A truncation design, encoding and decoding method

Abstract

The present invention provides a truncation design, encoding, and decoding method for Hadamard-LDPC codes. The truncation design method comprises the following steps: using a differential evolution algorithm to determine the optimal variable node truncation degree distribution and the optimal check node degree distribution with a decoding threshold as the target; determining a first truncation node position based on the optimal variable node truncation degree distribution; randomly generating a second truncation node position with the same optimal variable node truncation degree distribution, calculating the distance between the check node degree distribution corresponding to the first truncation node position and the second truncation node position and the optimal check node degree distribution, and taking the truncation node position with the smaller distance as the more optimal truncation node position; and finding the optimal truncation node position through multiple iterations. The truncation node position designed by the present invention and the corresponding decoding method can support messages of different message lengths and reliable transmission in multiple types of low signal-to-noise ratio environments.

Description

Technical Field

[0001] The present invention relates to the field of electronic information technology, and in particular to a truncation design, encoding and decoding method for Hadamard-LDPC codes. Background Art

[0002] Anti-interference performance is a core indicator of communication systems. For ultra-long-distance communication scenarios such as deep space communication, offshore and remote communication, and for scenarios with strong electromagnetic interference in adversarial environments, there is an urgent need for a set of coding and decoding methods suitable for operating at extremely low signal-to-interference-and-noise ratios and supporting reliable communication transmission. Hadamard-LDPC codes are a type of coding and decoding method that can operate at extremely low signal-to-noise ratios and have the ability to approach the Shannon limit. Traditional Hadamard-LDPC codes are designed to support only one code rate. To adapt to different transmission rates, Hadamard-LDPC codes are often punctured and truncated to improve their adaptability to different scenarios. However, the current truncation method uses random truncation, which is relatively limited. Summary of the Invention

[0003] In view of the problems existing in the prior art, a truncation design, encoding and decoding method for Hadamard-LDPC codes is provided, which can effectively reduce the transmission code rate.

[0004] A first aspect of the present invention proposes a truncation design method for Hadamard-LDPC codes, comprising:

[0005] S1. Determine the variable node degree and degree distribution, and the check node degree and degree distribution according to the Hadamard-LDPC code check matrix;

[0006] S2. Set the initial truncated degree distribution and update the variable node degree distribution according to the initial truncated degree distribution;

[0007] S3. Using the differential evolution algorithm, the updated variable node truncated degree distribution and check node degree distribution are optimized with the decoding threshold as the target, to obtain the optimal variable node truncated degree distribution and the optimal check node degree distribution;

[0008] S4. Determine the position of the first truncation node according to the optimal variable node truncation degree distribution;

[0009] S5. Obtain a corresponding first check node degree distribution according to the first truncated node position and the Hadamard-LDPC code check matrix;

[0010] S6. Calculate a first distance between the first check node degree distribution and the optimal check node degree distribution;

[0011] S7. Randomly generate a second truncated node position with the same optimal variable node truncated degree distribution, and calculate a second distance between the second check node degree distribution corresponding to the second truncated node position and the optimal check node degree distribution. If the second distance is less than the first distance, replace the first truncated node position with the second truncated node position.

[0012] S8. Repeat S7 until a preset condition is met, and output the first truncated node position as the optimal truncated node position.

[0013] As a preferred solution, in S3, the decoding threshold is determined by an external information transfer function of a variable node and a check node.

[0014] As a preferred solution, in S7, randomly generating a second truncated node position with the same optimal variable node truncation degree distribution specifically includes: randomly exchanging truncated variable nodes and non-truncated variable nodes with the same degree to form the second truncated node position.

[0015] As a preferred solution, in S8, the preset condition is that the maximum number of iterations is reached or the second distance is 0.

[0016] A second aspect of the present invention provides an encoding method, comprising:

[0017] receiving a bit sequence to be transmitted;

[0018] According to the optimal truncation node position obtained in the first aspect, the bit sequence to be transmitted is combined with the corresponding all-zero sequence to generate a new bit sequence;

[0019] Channel coding is performed on the new bit sequence, and the filled all-zero bits are removed according to the optimal truncation node position to complete the truncation transmission.

[0020] The third aspect of the present invention provides a decoding method, comprising:

[0021] receiving channel LLR information of the transmitted coded bits;

[0022] Adding an all-zero bit LLR value to the code bit channel LLR information according to the optimal truncation node position obtained in the first aspect;

[0023] Decoding the added LLR values ​​to obtain coded bits;

[0024] The coded bits are truncated according to the optimal truncation node position to obtain a service message bit sequence.

[0025] As a preferred solution, the all-zero bit LLR value is .

[0026] Compared with the existing technology, the beneficial effects of adopting the above technical solution are:

[0027] The present invention designs a truncation position sequence, fills the Hadamard-LDPC code message with zero bits at the corresponding position, and then performs Hadamard-LDPC code channel coding, thereby reducing the transmission code rate. Since the receiving end knows the zero bit position, the channel prior information of the filling bit can be used to improve the decoding performance of the codec under extremely low signal-to-noise ratio. This truncation design method not only effectively alleviates the hardware resource limitations and transmission efficiency problems faced by Hadamard-LDPC codes in practical applications, but also greatly expands the application scope of Hadamard-LDPC codes in resource-constrained communication scenarios. More importantly, this design improves the flexibility of the system, making it possible to adjust the coding parameter configuration according to specific application requirements.

[0028] This invention addresses a gap in current technology by designing Hadamard-LDPC codes for extreme signal-to-noise ratio environments with their proposed truncation node locations. The proposed method, combined with the corresponding decoding method, supports reliable transmission of messages of varying lengths in a variety of signal-to-noise ratio environments. In particular, the proposed method enables Hadamard-LDPC code encoding and decoding to simultaneously support the reliable transmission of long messages in service channels and short messages in control channels. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 This is a flow chart of the truncation design method proposed in an embodiment of the present application.

[0030] Figure 2 Schematic diagram of different types of check nodes proposed in the embodiments of the present application, including: (1) degree 6 node; (2) degree 5 node; (3) degree 4 node;

[0031] Figure 3 This is the output external information of different check nodes when the signal-to-noise ratio is -10dB proposed in the embodiment of the present application.

[0032] Figure 4 This is a schematic diagram of a Hadamard-LDPC code check matrix with a code rate of 1 / 10 proposed in an embodiment of the present application.

[0033] Figure 5 This is the truncated Hadamard-LDPC code encoding process proposed in the embodiment of the present application.

[0034] Figure 6 This is the decoding process of the truncated Hadamard-LDPC code proposed in the embodiment of the present application.

[0035] Figure 7Bit error rate (BER) decoding performance curves of different Hadamard-LDPC codes proposed in the embodiments of this application.

[0036] Figure 8 Frame error rate (FER) decoding performance curves of different Hadamard-LDPC codes proposed in the embodiments of this application. DETAILED DESCRIPTION

[0037] The embodiments of the present application are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar modules or modules with the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application and are not to be construed as limiting the present application. On the contrary, the embodiments of the present application include all changes, modifications, and equivalents that fall within the spirit and scope of the appended claims.

[0038] This application embodiment proposes a truncation design method for Hadamard-LDPC codes, which can effectively improve the decoding performance of the codec under extremely low signal-to-noise ratio. Figure 1 , the specific steps are as follows:

[0039] S1. Determine the variable node degree and degree distribution, and the check node degree and degree distribution according to the Hadamard-LDPC code check matrix;

[0040] S2. Set the initial truncated degree distribution and update the variable node degree distribution according to the initial truncated degree distribution;

[0041] S3. Using the differential evolution algorithm, the updated variable node truncated degree distribution and check node degree distribution are optimized with the decoding threshold as the target to obtain the optimal variable node truncated degree distribution and the optimal check node degree distribution; wherein the decoding threshold is determined by the external information transfer function of the variable node and the check node.

[0042] S4. Determine the position of the first truncation node according to the optimal variable node truncation degree distribution;

[0043] S5. Obtain a corresponding first check node degree distribution according to the first truncated node position and the Hadamard-LDPC code check matrix;

[0044] S6. Calculate a first distance between the first check node degree distribution and the optimal check node degree distribution;

[0045] S7. Randomly generate a second truncated node position having the same optimal variable node truncated degree distribution, and calculate a second distance between the second check node degree distribution and the optimal check node degree distribution corresponding to the second truncated node position. If the second distance is less than the first distance, replace the first truncated node position with the second truncated node position. Randomly generating the second truncated node position having the same optimal variable node truncated degree distribution includes randomly exchanging a truncated variable node and a non-truncated variable node having the same degree to form the second truncated node position.

[0046] S8, repeat S7 until a preset condition is met, outputting the first truncated node position as the optimal truncated node position. According to the truncated node position, the corresponding Hadamard-LDPC codeword is truncated to truncate the node position.

[0047] The following is a detailed description of the truncation method:

[0048] Sequence truncation essentially involves deleting columns from the Hadamard-LDPC parity check matrix. This process changes the truncation degree distribution of the encoded variable and check nodes. Therefore, deleting different columns results in different truncation degree distributions, leading to different decoding performance.

[0049] Assume that the variable node degree determined by the Hadamard-LDPC code check matrix is , the degree distribution is At this time, let the truncation degree distribution be ,in Indicates the deletion degree is The ratio of variable nodes to all variable nodes, After deleting the corresponding column, the variable node degree distribution is updated to The external information transfer function of the variable node decoder is:

[0050]

[0051] in, The degree is d i The variable node information is:

[0052]

[0053] in, is the external information transfer function, is the equivalent channel noise variance, Prior information passed to the check node decoder.

[0054] like Figure 2As shown, for Hadamard-LDPC codes, the degree of their check nodes is fixed to d c After deleting the columns of the check matrix, the degree of the check node will be reduced, that is, the degree is , the check node degree distribution is For degrees i The nodes have d c - i The edge is deleted.

[0055] like Figure 3 As shown in Figure 1, Monte Carlo simulation can be used to obtain the external information transfer curves for check nodes of different degrees. It can be found that the external information transfer curves corresponding to check nodes of different degrees are different, and all external information transfer curves are convex functions. Therefore, the external information transfer graph can be used to calculate the decoding threshold and thus optimize the truncation degree distribution. Based on the check node truncation degree distribution, the external information transfer function of the check node decoder is:

[0056]

[0057] in, The degree is i The check node external information.

[0058] Since the edges connecting the variable nodes are consistent with the edges connecting the check nodes, the variable node truncation degree distribution and check node degree distribution The following constraints exist:

[0059]

[0060] At this time, the differential evolution algorithm is used to obtain the optimal variable node truncation degree distribution with the decoding threshold as the target. and optimal check node degree distribution The decoding threshold is determined according to the external information transfer function of the variable node and the check node, which is a conventional method and will not be described in detail here.

[0061] Then, the first truncation node position that satisfies the truncation degree distribution is generated according to the optimal variable node truncation degree distribution. ,in Record No. i The truncated bits are located at the position of the encoded message.

[0062] Then, according to the first truncated node position S and the Hadamard-LDPC code check matrix, the check node degree distribution corresponding to the first truncated node position is obtained: , calculate the degree distribution and optimal check node degree distribution The distance, that is:

[0063]

[0064] For truncated variable nodes and non-truncated variable nodes with the same degree, swapping them will not change the variable node truncation degree distribution, but will change the check node degree distribution. Randomly swapping truncated variable nodes and non-truncated variable nodes with the same degree to obtain the second truncated node position , and calculate the second truncation node position Corresponding to the second check node degree distribution and optimal check node degree distribution The second distance .like , then use the second truncation node position Replace the first truncation node position .

[0065] Continue to iterate the above process until dis =0 or the maximum number of iterations is reached T . Position the first truncated node Output as the optimal truncated node position.

[0066] Furthermore, an embodiment of the present application also proposes a coding method, including: receiving a bit sequence to be transmitted; merging the bit sequence to be transmitted with a corresponding all-zero sequence according to the optimal truncation node position obtained by the aforementioned method to generate a new bit sequence; channel coding the new bit sequence, and removing the filled all-zero bits according to the optimal truncation node position to complete truncated transmission.

[0067] by Figure 4 As an example, the Hadamard-LDPC code check matrix shown in FIG. has a quasi-cyclic structure, and its parameters are the number of rows of the basis matrix and , number of columns , bit rate 1 / 10, expansion factor , Hadamard order 4, message length 1040, code length 11336.

[0068] Please continue to refer to Figure 5By truncating the Hadamard-LDPC code, lower code rate message transmission and reception can be achieved. For example, a transmission scheme with a message length of 520 bits is implemented. The encoding process is as follows: First, the upper-layer application passes 520 bits to the encoding module. Based on the 520 known truncation node positions, the 520-bit service message is combined with 520 all-zero sequences to form a 1024-bit sequence. Channel coding is performed to obtain 11336 coded bits. Based on the truncation node positions, the padded all-zero bits are removed, shortening the coded bits to 10816 bits, resulting in an overall code rate of 0.0481.

[0069] Furthermore, an embodiment of the present application also provides a decoding method, including: receiving channel LLR information of transmitted coded bits; adding an all-zero bit LLR value to the code bit channel LLR information according to the optimal truncation node position obtained by the aforementioned method; decoding the added LLR value to obtain the coded bit; and truncating the coded bit according to the optimal truncation node position to obtain a service message bit sequence.

[0070] Please refer to Figure 6 Continuing with the above encoding scenario, the decoding process is as follows: First, the channel LLR information of 10816 coded bits is received. At the same time, according to the designed truncation node position, the all-zero bit LLR value is added to the coded bit channel LLR information, and the code length is restored to 11336. In this embodiment, it is known that the all-zero bit value is ,in b The decoder quantizes the number of bits. Next, the 11,336 codeword LLR values ​​are fed into the decoder, yielding 1,024 coded bits. Finally, the 520 transmitted service messages are obtained based on the truncated node positions.

[0071] Please refer to Figure 7 By comparing with the original Hadamard-LDPC code simulation, the truncation method proposed in this invention can further reduce the code rate and achieve correct decoding at a lower signal-to-noise ratio. Compared with the random truncation method simulation, at a BER of 10 -5 The designed truncated node position can achieve a gain of about 0.6dB. Figure 8 , compared with the random truncation method, at an FER of 10 -3 When , the designed truncated node position can achieve a gain of about 0.6dB.

[0072] The present invention proposes for the first time a truncation design method, including Hadamard-LDPC code truncation design and truncation encoding and decoding method, filling the gap in the current Hadamard-LDPC code truncation design method. Through the truncation node position and corresponding decoding method designed by the present invention, it can support messages of different message lengths and reliable transmission in multiple types of low signal-to-noise ratio environments. In particular, through the method proposed by the present invention, the Hadamard-LDPC code encoding and decoding can simultaneously support the reliable transmission of long messages in the service channel and short messages in the control channel, realizing a set of Hadamard-LDPC code encoders and decoders with the ability to support multiple types of transmission requirements. Simulation results show that the truncation node position designed by the present invention achieves a gain of about 0.6dB in both bit error rate and frame error rate compared to the random truncation method.

[0073] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A truncation design method for Hadamard-LDPC codes, characterized in that: include: S1. Determine the variable node degree and degree distribution, and the check node degree and degree distribution according to the Hadamard-LDPC code check matrix; S2. Set the initial truncated degree distribution and update the variable node degree distribution according to the initial truncated degree distribution; S3. Using the differential evolution algorithm, the updated variable node truncation degree distribution and check node degree distribution are optimized with the decoding threshold as the target, to obtain the optimal variable node truncation degree distribution and the optimal check node degree distribution; S4. Determine the position of the first truncation node according to the optimal variable node truncation degree distribution; S5. Obtain a corresponding first check node degree distribution according to the first truncated node position and the Hadamard-LDPC code check matrix; S6. Calculate a first distance between the first check node degree distribution and the optimal check node degree distribution; S7. Randomly generate a second truncated node position with the same optimal variable node truncated degree distribution, and calculate a second distance between the second check node degree distribution corresponding to the second truncated node position and the optimal check node degree distribution. If the second distance is less than the first distance, replace the first truncated node position with the second truncated node position. S8, repeat S7 until the preset condition is met, and output the first truncated node position as the optimal truncated node position; The preset condition is that the maximum number of iterations is reached or the second distance is 0.

2. The truncation design method according to claim 1, characterized in that: In S3, the decoding threshold is determined by the external information transfer function of the variable node and the check node.

3. The truncation design method according to claim 1, characterized in that: In S7, randomly generating a second truncated node position having the same optimal variable node truncation degree distribution specifically includes: randomly exchanging truncated variable nodes and non-truncated variable nodes having the same degree to form the second truncated node position.

4. A coding method, characterized in that include: receiving a bit sequence to be transmitted; According to the optimal truncation node position obtained by the truncation design method according to any one of claims 1 to 3, the bit sequence to be transmitted is combined with the corresponding all-zero sequence to generate a new bit sequence; Channel coding is performed on the new bit sequence, and the filled all-zero bits are removed according to the optimal truncation node position to complete the truncation transmission.

5. A decoding method, characterized in that: include: receiving channel LLR information of the transmitted coded bits; According to the position of the optimal truncation node position obtained by the truncation design method according to any one of claims 1 to 3, an all-zero bit LLR value is added to the code bit channel LLR information; Decoding the added LLR values ​​to obtain coded bits; The coded bits are truncated according to the optimal truncation node position to obtain a service message bit sequence.

6. The decoding method according to claim 5, characterized in that: The all-zero bit LLR value is , where b is the number of decoder quantization bits.

Images