Intermediate Channel Selection and Decoding Method for an LDPC-Polar Cascade System

Abstract

The present invention relates to channel coding technology, and specifically relates to an intermediate channel selection and decoding method for an LDPC-Polar concatenated system; this method uses Rate-1 nodes to extract the selection set of the intermediate channel, and sorts the selection set according to the leaf set size and the Gaussian construction error probability; the sorted selection set is allocated to LDPC variable nodes; during the decoding process, based on the joint factor graph, according to the BP decoding information update rule, the process of right-transmitting the prior information of nodes such as Rate-1 and Rate-0 is simplified; the present invention improves the error performance of the LDPC-Polar concatenated system and reduces the computational complexity of the entire system according to special nodes.

Description

Technical Field

[0001] The present invention relates to channel coding technology, in particular to polar codes, and specifically relates to an intermediate channel selection and decoding method for an LDPC-Polar concatenated system. Background Art

[0002] Polar codes were proposed by Arikan and are the first channel coding scheme that theoretically achieves the Shannon limit for binary input discrete memoryless symmetric channels (BI-DMSC). Two well-known decoding algorithms are the successive cancellation (SC) and belief propagation (BP) decoding algorithms. However, the SC decoding algorithm is a serial decoding algorithm with a high decoding delay, and the error performance of the SC decoding algorithm at finite block lengths is not ideal. In addition to the SC decoding algorithm, the BP decoding algorithm is an iterative parallel decoding algorithm. The parallel nature makes it suitable for low-delay and high-throughput systems, but it also has the characteristic of relatively high complexity.

[0003] The error correction performance of polar codes can be improved by concatenating with other codewords, such as the concatenation of Polar codes and (Reed-solomon) RS codes, and the concatenation of LDPC codes and Polar codes. In the present invention, the concatenated system uses an LDPC code as the outer code and a Polar code as the inner code. The outer LDPC code improves the decoding performance by protecting the intermediate channels of the inner Polar code, and the rightward message calculation is simplified through a special node matrix during the decoding process.

[0004] However, the intermediate channel selection method proposed in the article (Abbas S M, Fan Y Z, Chen J, et al. Concatenated LDPC-Polar Codes Decoding Through Belief Propagation[C]. ISCAS, 2017.) uses the Bhattacharyya parameter as part of the selection method. However, the Bhattacharyya parameter can only be accurately calculated under the Binary Erasure Channel (BEC). Therefore, it is not appropriate to use the Bhattacharyya parameter to select the intermediate channel under non-BEC channels, which will lead to poor decoding performance of the concatenated system. The bit mapping method (Yu Q P, Shi Z P, Deng L, et al. An improved belief propagation decoding of concatenated polar codes with bit mapping[J]. IEEE Communications Letters, 2018, 22(6): 1160-1163.) does not consider the ranking of the reliability of the intermediate channels in detail when corresponding the intermediate channels and the variable nodes of the LDPC code, but only divides them into reliable and unreliable, resulting in poor bit error performance of the concatenated system. Summary of the Invention

[0005] To solve the above problems, the present invention provides an intermediate channel selection and decoding method for an LDPC-Polar concatenated system, including the following steps:

[0006] Construct a polar code with a code length of N through the Gaussian approximation construction method, which includes K information bits and N-K frozen bits; obtain all Rate-1 nodes of the polar code, and respectively form sets CS and CS2 with the first information bit and the first two information bits of all Rate-1 nodes;

[0007] Based on sets CS and CS2, use the improved intermediate channel selection method to obtain a new set CS2'; extract several Polar intermediate channels according to the new set CS2' and arrange them in descending order of reliability to obtain the Polar intermediate channel selection sequence;

[0008] Construct a regular LDPC code with a code length of N through the Mackey construction method P which includes K P information bits and N P LDPC variable nodes; based on the Polar intermediate channel selection sequence, assign a Polar intermediate channel to each LDPC variable node through the intermediate channel allocation method; where K P <K, NP <N;

[0009] The transmitting end divides the K information bits to be transmitted into two parts, selects K P information bits for LDPC coding, and transmits them through the corresponding Polar intermediate channel; the remaining K - K P information bits are directly transmitted through the high-reliability channel of the polar code;

[0010] The receiving end uses a low-complexity decoding method for joint decoding of the polar code and the LDPC code.

[0011] Furthermore, an improved intermediate channel selection method is adopted to obtain a new set CS2, including:

[0012] Use the first information bit of all Rate-1 nodes to form a set CS = {c 1 , c 2 ,..., c n}, where c i represents the i-th element in the set CS, and n represents the total number of elements in the set CS; use the first two information bits of all Rate-1 nodes to form a set where represents the i-th element in the set CS2, and m represents the total number of elements in the set, and m > n;

[0013] The first round of screening: Determine whether there is an element in the set CS If it exists, retain the element If it does not exist, the element goes to the second round of screening;

[0014] The second round of screening: Set the leaf set threshold. If the leaf set of the element is not less than the leaf set threshold, retain the element Otherwise, delete the element

[0015] The third round of screening: Obtain the minimum polarization weight of the set CS, and compare it with the polarization weight of each element in the updated set CS2 after the second round of screening; if the polarization weight of the element is not less than the minimum polarization weight of the set CS, delete the element;

[0016] Arrange each element in the updated set CS2 after the third round of screening in ascending order according to the size of its leaf set to obtain a new set CS2'; extract the first N P elements in the new set CS2', and arrange them in descending order according to the Gaussian construction error probability of the elements, and finally obtain the Polar intermediate channel selection sequence.

[0017] Further, for the K information bits to be transmitted, first select K P information bits from them for LDPC encoding to generate an LDPC code with a code length of N P Select N P matched Polar intermediate channels from the polarization channels of the polar code to transmit the LDPC code, and the remaining K - K P information bits are directly transmitted through the remaining K - K P polarization channels of the polar code; then the K information bits and N - K frozen bits input into the polarization channels are encoded by the polar code to form N transmission codewords, and the N transmission codewords are modulated by BPSK and then transmitted through the AWGN channel.

[0018] Further, the receiving end uses a low-complexity decoding method for joint decoding of the polar code and the LDPC code, including:

[0019] The receiving end obtains the modulated polar code information and performs the first-round decoding, including:

[0020] S1. According to the calculation rules of the BP decoding operation unit, perform message passing from the rightmost side to the left side of the LDPC-Polar joint factor graph;

[0021] S2. When the message passing reaches the leftmost side of the LDPC-Polar joint factor graph, find the LPDC code corresponding to and connected to the Polar intermediate channel. The variable nodes of this LPDC code receive the soft information input from the polar code and perform one round of BP decoding;

[0022] S3. After the LPDC code completes one round of BP decoding, transmit the updated soft information of the LDPC code to the leftmost side of the LDPC-Polar joint factor graph, and then perform message passing from the leftmost side to the right side of the LDPC-Polar joint factor graph in combination with the special node update rules;

[0023] After the first-round decoding is completed, continue to perform multiple rounds of decoding in a process similar to the first-round decoding until the maximum number of iterations is reached, and then perform a hard decision on the soft information to output the decoding result.

[0024] Further, construct a special matrix through special nodes, and construct special node update rules through the special matrix; the special matrix is a description of the situation where the special node schedules the PE nodes corresponding to its child nodes, and the special node refers to the Rate-1 node, Rate-0 node, and Rep node obtained when constructing the polar code.

[0025] Further, the special node update rules include the update rules for the Rate-1 node, the update rules for the Rate-0 node, and the update rules for the Rep node;

[0026] When the special matrix indicates that the current PE unit is calculating a Rate-1 node, the update rule for the Rate-1 node is adopted as follows:

[0027] R c = g(R a , L d + R b ) = s × sign(∞) × sign(L d + ∞) × min(∞, ∞) = ∞

[0028] R d = g(R a , L c ) + R b = ∞

[0029] When the special matrix indicates that the current PE unit is calculating a Rate-0 node, the update rule for the Rate-0 node is adopted as follows:

[0030] R c = g(R a , L d + R b ) = s × sign(0) sign(L d + 0) min(0, 0) = 0

[0031] R d = g(R a , L c ) + R b = s × sign(0) × sign(L c ) × min(|L c |, 0) + 0 = 0

[0032] When the special matrix indicates that the current PE unit is calculating a Rep node, the update rule for the Rep node is adopted as follows:

[0033] R c = g(R a , L d + R b ) = s × sign(∞) sign(L d + ∞) min(∞, ∞) = ∞

[0034] R d = g(R a , L c ) + R b = ∞

[0035] Among them, R c represents the right-pass information on the c node in the current PE unit, and R a represents the right-pass information on the d node in the current PE unit, and Ld Indicates the left-pass information on the d node in the current PE unit, R b Indicates the right-pass information on the b node in the current PE unit, L c Indicates the left-pass information on the c node in the current PE unit, sign represents the sign function, s represents the value 0.9375, g(x,y) = ln[1 + xy / (x + y)], R d Indicates the right-pass information on the d node in the current PE unit.

[0036] Advantages of the present invention:

[0037] Based on the intermediate channel selection and decoding method of the LDPC-Polar concatenated system provided by the present invention, a simulation analysis is carried out with the traditional concatenated system method. The simulation parameter settings are as follows: The polar code is used as the inner code, and the length of the polar code N = 1024, K = 560. The LDPC code is used as the outer code, and a (3,6) regular LDPC code with a constructed code length of 96 and an information bit of 48 is used. The simulation results show that when the length of the polar code is 1024, at BLER = 10 -3 The proposed intermediate channel selection plus allocation method has a gain of about 0.12 dB compared with the bit mapping method. And this concatenated system has a lower decoding delay compared with the CA-SCL decoding algorithm. Moreover, for the polar code with N = 1024, 10240 computing units need to be calculated in one round of iteration, but through the simplified algorithm proposed by the present invention, the calculation of 1381 PE computing units is reduced, that is, the computational complexity of the polar code is reduced by 13.5%. Brief Description of the Drawings

[0038] Figure 1 It is a flowchart of the intermediate channel selection and decoding method of the LDPC-Polar concatenated system of the present invention;

[0039] Figure 2 It is a low-complexity decoding joint factor graph of the LDPC-Polar concatenated code of an embodiment of the present invention;

[0040] Figure 3 It is a diagram of the allocation method of the intermediate channel and the LDPC variable node of an embodiment of the present invention;

[0041] Figure 4 It is a diagram of the special node type of the present invention;

[0042] Figure 5 It is a diagram of the related simplified calculation BP decoding unit of the present invention;

[0043] Figure 6 It is a diagram of the simulation results of an embodiment of the present invention. Detailed Embodiment

[0044] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0045] The present invention provides an intermediate channel selection and decoding method for an LDPC-Polar concatenated system, as Figure 1 shown, which includes the following steps:

[0046] Step 1: Construct a regular LDPC code with a code length N P = 96 and an information bit K P = 48 by the Mackey construction method.

[0047] Step 2: Construct a polar code with a code length N = 1024 by the Gaussian approximation construction method, which includes K = 560 information bits and N - K = 464 frozen bits; then perform channel polarization on the channels through which the polar code is transmitted. When the number of merged channels approaches infinity, a part of the channels tend to be noiseless channels, and another part of the channels tend to be fully noisy channels. The transmission rate of the noiseless channels will reach the channel capacity I(W), while the transmission rate of the fully noisy channels tends to 0. Utilizing this characteristic, sort the polar channels from small to large according to the reliability to obtain the set v = {v0,..., v 1023}, where v i represents the i-th polar channel in the set v. Divide the set v into two parts, that is, select 560 polar channels to transmit the information bits of the information bits, and the remaining 464 polar channels to transmit the frozen bits of the frozen bits.

[0048] Step 3: Obtain all the Rate-1 nodes of the polar code, and respectively form sets CS and CS2 with the first information bit and the first two information bits of all the Rate-1 nodes.

[0049] Specifically, a Rate-1 node is a node whose leaf nodes are all information bits; at the same time, obtain the Rate-0 nodes and Rep nodes of the polar code. A Rate-0 node is a node whose leaf nodes are all frozen bits; a Rep node is a node whose rightmost leaf node is an information bit and the remaining leaf nodes are all frozen bits. As Figure 4 shown, the black circles represent information bits, the white circles represent frozen bits, and the gray circles represent those containing both information bits and frozen bits.

[0050] Step 4: Based on set CS and set CS2, use an improved intermediate channel selection method to obtain a new set CS2'; extract several Polar intermediate channels from the new set CS2' and arrange them in descending order of reliability to obtain a Polar intermediate channel selection sequence.

[0051] Specifically, obtaining the new set CS2 by using the improved intermediate channel selection method includes:

[0052] Use the first information bits of all Rate-1 nodes to form set CS = {c 1 , c 2 ,..., c n}, where c i represents the i-th element in set CS, and n represents the total number of elements in set CS; use the first two information bits of all Rate-1 nodes to form set where represents the i-th element in set CS2, and m represents the total number of elements in the set, and m > n. In this embodiment, n = 122, and the channel indices corresponding to the Polar intermediate channels of set CS are {128, 190,..., 961}; m = 194, and the channel indices corresponding to the Polar intermediate channels of set CS2 are {128, 190,..., 962}.

[0053] First-round screening: For the elements in set CS2 judge whether the element belongs to set CS. If it belongs, retain the element If it does not belong, then the element goes through the second-round screening.

[0054] Second-round screening: Set the leaf set threshold. If the leaf set of the element is not less than the leaf set threshold, retain the element Otherwise, delete the element In this embodiment, the leaf set threshold is the minimum leaf set size in set CS, that is, 8.

[0055] Third-round screening: Obtain the minimum polarization weight of set CS, and compare it with the polarization weight of each element in the updated set CS2 after the second-round screening; if the polarization weight of the element is not less than the minimum polarization weight of set CS, delete the element. In this embodiment, the updated set CS2 after the third-round screening includes 112 elements in total, and their corresponding channel indices are {128, 190,..., 961}.

[0056] Arrange each element in the updated CS2 set after the third-round screening in ascending order of its leaf set size to obtain a new set CS2'; extract the first N in the new set CS2' PElements, and arrange them in descending order according to the Gaussian construction error probability of the elements, and finally obtain the Polar intermediate channel selection sequence.

[0057] Step 5: Based on the Polar intermediate channel selection sequence, assign a Polar intermediate channel to each LDPC variable node through the intermediate channel allocation method.

[0058] Specifically, as Figure 3 shown, the allocation process of the Polar intermediate channel includes:

[0059] Label the LPDC variable nodes in sequence according to 1 to N LDPC in sequence, and label the elements in the Polar intermediate channel selection sequence from left to right according to 1 to N LDPC in sequence. In this embodiment, N LDPC = 96.

[0060] Assign the Polar intermediate channel corresponding to the element labeled 1 to the LDPC variable node labeled 1, and the Polar intermediate channel corresponding to the element labeled 1 is the most unreliable. In this embodiment, the channel index of the Polar intermediate channel corresponding to the element labeled 1 is 961, and the LDPC variable node labeled 1 is connected to the 9th LDPC check node and the 11th LDPC check node.

[0061] Obtain all LDPC variable nodes connected to the same LDPC check node as the LDPC variable node labeled 1, and assign the Polar intermediate channels corresponding to the elements labeled {N LDPC , N LDPC - 1,..., N LDPC - k1 + 1} to them, where k1 represents the number of LDPC variable nodes connected to the same LDPC check node as the LDPC variable node labeled 1. Specifically, in this embodiment, the labels of the LDPC variable nodes connected to the 9th LDPC check node together with the LDPC variable node labeled 1 are {20, 37, 55, 63, 94}, and assign the Polar intermediate channels corresponding to the elements labeled {96, 95, 94, 93, 92} to them, and the channel indices corresponding to these Polar intermediate channels are {365, 370, 371, 373, 377}; the labels of the LDPC variable nodes connected to the 11th LDPC check node together with the LDPC variable node labeled 1 are {2, 4, 50, 52, 83}, and assign the Polar intermediate channels corresponding to the elements labeled {91, 90, 89, 88, 87} to them, and the channel indices corresponding to these Polar intermediate channels are {406, 407, 410, 411, 413}.

[0062] After all LDPC variable nodes connected to the same LDPC check node as the LDPC variable node labeled 1 are allocated; find the next unmatched LDPC variable node with the smallest label, allocate it to the Polar intermediate channel corresponding to the element labeled 2, and perform the search and matching according to the above process until all LDPC variable nodes are matched. In this embodiment, after the relevant search and matching of the LDPC variable node labeled 1 are completed, the smallest label of the remaining unmatched LDPC variable nodes is 3. Therefore, the Polar intermediate channel corresponding to the element labeled 2 is allocated to the LDPC variable node labeled 3, and the channel index of this Polar intermediate channel is 929.

[0063] Step Six: The transmitting end divides the K information bits to be transmitted into two parts. First, select K P information bits for LDPC encoding to generate an LDPC code with a code length of N P , and select N P matched Polar intermediate channels from the polarization channels of the polar code for transmission through the above operations; the remaining K - K P information bits are directly transmitted through the remaining K - K P high-reliability channels of the polar code. In this embodiment, K = 560 information bits need to be transmitted. Select K P = 48 information bits for LDPC encoding, and the resulting code length N P = 96. Then, transmit this LDPC code through the Polar intermediate channel matched by the LDPC variable node, while the remaining K - K P = 512 information bits are directly transmitted through 512 polarization channels except for the already allocated Polar intermediate channels; then, the 560 information bits and N - K = 464 frozen bits introduced into the polarization channels are encoded by the polar code to form N = 1024 transmission codewords.

[0064] Step Seven: Modulate the 1024 transmission codewords by BPSK, convert the 0 and 1 codewords into 1 and -1, and transmit them through the AWGN channel.

[0065] Step Eight: The receiving end performs joint decoding of the polar code and the LDPC code using a low-complexity decoding method.

[0066] Specifically, as Figure 2 shown, the low-complexity decoding method includes:

[0067] The receiving end obtains the modulated polar code information and performs the first-round decoding, including:

[0068] S1. According to as Figure 5The calculation rule of the PE unit of the BP decoder shown is to perform information transfer from the rightmost side to the left of the LDPC-Polar joint factor graph;

[0069] S2. When the information transfer reaches the leftmost side of the LDPC-Polar joint factor graph, search for the LPDC code corresponding to the connection with the Polar intermediate channel. The variable nodes of this LPDC code receive the soft information transmitted by the polar code and perform one round of BP decoding;

[0070] S3. After the LPDC code completes one round of BP decoding, the soft information after iterative update of the LDPC code is transmitted to the leftmost side of the LDPC-Polar joint factor graph, and then information transfer is performed from the leftmost side to the right of the LDPC-Polar joint factor graph in combination with the special node update rule;

[0071] After the first round of decoding is completed, continue to perform multiple rounds of decoding in a process similar to the first round of decoding until the maximum number of iterations is reached, and then perform a hard decision on the soft information to output the decoding result. Specifically, the prior right-transmitted information corresponding to all child nodes of the Rate-0 node is infinite. Therefore, in subsequent calculation processes, the values transmitted by all child nodes of its Rate-0 node in the corresponding BP decoding unit are also infinite. There is a special child node containing information bits in the Rep node. After excluding this special child node, the prior right-transmitted information of the remaining child nodes of the Rep node is infinite. Therefore, the values calculated by this part of the child nodes of the Rep node in subsequent calculation processes are also infinite. For the Rate-1 node, excluding the part of the Rate-1 nodes connected to the LDPC codeword, the prior right information of the remaining child nodes of the Rate-1 node is 0. Therefore, the values calculated in subsequent transfer processes are 0. The special node is used to determine the distribution of its child nodes and schedule to simplify the decoding process. Based on the above analysis, designing the special node update rule to directly set the value to 0 or infinite can achieve the purpose of simplifying the calculation process and continue to complete the second half of the iterative process.

[0072] Specifically, a special matrix is constructed through special nodes. The special matrix is a description of the situation where the special node schedules the PE node. BP decoding completes one round of scheduling process according to Nlog2(N) / 2 PE nodes, and uses the special matrix of [N / 2, log2(N)] to correspond to the scheduling process of BP decoding. If the length of the child nodes of the special node is Nspecial, the size of its corresponding special matrix is [N special / 2, log2(N special )], that is, the dimensions of the row and column are N special / 2 and log2(N special ), respectively. When the special matrix shows that the current PE unit is of the special node type, that is, an R0 node, or a Rep node, or a Rep node, the operation is simplified according to the following update rules.

[0073] Specifically, the Rate-1 node update rule is as follows:

[0074] R c = g(R a , L d + R b ) = s × sign(∞) × sign(L d + ∞) × min(∞, ∞) = ∞

[0075] R d = g(R a , L c ) + R b = ∞

[0076] The Rate-0 node update rule is as follows:

[0077] R c = g(R a , L d + R b ) = s × sign(0) sign(L d + 0) min(0, 0) = 0

[0078] R d = g(R a , L c ) + R b = s × sign(0) × sign(L c ) × min(|L c |, 0) + 0 = 0

[0079] The Rep node update rule is as follows:

[0080] R c = g(R a , L d + R b ) = s × sign(∞) sign(L d + ∞) min(∞, ∞) = ∞

[0081] R d = g(R a , L c ) + R b = ∞

[0082] The LDPC-Polar cascade system of the present invention uses a BP decoder for decoding. The BP decoder is built by multiple PE units. As Figure 5 shown is the calculation process of a PE unit. R c represents the rightward message on the c node in the PE unit. R aRepresents the right - passed information on the d node in the PE unit, L d Represents the left - passed information on the d node in the PE unit, R b Represents the right - passed information on the b node in the PE unit, L c Represents the left - passed information on the c node in the PE unit, sign represents the sign function, s represents the value 0.9375, g(x,y)=ln[1 + xy / (x + y)], R d Represents the right - passed information on the d node in the PE unit.

[0083] Based on the intermediate channel selection and decoding method of the LDPC - Polar concatenated system provided by the present invention, a simulation analysis is carried out with the traditional concatenated system method. The simulation parameter settings are as follows: The polar code is used as the inner code, and the length of the polar code N = 1024, K = 560. The LDPC code is used as the outer code, and a (3,6) regular LDPC code with a constructed code length of 96 and 48 information bits is used. The simulation results Figure 6 show that when the length of the polar code is 1024, at BLER = 10 -3 the proposed intermediate channel selection plus allocation method has a gain of about 0.12 dB compared with the bit - mapping method. And this concatenated system has a lower decoding delay compared with the CA - SCL decoding algorithm.

[0084] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intermediate channel selection and decoding method for an LDPC-Polar concatenated system, characterized in that, Including the following steps: Construct a polar code with a code length of N through the Gaussian approximation construction method, which includes K information bits and N-K frozen bits; obtain all Rate-1 nodes of the polar code, and respectively form sets CS and CS2 with the first information bit and the first two information bits of all Rate-1 nodes; Based on sets CS and CS2, obtain a new set CS2` by using the improved intermediate channel selection method; extract several Polar intermediate channels according to the new set CS2` and arrange them in descending order of reliability to obtain the Polar intermediate channel selection sequence; Obtaining a new set CS2 by using the improved intermediate channel selection method includes: Using the first information bits of all Rate-1 nodes to form a set CS = {c 1 , c 2 ,..., c n}, where c i represents the i-th element in the set CS, and n represents the total number of elements in the set CS; using the first two information bits of all Rate-1 nodes to form a set where represents the i-th element in the set CS2, and m represents the total number of elements in the set, and m > n; First-round screening: Determine whether there is an element in the set CS If there is, retain the element If not, the element Undergo the second-round screening; Second-round screening: Set the leaf set threshold. If the leaf set of an element is not less than the leaf set threshold, then retain the element Otherwise, delete the element The third round of screening: obtain the minimum polarization weight of set CS, and compare it with the polarization weight of each element in the updated set CS2 after the second round of screening; if the polarization weight of the element is not less than the minimum polarization weight of set CS, delete the element; Arrange each element in the updated set CS2 after the third round of screening in ascending order according to the size of its leaf set to obtain a new set CS2`; extract the first N P elements in the new set CS2` and arrange them in descending order according to the Gaussian construction error probability of the elements to finally obtain the Polar intermediate channel selection sequence; Construct a regular LDPC code with code length N by Mackey construction method P , which includes K P information bits and N P LDPC variable nodes; Based on the Polar intermediate channel selection sequence, assign a Polar intermediate channel to each LDPC variable node through the intermediate channel allocation method; where K P < K, N P < N; The transmitter divides the K information bits to be transmitted into two parts, and selects K P information bits for LDPC coding and transmits them through the corresponding Polar intermediate channel; the remaining K - K P information bits are directly transmitted through the high-reliability channel of the polar code; The receiver uses a low-complexity decoding method to perform joint decoding of the polar code and the LDPC code, including: The receiver obtains the modulated polar code information and performs the first round of decoding, including: S1. According to the calculation rules of the BP decoding operation unit, perform information transfer from the rightmost side to the left side of the LDPC-Polar joint factor graph; S2. When the information is transferred to the leftmost side of the LDPC-Polar joint factor graph, find the LPDC code corresponding to and connected to the Polar intermediate channel. The variable nodes of this LPDC code receive the soft information transmitted by the polar code and perform one round of BP decoding; S3. After the LPDC code completes one round of BP decoding, transmit the soft information generated by the LDPC code to the leftmost side of the LDPC-Polar joint factor graph, and then perform information transfer from the leftmost side to the right side of the LDPC-Polar joint factor graph in combination with the special node update rules; After the first round of decoding is completed, continue to perform multiple rounds of decoding in a process similar to the first round of decoding until the maximum number of iterations is reached, and then perform a hard decision on the soft information to output the decoding result; Construct a special matrix through special nodes, and construct special node update rules through the special matrix; the special matrix is a description of the situation where the special node corresponds to the sub-node scheduling PE node, and the special nodes refer to the Rate-1 node, Rate-0 node, and Rep node obtained when constructing the polar code; The special node update rules include the update rules for Rate-1 nodes, the update rules for Rate-0 nodes, and the update rules for Rep nodes; When the special matrix shows that the current PE unit is calculating a Rate-1 node, the update rule for the Rate-1 node is adopted, as follows: R c = g(R a , L d + R b ) = s × sign(∞) × sign(L d + ∞) × min(∞, ∞) = ∞ R d = g(R a , L c ) + R b = ∞ When the special matrix shows that the current PE unit is calculating a Rate-0 node, the update rule for the Rate-0 node is adopted, as follows: R c = g(R a , L d + R b ) = s × sign(0) sign(L d + 0) min(0, 0) = 0 R d = g(R a , L c ) + R b = s × sign(0) × sign(L c ) × min(|L c |, 0) + 0 = 0 When the special matrix shows that the current PE unit is calculating a Rep node, the update rule for the Rep node is adopted, as follows: R c = g(R a , L d + R b ) = s × sign(∞)sign(L d + ∞)min(∞, ∞) = ∞ R d = g(R a , L c ) + R b = ∞ Among them, R c represents the right transfer information on the c node in the current PE unit, R a represents the right transfer information on the d node in the current PE unit, L d represents the left transfer information on the d node in the current PE unit, R b represents the right transfer information on the b node in the current PE unit, L c represents the left transfer information on the c node in the current PE unit, sign represents the sign function, s represents the value 0.9375, g(x, y) = ln[1 + xy / (x + y)], R d represents the right transfer information on the d node in the current PE unit.

2. The intermediate channel selection and decoding method of an LDPC-Polar concatenated system according to claim 1, characterized in that For the K information bits to be transmitted, first select K P of them for LDPC encoding to generate an LDPC code with a code length of N P . Select N P matched Polar intermediate channels from the polarization channels of the polar code to transmit this LDPC code, while the remaining K - K P information bits are directly transmitted through the remaining K - K P polarization channels of the polar code; then the K information bits and N - K frozen bits entering the polarization channels are encoded by the polar code to form N transmission codewords, and the N transmission codewords are transmitted through the AWGN channel after BPSK modulation.

Images