channel capacity in binary symmetric channels, have been widely used in the field of wireless communication due to their good block error rate performance under short to medium code lengths. Polar codes are usually decoded using the Successive Cancellation List (SCL) decoding algorithm. In concatenated polar code systems, iterative processing is a key means to improve decoding performance. In iterative processing, a soft decoding algorithm based on SCL is used. For example, the literature (Yuan P, Duffy KR, Médard M. Soft-output successive cancellationlist decoding[J]. IEEE Transactions on Information Theory, 2025,71(2):1007–1017.) proposes the SO-SCL algorithm, which provides a relatively accurate soft output sequence based on the SCL decoding tree, but can only provide one soft output sequence. The literature (Gao L, Wu X, Yang Y, et al. Soft-Out Successive Cancellation Decoding and Its List Version for Polar Codes[J]. IEEE Wireless Communications Letters, 2024,14(3):651–655.) proposes the SoSCL algorithm, which can provide multiple extrinsic information sequences and provide the final soft output sequence through the weighted sum of multiple extrinsic information sequences. Although SoSCL provides excellent soft output information and meets the needs of cascaded systems to a certain extent, there is still room for improvement in the accuracy of its soft information. To this end, this patent proposes an improved mechanism for the SoSCL decoder, which focuses on introducing fading factors and correction factors based on the characteristics of multiple extrinsic information sequences provided by SoSCL to effectively correct the soft output information, and ultimately effectively improves the soft input and soft output decoding capability of SoSCL.

[0003] A search revealed that Chinese invention patent CN119402017B discloses a polar code soft-input soft-output decoding method, decoding device, and communication equipment. The decoding method includes: (S1) performing fast serial cancellation list decoding under dynamic frozen bit constraints on all special nodes constituting a polar code, wherein the types of special nodes include: rate zero, rate one, repetition, and single parity check; (S2) after decoding each special node, calculating an approximation of the sum of the posterior probabilities of the unvisited legal codewords under that node; (S3) using the approximation obtained in step (S2) to estimate the codebook probability and calculate the bit-level soft information of the output polar code.

[0004] The aforementioned prior art differs from this application in the following ways:

[0005] The core objective of the prior art is low-latency decoding, and its application is a serial cancellation list decoder. By designing fast serial cancellation list decoders for special nodes such as Rate0, Rate1, REP, and SPC, it increases parallelism without sacrificing soft output accuracy, thereby reducing decoding latency. In contrast, the core objective of this application is to correct the soft output of a given soft-in, soft-out decoder, and its application is a SoSCL decoder. By introducing fading factors and correction factors to construct a weighted bit-level soft output information sequence, it specifically improves the accuracy of the original soft output. The core objectives and application targets of the two are clearly different.

[0006] The decoding logic of the comparison file involves step-by-step processing of special nodes and codebook probability calculation. First, decoding is performed on all special nodes under dynamic frozen bit constraints. Then, the approximate codebook probability sum of unvisited legal codewords is calculated. Finally, soft information is output based on this approximate estimated codebook probability. In contrast, the logic of this invention involves weighted extrinsic information, combination of multiple path posterior information sequences, and correction factor improvement. First, multiple extrinsic information sequences are output through the SoSCL decoder. Multiple posterior information sequences are calculated by weighting with fading factors. The correction factor is obtained by combining the path metric and the frozen bit set. Finally, the calculation and correction of soft output information are completed by combining multiple posterior information sequences and the correction factor. The decoding logics of the two are completely different.

[0007] The technical focus of the prior art is node parallel optimization, which improves parallel processing capabilities by optimizing the decoding process of special nodes, thereby reducing decoding latency; while the technical focus of this application is the soft output correction mechanism, which innovatively introduces two key parameters, fading factor and correction factor, and improves the soft output accuracy of a given soft-input soft-output decoder through weighted calculation and correction process. The two have significantly different focuses of technical innovation.

[0008] A search revealed that Chinese invention patent CN114584152B discloses a sliding window soft serial cancellation list decoding method suitable for polar code modulation, including the following steps: receiver initialization; and decoding a source information bit sequence θ of length K at time t. t Perform sliding window decoding; if the decoding state s0 = 0, execute the second iteration process; output θ. t If s0 = 1, then code reconstruction yields w. t The estimated update makes it consistent with; let t′=t+L w Calculate and update s; let t = t + 1, return to step two until the final estimated sequence of all sent source information bit sequences is output.

[0009] The aforementioned prior art differs from this patent in the following ways:

[0010] The application scenario of the comparison document focuses on asynchronous polar code coded modulation (A-PCM) systems. Decoding requires the cooperation of spatial coupling structures, and the core is adapted to the processing of time-series information under high-order modulation. It belongs to the system-level application of modulation and decoding co-optimization. In contrast, the application scenario of this invention is clearly SoSCL decoders. It does not rely on polar code modulation, spatial coupling structures or iterative decoding frameworks. It only corrects the soft output of SoSCL decoders. It belongs to the performance optimization of a single decoding module. Its adaptation range is more focused on specific decoder types. The application scenarios of the two are completely different.

[0011] The decoding process of the comparison file relies on a sliding window mechanism and state iteration updates. It decodes the time-series source information bit sequence through a sliding window, determines whether to perform a second iteration based on the decoding state, and also needs to ensure sequence consistency through code reconstruction and state updates. In contrast, the decoding process of this invention uses external information weighting, combination of multi-path posterior information sequences, and correction factor improvement, without the need for sliding window segmentation, second iteration, or state update mechanisms. The decoding process architectures of the two are significantly different.

[0012] The soft information generation of the comparison file relies on path metric differences and iterative updates of extrinsic information. The posterior information is derived from the minimum path metric difference of multiple estimated paths, and the extrinsic information is obtained from the difference between the posterior and prior information. Furthermore, it requires the combination of system parameters such as interleaver and channel mapping mode to complete the information conversion. The core parameters are path metric, prior information, and posterior information. In contrast, the soft information generation of this invention relies on two core parameters: fading factor and correction factor. The fading factor is used to balance the weights of the LLR sequence received by the channel and the extrinsic information sequence to obtain the posterior information sequence. The soft output sequence is calculated using the multi-path posterior information sequence and the correction factor. It does not rely on path iterative updates or system-level mapping parameters. The core focus is on the adaptation of factor weights and the construction of correction factors. The soft information generation logic of the two is significantly different. Summary of the Invention

[0013] To address the shortcomings mentioned in the background section, this invention provides an enhanced soft-input soft-output decoding method and apparatus suitable for SoSCL decoders of polar codes, which can enhance the soft-input soft-output decoding performance of SoSCL decoders.

[0014] The specific plan is as follows:

[0015] A soft-input soft-output decoding method applicable to SoSCL polar code decoders, comprising the following steps:

[0016] S1. Use a SoSCL decoder to receive the LLR sequence from the channel and output the external information sequence.

[0017] S2. Introduce a fading factor to perform weighted calculations on the channel received LLR sequence and the extrinsic information sequence obtained in step S1, and output the posterior information sequence.

[0018] S3. Calculate the correction factor using the path metric and frozen bit set of the SoSCL decoder decoding module;

[0019] S4. Calculate the bit-level soft output information sequence using the posterior information sequence obtained in step S2, the correction factor obtained in step S3, and the channel received LLR sequence.

[0020] Further, step S1 is as follows:

[0021] Input length is The channel receives LLR sequences, with a maximum number of paths. When the decoding module of the SoSCL decoder performs path expansion to increase the existing number of paths... At that time, retain The path metrics of each node along the path are calculated and sorted in ascending order. The nodes with the smallest path metrics are retained. The node information of the path, based on the previous The decoding result of each path is output as follows External information sequence.

[0022] Furthermore, S11, the decoding module is initialized, including:

[0023] definition and All A three-dimensional matrix consisting entirely of zeros; The maximum number of paths; For the factor graph of the decoding module Line number Liede The information passed to the left by the nodes of the path; For the factor graph of the decoding module Line number Liede The information passed to the right by the nodes on the path, where , ;Will Matrix number The column is initialized as a channel receive LLR sequence. , The node with the index of the frozen bit in the first column of the matrix is ​​initialized to infinity; the first... The input and output nodes of the path are respectively and ,Will Initialize to the first row of the decoding module factor graph. Column nodes, Initialize it as the node in the first row and first column of the decoding module factor graph;

[0024] S12, For each bit to be encoded Perform serial cancellation list decoding when accessing an information node. Extend the path at time, where, An ascending set composed of information bit indices, i.e. ; For information bit index set The Middle Each element, from the information bit node Time Retention The path metrics of each path are sorted in ascending order, where... ; Retain the path with the smallest metric value Node information for each path;

[0025] S13, After all bits to be encoded have been translated At that time, the serial cancellation list decoding module is used to update the input and output nodes of each path, that is... and ,calculate The information of each node in the matrix, from the input node to the output node, is calculated as follows:

[0026] ;

[0027] ;

[0028] in, , and represent The function takes two arbitrary real numbers as input, and in the above formula... represent or , represent The sign of the value, represent The sign of the value, Represents input and The minimum absolute value;

[0029] S14. Repeat steps S12 and S13 until all nodes have been visited. The rightmost nodes of the decoder module factor graph... This constitutes the external information output sequence, i.e. ;in , .

[0030] Furthermore, in step S2, the weighted calculation is performed according to the following formula:

[0031] ;

[0032] in, For posterior information sequence, , Represents the fading factor.

[0033] Further, in step S3, the correction factor is calculated according to the following formula:

[0034] ;

[0035] in, As a correction factor, For path probability, For decoding module path metric, representing the first... The first bit The path metric for each path. The ascending set of frozen bit indices, i.e. , For set , That is, set The Middle The value of each element;

[0036] And path probability Path measurement with decoding module The relationship is: .

[0037] Further, in step S4, the bit-level soft output information sequence is calculated according to the following formula:

[0038] ;

[0039] in, For bit-level soft output information sequences, subscript , and The relationship is: , and The relationship is: , probability From the channel LLR value Calculated.

[0040] The present invention also provides a terminal device, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor. The memory stores the computer program capable of running on the processor. When the processor loads and executes the computer program, it employs the soft-input soft-output decoding method described above for a polar code SoSCL decoder.

[0041] The present invention also provides a computer-readable storage medium storing a computer program, wherein the computer program, when running, controls the device or apparatus containing the computer-readable storage medium to execute the above-described soft-input soft-output decoding method applicable to a polar code SoSCL decoder.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] 1. By introducing a fading factor to weight and fuse the channel received LLR sequence and the external information sequence, and combining the path metric and the frozen bit set to calculate the correction factor, an enhanced bit-level soft output information calculation method is constructed, which more accurately corrects the soft output of the SoSCL decoder and improves the overall performance of the soft input soft output decoding module.

[0044] 2. This invention can generate more reliable soft information, which helps to improve the iterative decoding efficiency of cascaded systems and further reduce the block error rate;

[0045] 3. The fading factor can be optimized and adjusted through simulation or experiment according to the actual channel conditions, so that the decoding method has good environmental adaptability and robustness, and is suitable for a variety of wireless communication scenarios;

[0046] 4. The proposed method has a clear flow and controllable computational complexity. It can be easily integrated into the existing SoSCL decoder architecture without significant changes to the hardware structure, and has high practicality and feasibility. Attached Figure Description

[0047] Figure 1 This is a flowchart illustrating the soft-input soft-output decoding method for a polar code SoSCL decoder provided in this application;

[0048] Figure 2 A factor diagram and decoding tree diagram of a soft-input soft-output decoding method for a polar code SoSCL decoder with a length of 8 provided in the embodiments of this application;

[0049] Figure 3 The diagram shows a comparison of the soft-input soft-output decoding method for polar code SoSCL decoders provided in this application embodiment and the decoding performance of the SoSCL decoder in a parallel cascaded iterative system. Detailed Implementation

[0050] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. The specific implementation methods of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0051] Example 1:

[0052] Consider a code length of ( Polar codes (where the integers are positive), and the bit sequence to be encoded. The codeword is obtained after polarization transformation. ,in Generate a matrix for the polar code. Indicates the Kronecker product. .

[0053] This embodiment provides a soft-input soft-output decoding method for the above-mentioned polar codes. The specific implementation process is as follows: Figure 1 As shown, it includes the following steps:

[0054] S1. Use a SoSCL decoder on the received LLR sequence to output the external information sequence, specifically including:

[0055] S11. Initialize the decoding module, specifically including:

[0056] definition and All A three-dimensional matrix consisting entirely of zeros; Represents the maximum number of paths; The factor graph representing the decoding module is shown. Line number Liede The message passed to the left by the nodes on the path; The factor graph representing the decoding module is shown. Line number Liede The information passed to the right by the nodes on the path, where , ; Matrix number The column is initialized as a channel receive LLR sequence. ; The node with the index of the frozen bit in the first column of the matrix is ​​initialized to infinity; the first... The input and output nodes of the path are respectively and ,like Figure 2 As shown, Initialize to the first row of the decoding module factor graph. Column nodes, Initialize it as the node in the first row and first column of the decoding module factor graph;

[0057] S12, For each bit to be encoded Perform serial cancellation list decoding when accessing an information node. Extend the path at time, where, This represents the ascending set of information bit indices, i.e. , Indicates information bit index set The Middle Each element, from the information bit node Time Retention The path metrics of each path are sorted in ascending order, where... Retain the path with the smallest metric value. Node information for each path;

[0058] S13, After all bits to be encoded have been translated At that time, the serial cancellation list decoding module is used to update the input and output nodes of each path, that is... and ,calculate The information of each node in the matrix, from the input node to the output node, is calculated as follows:

[0059] ;

[0060] ;

[0061] in, , and represent The function takes two arbitrary real numbers as input, and in the above formula... represent or , represent The sign of the value, represent The sign of the value, Represents input and The minimum absolute value;

[0062] S14. Repeat steps S12 and S13 until all nodes have been visited. The rightmost nodes of the decoder module factor graph... This constitutes the external information output sequence, i.e. ,in , .

[0063] S2. Introduce a fading factor. The a posteriori information sequence is calculated by weighting the channel received LLR sequence and the extrinsic information sequence obtained in step S1, specifically including:

[0064] Introducing a fading factor Receive LLR sequence for channel and external information sequence Perform weighted summation to calculate the posterior information sequence The calculation method is as follows:

[0065] ;

[0066] in , This represents the fading factor, the specific value of which can be determined through simulation or experimentation. .

[0067] S3. Calculate the correction factor using the path metric and frozen bit set of the SoSCL decoder decoding module, specifically including:

[0068] ;

[0069] in, Represents the correction factor, path probability and decoding module path measurement The relationship is , Representing the The first bit The path metric for each path. This represents the ascending set of frozen bit indices, i.e. , Representative set , That is, set The Middle The value of each element.

[0070] S4. Calculate the bit-level soft output information sequence using the posterior information sequence obtained in step S2, the correction factor obtained in step S3, and the channel received LLR sequence. Specifically, this includes:

[0071] ;

[0072] in, Represents a bit-level soft output information sequence, subscript , and The relationship is , and The relationship is , probability From the channel LLR value Calculated.

[0073] The soft-input soft-output decoding method for SoSCL decoders of polar codes provided in this embodiment can be applied to parallel cascaded iterative decoding systems. Figure 3 This paper presents a performance comparison between the enhanced soft-input soft-output decoding method proposed in this application, applicable to SoSCL decoders of polar codes, and the decoding performance of the SoSCL decoder in a parallel cascaded iterative decoding system. Figure 3 It can be seen that the decoding method provided in this application effectively improves the algorithm performance of the polar code soft-input soft-output decoding module.

[0074] Example 2:

[0075] This embodiment provides a terminal device, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor. The memory stores the computer program that can run on the processor. When the processor loads and executes the computer program, it employs the decoding method described in Embodiment 1. The processor may be a central processing unit (CPU), a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices. It serves as the computing and control core, used to load and execute instructions stored in a computer-readable medium, thereby implementing the various technical solutions described in this invention.

[0076] Example 3:

[0077] This embodiment provides a computer-readable storage medium storing a computer program. When the computer program runs, it controls the device or apparatus containing the computer-readable storage medium to execute the decoding method described in Embodiment 1. The storage medium can be any combination of one or more computer-readable media, such as, but not limited to, electrical, magnetic, optical, infrared, or semiconductor systems, devices, or components, or any combination thereof. More specific examples include, but are not limited to, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, or magnetic storage devices. Without departing from the spirit of the invention, any tangible medium containing or storing a program can serve as the computer-readable storage medium of this invention.

[0078] The above embodiments are for illustrative purposes only and are not intended to limit the scope of this application. Although this application has been described in detail with reference to the embodiments, those skilled in the art should understand that various combinations, modifications, or equivalent substitutions of the technical solutions of this application do not depart from the spirit and scope of the technical solutions of this application and should be covered within the scope of the claims of this application.

Claims

1. A soft-input soft-output decoding method suitable for a polar code SoSCL decoder, characterized in that, The method comprises the following steps: S1, using a SoSCL decoder to decode the channel received LLR sequence to output an extrinsic information sequence; The step S1 is as follows: Input a channel received LLR sequence with length , the maximum path number is , when the decoding module of the SoSCL decoder carries out path extension so that the existing path number reaches , the path metric values of the nodes of the first paths are reserved and arranged in ascending order, the node information of the first paths with the smallest path metric values is reserved, and the corresponding extrinsic information sequences are output according to the decoding results of the first paths. The specific steps of the step S1 are as follows: S11, initializing the decoding module, comprising: definition and All A three-dimensional matrix consisting entirely of zeros; The maximum number of paths; For the factor graph of the decoding module Line number Liede The information passed to the left by the nodes of the path; For the factor graph of the decoding module Line number Liede The information passed to the right by the nodes on the path, where , ;Will Matrix number The column is initialized as a channel receive LLR sequence. , The node with the index of the frozen bit in the first column of the matrix is ​​initialized to infinity; the first... The input and output nodes of the path are respectively and ,Will Initialize to the first row of the decoding module factor graph. Column nodes, Initialize it as the node in the first row and first column of the decoding module factor graph; S12, For each bit to be encoded Perform serial cancellation list decoding when accessing an information node. Extend the path at time, where, An ascending set composed of information bit indices, i.e. ; For information bit index set The Middle Each element, from the information bit node Time Retention The path metrics of each path are sorted in ascending order, where... ; Retain the path with the smallest metric value Node information for each path; S13, After all bits to be encoded have been translated At that time, the serial cancellation list decoding module is used to update the input and output nodes of each path, that is... and ,calculate The information of each node in the matrix, from the input node to the output node, is calculated as follows: ； ； wherein , and represent two arbitrary real numbers input to the function, in the above equation represent or , represent the sign of the value, represent the sign of the value, represent the input and the minimum of the absolute values in S14, repeat steps S12 and S13 until all nodes are visited, the decoder module factors the rightmost nodes of the factor graph constitute the extrinsic information output sequence, i.e. ; wherein , ; S2, introducing a fading factor to weight the channel received LLR sequence and the extrinsic information sequence obtained in the step S1 to calculate a posteriori information sequence; The weighting calculation in the step S2 is performed according to the following formula: ； wherein is a sequence of a posteriori information, , denotes a fading factor; S3, using the path metric value of the SoSCL decoder decoding module and the frozen bit set to calculate a correction factor; The correction factor in the step S3 is calculated according to the following formula: ； wherein, is a correction factor, is a path probability, is a path metric of a decoding module representing the path metric value of the bit path, bit path, is an ascending set of frozen bit indices, i.e. , is a set , is the value of the element in the set ; And path probability Relationship with decoding module path metric is: ; S4, using the a posteriori information sequence obtained in the step S2, the correction factor obtained in the step S3 and the channel received LLR sequence to calculate a bit-level soft output information sequence; The bit-level soft output information sequence in the step S4 is calculated according to the following formula: ； wherein, is a sequence of bit-level soft output information, subscript , and have the following relationship: , and have the following relationship: , , the probability is calculated from the channel LLR value .

2. A terminal device comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that, The memory stores a computer program capable of running on the processor, and when the processor loads and executes the computer program, the soft input soft output decoding method for the SoSCL decoder of the polar code is adopted.

3. A computer-readable storage medium having stored therein a computer program, characterized in that, The computer program controls the device or apparatus where the computer readable storage medium is located to execute the soft input soft output decoding method for the SoSCL decoder of the polar code when running.

Images