A 5G-LDPC code closed set blind identification method based on BP iteration

Abstract

The present application relates to the blind identification of LDPC code closed set, in particular to a kind of 5G-LDPC code closed set blind identification method based on BP iteration.The main steps include the following: the present application is first operated to the redundancy column and padding column involved in the matrix to be identified, the likelihood value of redundancy column is set to 0, the likelihood value of padding column is set to maximum value, the likelihood value of remaining column is obtained by calculation, then BP iteration calculation is sequentially carried out, and the updated likelihood value is obtained, which is used for the calculation of blind identification LLR, based on the characteristics that correct matrix will converge and error matrix will not converge or converge slowly, the correct matrix can be identified.The present application solves the problem that the traditional blind identification method cannot identify when involving redundancy and padding, and does not transform the original matrix, so the characteristics of low row weight of the original matrix are retained, and the identification performance is guaranteed, thereby improving the identification accuracy when the matrix in the candidate set is small and the similarity is high, and the number of received code words is small.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of blind recognition of channel coding, and particularly relates to a 5G-LDPC code closed set recognition method based on BP iteration. BACKGROUND

[0002] In the field of channel coding, blind recognition technology can recover the encoding parameters used by the sender and even the specific code word from the received signal, so it plays an important role in automatic modulation coding (AMC) and communication countermeasures. Among many coding schemes, LDPC codes are used more and more widely in practice because of their diverse construction methods and excellent performance, but at the same time, they also bring higher blind recognition difficulty, so the blind recognition technology of LDPC codes has attracted more and more attention.

[0003] LDPC code blind recognition technology is mainly divided into two types. The first type is open set (full blind) blind recognition, that is, the recognizer only has a received signal and no other prior knowledge. At present, the open set blind recognition methods mainly include rank criterion method and dual vector based method, which have great recognition difficulty and are difficult to restore the check matrix used by the sender 100%, so they are more commonly used in the field of communication countermeasures. The other type is closed set (semi-blind) blind recognition, in addition to the received signal, the recognizer also knows the candidate set used by the sender, that is, a set consisting of several LDPC codes (such as known protocol adopted by the sender, which only specifies several LDPC codes). At this time, the recognizer returns the most likely candidate. The closed set blind recognition has lower difficulty, higher success rate and lower complexity, but needs more prior knowledge, so it is commonly used in automatic modulation and coding technology, at this time, the sender dynamically selects the modulation and coding scheme according to the channel condition, and the receiver needs to confirm the modulation and coding scheme used by the sender, so as to improve the channel utilization efficiency. The closed set blind recognition mainly uses parity check verification, based on the log likelihood ratio (LLR) obtained from the check relation (SPC), based on the maximum a posteriori probability (APP) described from different angles, and selects the code word with the maximum APP. From the average log likelihood ratio method at the beginning, to the average likelihood difference (LD) method omitting arctanh calculation, to the improved method based on cosine transform (CC), and to the latest two-stage APP estimation (TS) discrimination method, the accuracy of closed set recognition of LDPC codes is getting higher and higher.

[0004] In the field of 5G communication, due to the involvement of deletion and padding, the traditional identification method cannot be used, and at present, there is only one identification method based on linear transformation, which changes the deleted column in the to-be-identified matrix into all 0 through linear transformation, at this time, the deleted column is linearly independent of the received code word, and the original blind identification method can be used for identification.

[0005] However, due to the limitations of the above method, the check equation used for blind identification after linear transformation often has poor identification effect due to large row weight. On the other hand, when the matrix in the candidate set is small and the similarity is high, and the number of received code words is small, the effect is not ideal. SUMMARY

[0006] Therefore, the present application provides a 5G-LDPC code closed set blind identification method based on BP iteration, which improves the accuracy when the candidate set matrix is small and the similarity is high, and the number of received code words is small.

[0007] The 5G-LDPC code closed set blind identification method based on BP iteration of the present application comprises the following steps:

[0008] Step 1: selecting a check matrix from the 5G-LDPC check matrix candidate set, and dividing the received code word according to the size of the selected check matrix;

[0009] In the initial check matrix for BP decoding, the corresponding deleted column likelihood value is 0, the padding column likelihood value is infinite, and the likelihood values of the remaining bits are filled into the likelihood value calculation results of the corresponding received code word;

[0010] Step 2: performing a certain number of BP iterations to update the likelihood values of the variable nodes and check nodes in the check matrix;

[0011] Step 3: based on the new likelihood value of each bit obtained, using the LLR blind identification method to obtain the index value of the current check matrix;

[0012] Step 4: repeating steps 1-3 for each check matrix in the 5G-LDPC check matrix candidate set, and selecting the LDPC check matrix with the maximum index value as the identification result.

[0013] Further, in step 1, the padding node is set to an upper limit value for subsequent calculation.

[0014] Further, in step 2, on the one hand, each check node calculates an updated confidence degree according to the messages received from all other connected variable nodes and transmits it to the target variable node;

[0015] Specifically, the check node will multiply the received belief messages from other connected variable nodes after conversion by the hyperbolic tangent function to represent the constraints of these variables satisfying the check equation; then, the check node converts the result into the corresponding log-likelihood ratio form and transmits it to the target variable node for further iterative update.

[0016] On the other hand, each variable node calculates and updates the information to be sent to the target check node according to the likelihood value of the received initial signal and the messages transmitted from other connected check nodes;

[0017] Specifically, the message sent by the variable node to the check node is the sum of its initial belief and all the messages received from the remaining connected check nodes; this updated belief information represents the current estimate of the variable node for the transmitted bits for the next iteration propagation.

[0018] In step 3, the index value of the check matrix is the index value y i Take the arithmetic mean;

[0019] The index value y i According to the following formula:

[0020]

[0021] N v (i) represents the set of variable nodes participating in the i-th check node, and the likelihood value L(v l ∣r l is obtained by BP iterative calculation.

[0022] The benefits of the present application are: through the idea of BP decoding, the parity bits and padding bits obtain relatively accurate belief information through information propagation, and based on this, blind recognition is performed, solving the problem that 5G-LDPC cannot be blindly recognized due to the existence of parity and padding. And unlike the idea of linear transformation, the original matrix is not transformed, the characteristics of low row weight are retained, and the performance of the identification is guaranteed. The algorithm has obvious advantages compared with the idea of linear transformation when the matrices in the candidate set are small and the similarity is high, and the number of received codewords is small. Our method has good results on the matrix based on the 5G standard, and on the data set, it is improved by 1.5 dB at most than the previous method, and the overall identification performance is better than any previous traditional algorithm. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 is a flowchart of the 5G-LDPC code closed set blind recognition method based on BP iteration in the embodiments of the present application. DETAILED DESCRIPTION

[0024] In this example, the transmission of the codeword is exemplary of BPSK modulation under an AWGN channel.

[0025] First, one of the candidate check matrices is taken, and the received codeword is segmented according to the size of the check matrix. In the process of constructing the initial matrix for BP iteration, the likelihood value of the corresponding punctured bit in the check matrix is taken as 0, the likelihood value of the padding bit is taken as infinity, and the remaining bit is filled according to the received codeword. The likelihood value (log likelihood ratio, LLR) can be calculated according to the following formula,

[0026]

[0027] where r i represents the received signal value, v i represents the actual code symbol, and σ 2 represents the noise variance of the channel.

[0028] Here, BPSK modulation under the traditional AWGN channel is considered. In terms of initial information, since the puncturing and padding need to be considered under the 5G standard, the default punctured node does not have any prior information because it does not actually transmit. Therefore, the initialization likelihood value is 0. Although the padding node does not transmit, it is assumed that all nodes in this part are 0 under this model, which is known information. Therefore, the initialization likelihood value of this part of the node is infinity, which is set as an upper limit value for subsequent calculation. The remaining nodes can also calculate the corresponding likelihood value based on the known channel model and modulation. Based on this, the BP decoding initialization is completed, and normal iteration can be performed.

[0029] Next, each check node calculates an updated belief based on the messages it receives from all other connected variable nodes and passes it to the target variable node. Specifically, the check node will multiply the belief messages received from other connected variable nodes (after hyperbolic tangent function conversion) to represent the constraints of these variables on the check equation. Then, the check node converts the result into the corresponding log likelihood ratio form and passes it to the target variable node for further iteration update.

[0030] Each variable node calculates and updates the information it wants to send to the target check node based on the belief of the received initial signal (log likelihood ratio, LLR value) and the messages passed from other connected check nodes. Specifically, the message sent by the variable node to the check node is the sum of its initial belief and all the messages received from the remaining connected check nodes. This updated belief information represents the current estimate of the variable node on the transmitted bits, which is used for the next iteration propagation.

[0031] The more detailed BP iteration is well known to those skilled in the art and will not be described here.

[0032] For example, consider a check node case, a check node represents a parity check constraint (SPC), consider the case of code length d, satisfying a certain SPC constraint, for bit v0, we have:

[0033]

[0034] After the above formula is arranged, we get:

[0035]

[0036] And for a general binary random variable with probabilities p1 and p0, there is the following relationship

[0037]

[0038]

[0039]

[0040] Here tanh represents the hyperbolic tangent function.

[0041] So for bit v0, we can get the following expression,

[0042]

[0043] Where tanh -1 represents the inverse hyperbolic tangent function, and the generalization of the above formula is:

[0044]

[0045] N v (i) represents the set of variable nodes participating in the ith check node. It is not difficult to find that the above formula is the core formula involved in the transmission of information in BP decoding, that is, the information transmitted by the check node to the variable node. At the same time, there is also a similar identification formula for the LLR blind recognition algorithm, that is, for each check node i, its index value y i According to the following formula:

[0046]

[0047] Where L(v l | r lThe value of the LLR blind identification is calculated based on the BP iteration, so that the problems of the redundancy bit being 0 and the padding bit being infinite are avoided, and the BP decoding iteration part used for blind identification in the whole process can be used as part of the decoding iteration after the correct matrix is identified, so as to further reduce the decoding time.

[0048] After the above iteration, the confidence information of the redundancy node and the padding node is no longer 0 and infinite, but is updated to a new value through the above information propagation process, and in this process, the redundancy node and the padding node will finally converge for the correct matrix. At this time, the LLR blind identification method is used to identify whether the constraint form of the check equation is satisfied, that is, the correct matrix will converge to a larger value, and the incorrect matrix will not converge or converge slowly. Based on this, the correct matrix can be identified by the LLR blind identification method based on the setting of the appropriate number of cycles.

[0049] For the correct matrix, the bit likelihood value converges in the process of information transmission in the BP decoding iteration, and the y i corresponding to the correct matrix also converges, and because it satisfies the SPC check, the final result will make the value calculated by the LLR blind identification also converge to a larger value.

[0050] For the incorrect matrix, because the BP decoding iteration will not finally converge or will converge slowly, when the correct matrix has converged, the LLR index value calculated by the incorrect matrix will be smaller.

[0051] Therefore, at this time, the LLR blind identification algorithm finally takes the average of the LLR value of each check node to obtain the identification index value of a single matrix, based on which the correct check matrix can be selected to complete the purpose of blind identification.

[0052] In this experiment, the information data is randomly generated, and the candidate set matrix is generated according to the 5G standard, a total of 8 matrices are generated as the candidate set, and the specific parameters are as follows,

[0053]

[0054] The 8 matrices are commonly used in actual situations, so the 8 matrices are selected to form the candidate set for blind identification experiment. The number of code words is equal to 10, and the number of iterations is equal to 1.

[0055] For the matrix with code length 1536 and code rate 1 / 3, the following results are obtained

[0056]

[0057] The matrix of code length 3072 and code rate 1 / 3 has the following results

[0058]

[0059] The above is only the embodiment of the present application, and the specific technical solutions and / or common knowledge of the scheme are not described in detail. It should be pointed out that for those skilled in the art, without departing from the technical solutions of the present application, a number of modifications and improvements can be made, which should also be considered as the protection scope of the present application, which will not affect the effect and practicality of the present application. The protection scope of the present application should be subject to the content of its claims, and the specific implementation mode and the like recorded in the specification can be used to explain the content of the claims.

Claims

1. A 5G-LDPC code closed-set blind identification method based on BP iteration, characterized in that, Includes the following steps: Step 1: Select a parity check matrix from the 5G-LDPC parity check matrix candidate set, and divide the received codewords according to the size of the selected parity check matrix; In the initial parity check matrix used for BP decoding, the likelihood values ​​of the corresponding pruned columns are set to 0, the likelihood values ​​of the filled columns are set to infinity, and the likelihood values ​​of the remaining bits are filled into the likelihood value calculation results of the corresponding received codewords. Step 2: Perform a certain number of backpropagation (BP) iterations, updating the likelihood values ​​of the variable nodes and check nodes in the check matrix, including: Each verification node calculates an updated confidence score based on the messages it receives from all other connected variable nodes and passes it to the target variable node. The verification node multiplies the confidence messages received from other connected variable nodes after transforming them with the hyperbolic tangent function to indicate that these variables satisfy the constraints of the verification equation. Then, the verification node converts the result into the corresponding log-likelihood ratio form and passes it to the target variable node for further iterative updates. Each variable node calculates and updates the information it wants to send to the target check node based on the likelihood value of the received initial signal and the messages passed from other connected check nodes. The message that the variable node sends to the check node is the sum of its initial confidence and all messages received from the other connected check nodes. This updated confidence information represents the variable node's current estimate of the transmitted bits and is used for the next iteration of propagation. Step 3: Based on the new likelihood value obtained for each bit, use the LLR blind identification method to obtain the index value of the current parity check matrix; the index value of the parity check matrix is ​​the index value for each parity check node. Take the arithmetic mean; Indicator value Calculate according to the following formula: , This represents the set of variable nodes participating in the i-th verification node, where This represents the received codeword signal value. Represents the actual code character symbol, likelihood value Obtained through BP iteration calculation; Step 4: Repeat steps 1-3 for each verification matrix in the 5G-LDPC verification matrix candidate set, and select the verification matrix with the largest index value as the recognition result.

2. The method according to claim 1, characterized in that, In step 1, the fill node is set to an upper limit value for subsequent calculations.

Images