A decision feedback equilibrium method and system based on coding constraints

By introducing iterative collaborative processing of the XOR-extended Hamming code decoder into the decision feedback equalizer and using coding constraints to eliminate erroneous candidate symbols, the error accumulation problem of the decision feedback equalizer in low signal-to-noise ratio and strong inter-symbol interference environments is solved, thus improving the receiving performance.

CN123123080APending Publication Date: 2026-11-17BEIJING UNIV OF POSTS & TELECOMM
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202610955911.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-06-30
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing decision feedback equalizers are prone to accumulating and propagating erroneous decisions in environments with low signal-to-noise ratio, strong inter-symbol interference, or significant distortion in the preceding stage devices, leading to a decline in receiver performance.

Method used

A decision feedback equalization method based on coding constraints is adopted, which combines the iterative collaborative processing of the decision feedback equalizer and the XOR-extended Hamming code decoder. The XOR bit constraint information participates in the PAM-4 symbol decision, and the coding constraints of the XOR-extended Hamming code decoder are used to eliminate erroneous candidate symbols, reduce the probability of erroneous decision, and update the filter coefficients of the decision feedback equalizer.

Benefits of technology

It effectively suppresses the accumulation and propagation of errors in the decision feedback equalizer, improves the transmission reliability at the receiver, and reduces the possibility of erroneous decisions, especially in environments with low signal-to-noise ratio and strong inter-symbol interference.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123123080A_ABST
    Figure CN123123080A_ABST
Patent Text Reader

Abstract

This invention provides a decision feedback equalization method and system based on coding constraints. The method includes multiple iterations, each iteration comprising the following steps: extracting the received sampled input vector corresponding to each symbol time; inputting it into a decision feedback equalizer and outputting an equalized output; determining a candidate decision set based on the decoded XOR bit sequence corresponding to the current iteration; determining the iterative decision symbol sequence for the current iteration; calculating the decision error based on the equalized output and the iterative decision symbol sequence; calculating the original bit log-likelihood ratio for each symbol time and generating an XOR domain soft information sequence based on the XOR relation; inputting the XOR domain soft information sequence into an XOR-extended Hamming code decoder to obtain a decoded XOR bit sequence and a correction flag sequence; and updating the forward filter coefficients and the feedback filter coefficients of the decision feedback equalizer based on the correction flag sequence output from the previous iteration and the decision error sequence of the current iteration.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of optical transmission technology, and in particular to a decision feedback equalization method and system based on coding constraints. Background Technology

[0002] With the rapid development of data center optical interconnects, passive optical networks, and short-distance high-speed optical communication systems, four-level pulse amplitude modulation (PAM-4) systems based on intensity modulation / direct detection are widely used in high-speed optical communication links due to their relatively simple structure, low cost, and high spectral efficiency. However, as transmission rates increase, limitations in device bandwidth, fiber transmission defects, nonlinear distortion, and receiver noise all contribute to the received signal, reducing the level spacing between PAM-4 symbols and generating stronger inter-symbol interference.

[0003] Existing receivers typically employ decision feedback equalizers to compensate for inter-symbol interference (ISI). Decision feedback equalizers use historical decision results to estimate and cancel the downstream interference experienced by the current symbol, improving receiver performance under certain signal-to-noise ratio (SNR) conditions. However, the feedback path of this type of equalizer relies on hard decision results. When the system is in an environment with low SNR, strong ISI, or significant distortion in the preceding components, an erroneous decision for a particular symbol may be used by the feedback filter to cancel interference in subsequent symbols, causing errors to accumulate and propagate in the feedback path, resulting in burst errors. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide a decision feedback equilibrium method and system based on coding constraints to eliminate or improve one or more defects existing in the prior art.

[0005] One aspect of the present invention provides a decision feedback equalization method based on coding constraints, the method being applied to the receiver of an optical transmission system, the receiver including a decision feedback equalizer and an XOR-extended Hamming code decoder, the method comprising multiple iteration rounds, each iteration round including the following steps: Obtain the received PAM-4 sampling sequence, and extract the received sampling input vector corresponding to each symbol time from the PAM-4 sampling sequence; Each of the received sampled input vectors is input into a decision feedback equalizer, and the decision feedback equalizer outputs an equalized output corresponding to the received sampled input vectors. The candidate decision set for the current iteration round is determined based on the decoded XOR bit sequence corresponding to the current iteration round, and the XOR bit constraint state corresponding to the current iteration round is output by the XOR-extended Hamming code decoder of the previous round. Based on the candidate decision set and the balanced output corresponding to each symbol time, the iterative decision symbol sequence for the current iteration round is determined; The decision error is calculated based on the balanced output corresponding to each symbol time and the iterative decision symbol sequence, and the decision error sequence of the current iteration is determined. The log-likelihood ratio of the original bits at each symbol time is calculated based on the equalization output corresponding to each symbol time, and the XOR domain soft information sequence is generated based on the XOR relationship between the two original bits in the PAM-4 Gray map. The XOR field soft information sequence is input into an XOR-extended Hamming code decoder to obtain the decoded XOR bit sequence and correction flag sequence for the current iteration round. The correction flag sequence includes multiple correction flags corresponding to the symbol time. Based on the correction flag sequence output from the previous iteration and the decision error sequence of the current iteration, the adaptive update step size corresponding to each symbol time is determined, and the forward filter coefficients and the feedback filter coefficients of the decision feedback equalizer are updated based on the adaptive update step size corresponding to each symbol time.

[0006] By adopting the above scheme, this invention introduces iterative collaborative processing between the decision feedback equalizer and the XOR-extended Hamming code decoder, enabling XOR bit constraint information to participate in the PAM-4 symbol decision process. In each iteration, the received sampling input vector corresponding to each symbol time is extracted from the received PAM-4 sampling sequence, and the decision feedback equalizer outputs the equalization output corresponding to each symbol time. Then, the XOR bit constraint state available in the current iteration is formed by using the output of the XOR-extended Hamming code decoder from the previous iteration, determining the candidate decision set, and obtaining the iterative decision symbol sequence based on the candidate decision set and the equalization output corresponding to each symbol time. Thus, PAM-4 symbol decision no longer depends solely on the distance relationship between the equalization output and the complete constellation points, but can combine coding constraints to exclude candidate symbols that are inconsistent with the XOR bits after decoding. This reduces the probability of erroneous decisions caused by noise disturbances in environments with low signal-to-noise ratio, strong inter-symbol interference, or heavy distortion of the preceding devices, and reduces the possibility of erroneous symbols entering the feedback filter and continuing to affect the subsequent symbol interference cancellation, thereby suppressing the accumulation and propagation of errors in the decision feedback equalizer from the source.

[0007] In some embodiments of the present invention, the method further includes decision feedback equalizer initialization, wherein initialization training is performed based on a preset training sequence to obtain the initial forward filter coefficients and feedback filter coefficients of the decision feedback equalizer.

[0008] In some embodiments of the present invention, in the step of determining the candidate decision set for the current iteration round based on the decoded XOR bit sequence corresponding to the current iteration round: The PAM-4 constellation set is divided into a first PAM-4 symbol subset and a second PAM-4 symbol subset; the first PAM-4 symbol subset includes -3 and +1, and the second PAM-4 symbol subset includes -1 and +3; For any symbol time, if the XOR bit in the decoded XOR bit sequence output from the previous iteration is 0, the first PAM-4 symbol subset is determined as the candidate decision set corresponding to that symbol time. If the XOR bit in the decoded bit sequence output from the previous iteration is 1, then the second PAM-4 symbol subset is determined as the candidate decision set corresponding to that symbol time.

[0009] In some embodiments of the present invention, in the step of determining the iterative decision symbol sequence of the current iteration round based on the candidate decision set and the equalization output corresponding to each symbol time, for any symbol time, the distance metric between the equalization output corresponding to the symbol time and each PAM-4 candidate symbol in the candidate decision set corresponding to the symbol time is calculated, and the PAM-4 candidate symbol with the smallest distance metric is determined as the iterative decision symbol of the symbol time; the iterative decision symbols of each symbol time are combined in the order of the symbol times to obtain the iterative decision symbol sequence of the current iteration round.

[0010] In some embodiments of the present invention, in the step of calculating the raw bit log-likelihood ratio of each symbol time based on the equalization output corresponding to each symbol time: calculating the conditional probability of each PAM-4 candidate symbol under each equalization output; and calculating the raw bit log-likelihood ratio of each symbol time based on the conditional probability.

[0011] In some embodiments of the present invention, in the step of calculating the conditional probability of each PAM-4 candidate symbol under each equalization output, the conditional probability of each PAM-4 candidate symbol under each equalization output is calculated using the following formula: in, Indicates in PAM-4 candidate symbols Achieve balanced output The conditional probability; Indicates the first Within the iteration rounds, the _th The balanced output corresponding to each symbol time; This represents the noise variance estimated during the initial training phase. The raw bit log-likelihood ratio at each symbol time is calculated using the following formula: in, Indicates the first candidate symbol in PAM-4 The second PAM-4 symbol subset with each original bit valued as 1; Indicates the first candidate symbol in PAM-4 The first subset of PAM-4 symbols with original bits set to 0. Indicates the first Within the iteration rounds, the _th The first symbol time corresponding to the th symbol time The log-likelihood ratio of the original bits.

[0012] In some embodiments of the present invention, in the step of generating an XOR domain soft information sequence based on the XOR relationship of two original bits in the PAM-4 Gray map, for any symbol time, based on the conditional probabilities that the two original bits are 0 and 1 respectively, a first XOR probability that the XOR result of the two original bits is 1 and a second XOR probability that the XOR result of the two original bits is 0 are determined; the ratio of the first XOR probability to the second XOR probability is used as the log-likelihood ratio of the XOR domain corresponding to that symbol time; the log-likelihood ratios of the XOR domain corresponding to each symbol time are sequentially combined to obtain the XOR domain soft information sequence.

[0013] In some embodiments of the present invention, in the step of inputting the XOR field soft information sequence into an XOR-extended Hamming code decoder to obtain the decoded XOR bit sequence and correction flag sequence for the current iteration, a hard decision is made on the XOR field soft information sequence to obtain an initial XOR codeword; a companion check is performed on the initial XOR codeword based on the parity check matrix of the XOR-extended Hamming code decoder; if the companion check result satisfies the check constraint, the initial XOR codeword is used as the decoded XOR bit sequence; if the companion check result does not satisfy the check constraint, multiple least reliable positions are determined based on the reliability ranking of the XOR field soft information sequence, a finite flip pattern is generated around the multiple least reliable positions, and a candidate XOR codeword set is generated based on the finite flip pattern; a decoded XOR bit sequence is selected from the candidate XOR codeword set according to the soft information metric; a correction flag sequence is generated based on the difference between the decoded XOR bit sequence and the initial XOR codeword.

[0014] In some embodiments of the present invention, in the steps of determining the adaptive update step size corresponding to each symbol time based on the correction flag sequence output from the previous iteration and the decision error sequence of the current iteration, and updating the forward filter coefficients and the feedback filter coefficients of the decision feedback equalizer based on the adaptive update step size corresponding to each symbol time, for the correction flag of the previous iteration representing an uncorrected symbol time, the basic update step size is determined as the adaptive update step size corresponding to that symbol time; for the correction flag of the previous iteration representing a corrected symbol time, the basic update step size is suppressed based on a step size suppression factor less than 1 to obtain the adaptive update step size corresponding to that symbol time; based on the adaptive update step size corresponding to that symbol time, the decision error, the received sampled input vector, and the historical decision symbol vector, the forward filter coefficients and the feedback filter coefficients are updated respectively.

[0015] In some embodiments of the present invention, in the step of updating the forward filter coefficients and the feedback filter coefficients based on the adaptive update step size, decision error, received sampling input vector, and historical decision symbol vector corresponding to the symbol time, the forward filter coefficients and the feedback filter coefficients are calculated using the following formula: in, Indicates the first Processing within the iteration rounds Forward filter coefficients at each symbol time; Indicates the first The forward filter coefficients are updated at time +1 symbol; This indicates an adaptive update step size; Indicates the error in the decision; Indicates the received sampled input vector; Indicates the first Processing within the iteration rounds Feedback filter coefficients at each symbol time; Indicates the first The feedback filter coefficients are updated at time +1 symbol. This represents the vector of historical decision symbols.

[0016] A second aspect of the present invention also provides a decision feedback equalization system based on coding constraints. The system includes a computer device, the computer device including a processor and a memory, the memory storing computer instructions, and the processor executing the computer instructions stored in the memory. When the computer instructions are executed by the processor, the system implements the steps of the method described above.

[0017] A third aspect of the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the aforementioned decision feedback equilibrium method based on coding constraints.

[0018] Additional advantages, objects, and features of the invention will be set forth in part in the description which follows, and will also become apparent in part to those skilled in the art upon studying the text, or may be learned by practice of the invention. The objects and other advantages of the invention will become apparent from the description and the accompanying drawings.

[0019] Those skilled in the art will understand that the objectives and advantages achievable with the present invention are not limited to those specifically described above, and that the above and other objectives achievable with the present invention will become clearer from the following detailed description. Attached Figure Description

[0020] The accompanying drawings, which are provided to further illustrate the invention and form part of this application, are not intended to limit the scope of the invention.

[0021] Figure 1 This is a flowchart illustrating one implementation of the decision feedback equilibrium method based on coding constraints in this scheme. Figure 2 This is a schematic diagram of the architecture of the decision feedback equilibrium method based on coding constraints in this scheme; Figure 3 This is a schematic diagram illustrating the experimental results of the decision feedback equilibrium method based on coding constraints proposed in this scheme. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the embodiments and accompanying drawings. Here, the illustrative embodiments and descriptions of this invention are used to explain the invention, but are not intended to limit the invention.

[0023] It should also be noted that, in order to avoid obscuring the invention with unnecessary details, only the structures and / or processing steps closely related to the solution according to the invention are shown in the accompanying drawings, while other details that are not closely related to the invention are omitted.

[0024] like Figure 1 As shown, this invention proposes a decision feedback equalization method based on coding constraints. This method is applied to the receiver of an optical transmission system, where the transmitter employs PAM-4 modulation and XOR-extended Hamming code encoding. A transport block can include multiple PAM-4 symbols. In one implementation, every 60 information symbols correspond to 120 original information bits, with an additional 8 parity bits, forming a transport block of 64 PAM-4 symbols. The receiving end performs photoelectric conversion, sampling, and symbol synchronization on the received optical signal to obtain a received PAM-4 sampling sequence. This received PAM-4 sampling sequence serves as the input data source for the decision feedback equalizer. The XOR-extended Hamming code decoder decodes the XOR-domain soft information sequence generated by the equalization output and feeds the decoding result back to the next iteration.

[0025] The receiver completes a closed-loop process in each iteration. This closed-loop process inputs the entire XOR field soft information sequence within a transport block into the XOR-extended Hamming code decoder. The decoded XOR bit sequence and the corrected positions output by the decoder are recorded and reused in the next iteration.

[0026] The receiving end includes a decision feedback equalizer and an XOR-extended Hamming code decoder. The method includes multiple iteration rounds, and the steps of each iteration round include: Step S100: Obtain the received PAM-4 sampling sequence, and extract the received sampling input vector corresponding to each symbol time from the PAM-4 sampling sequence; Step S200: Input each of the received sampling input vectors into the decision feedback equalizer, and the decision feedback equalizer outputs an equalized output corresponding to the received sampling input vectors; In the specific implementation process, the decision feedback equalizer adopts the decision feedback equalizer (DFE) algorithm, which includes a forward filter, a decision unit and a feedback filter. The forward filter processes the current and adjacent received sample values, and the feedback filter uses the historical symbols that have been decided to estimate and cancel the inter-run code interference.

[0027] By adopting the above scheme, the received sampling input vector is extracted from the received PAM-4 sampling sequence according to the symbol time, and the decision feedback equalization is performed in combination with the historical decision symbol vector. The forward filter can be used to compensate for the front-running inter-symbol interference, and the feedback filter can be used to cancel the back-running interference of the decided historical symbols on the current output, so that the equalized output at each symbol time is closer to the PAM-4 constellation point of the transmitter.

[0028] Step S300: Determine the candidate decision set for the current iteration round based on the decoded XOR bit sequence corresponding to the current iteration round. The XOR bit constraint state corresponding to the current iteration round is output by the XOR-extended Hamming code decoder of the previous round. Step S400: Based on the candidate decision set and the balanced output corresponding to each symbol time, determine the iterative decision symbol sequence for the current iteration round; Step S500: Calculate the decision error based on the equalization output corresponding to each symbol time and the iterative decision symbol sequence, and determine the decision error sequence for the current iteration round; Step S600: Calculate the log-likelihood ratio of the original bits at each symbol time based on the equalization output corresponding to each symbol time, and generate the XOR domain soft information sequence based on the XOR relationship of the two original bits in the PAM-4 Gray map. Step S700: Input the XOR field soft information sequence into the XOR-extended Hamming code decoder to obtain the decoded XOR bit sequence and correction flag sequence for the current iteration round. The correction flag sequence includes multiple correction flags corresponding to the symbol time. Step S800: Based on the correction flag sequence output from the previous iteration and the decision error sequence of the current iteration, determine the adaptive update step size corresponding to each symbol time, and update the forward filter coefficients and the feedback filter coefficients of the decision feedback equalizer based on the adaptive update step size corresponding to each symbol time.

[0029] By adopting the above scheme, this invention introduces iterative collaborative processing between the decision feedback equalizer and the XOR-extended Hamming code decoder, enabling XOR bit constraint information to participate in the PAM-4 symbol decision process. In each iteration, the received sampling input vector corresponding to each symbol time is extracted from the received PAM-4 sampling sequence, and the decision feedback equalizer outputs the equalization output corresponding to each symbol time. Then, the XOR bit constraint state available in the current iteration is formed by using the output of the XOR-extended Hamming code decoder from the previous iteration, determining the candidate decision set, and obtaining the iterative decision symbol sequence based on the candidate decision set and the equalization output corresponding to each symbol time. Thus, PAM-4 symbol decision no longer depends solely on the distance relationship between the equalization output and the complete constellation points, but can combine coding constraints to exclude candidate symbols that are inconsistent with the XOR bits after decoding. This reduces the probability of erroneous decisions caused by noise disturbances in environments with low signal-to-noise ratio, strong inter-symbol interference, or heavy distortion of the preceding devices, and reduces the possibility of erroneous symbols entering the feedback filter and continuing to affect the subsequent symbol interference cancellation, thereby suppressing the accumulation and propagation of errors in the decision feedback equalizer from the source.

[0030] In some embodiments of the present invention, the method further includes decision feedback equalizer initialization, wherein initialization training is performed based on a preset training sequence to obtain the initial forward filter coefficients and feedback filter coefficients of the decision feedback equalizer.

[0031] During the initial training phase, the receiver utilizes the known training sequence. Construct the training received sample input vector with the corresponding training received sample values. And construct the historical known symbol vector during the training phase. .

[0032] In one implementation, the training output of the decision feedback equalizer can be expressed as: in, Indicates the training phase Equal output at each symbol time; This represents the initialized forward filter coefficient vector; This represents the initialized feedback filter coefficient vector; Indicates the training phase The received sampled input vector corresponding to each symbol time; Indicates the training phase The historical training symbol vector prior to each symbol time is known; superscript This indicates transpose.

[0033] Initial training can be adjusted using either the DD-LMS algorithm or the LMS algorithm. and This makes the training output approximate the known training symbol.

[0034] The training error can be expressed as: in, Indicates the training phase The training error at each symbol time; Indicates the first training sequence Known PAM-4 training symbols at symbol times.

[0035] The noise variance during the training phase can be estimated based on the training mean square error as follows: in, This represents the noise variance estimated during the training phase. Indicates the number of training symbols contained in the training sequence; Indicates the first The balanced output corresponding to each training symbol; Indicates the first The known PAM-4 symbol values ​​of the training symbols are used for subsequent symbol conditional likelihood and log-likelihood ratio calculations.

[0036] In some embodiments of the present invention, in the step of determining the candidate decision set for the current iteration round based on the decoded XOR bit sequence corresponding to the current iteration round: The PAM-4 constellation set is divided into a first PAM-4 symbol subset and a second PAM-4 symbol subset; the first PAM-4 symbol subset includes -3 and +1, and the second PAM-4 symbol subset includes -1 and +3; For any symbol time, if the XOR bit in the decoded XOR bit sequence output from the previous iteration is 0, the first PAM-4 symbol subset is determined as the candidate decision set corresponding to that symbol time. If the XOR bit in the decoded bit sequence output from the previous iteration is 1, then the second PAM-4 symbol subset is determined as the candidate decision set corresponding to that symbol time.

[0037] The PAM-4 Gray mapping relationship can be expressed as: For each PAM-4 symbol, there are two raw bits. and Define the XOR bit: in, This represents the XOR bit corresponding to the PAM-4 symbol; This represents the high-order original bits corresponding to the PAM-4 symbol; This represents the low-order original bits corresponding to the PAM-4 symbol; This represents the XOR operation.

[0038] From the Gray mapping above, we can see that if If the bit combination is 00 or 11, the corresponding PAM-4 symbol belongs to the first PAM-4 symbol subset; if If the bit combination is 01 or 10, the corresponding PAM-4 symbol belongs to the second subset of PAM-4 symbols: in, This represents the first subset of PAM-4 symbols corresponding to when the XOR bit value is 0; This represents the second PAM-4 symbol subset corresponding to the XOR bit value of 1.

[0039] In some embodiments of the present invention, in the step of determining the iterative decision symbol sequence of the current iteration round based on the candidate decision set and the equalization output corresponding to each symbol time, for any symbol time, the distance metric between the equalization output corresponding to the symbol time and each PAM-4 candidate symbol in the candidate decision set corresponding to the symbol time is calculated, and the PAM-4 candidate symbol with the smallest distance metric is determined as the iterative decision symbol of the symbol time; the iterative decision symbols of each symbol time are combined in the order of the symbol times to obtain the iterative decision symbol sequence of the current iteration round.

[0040] Specifically, for the first At any symbol time in each iteration round The receiving end is in the candidate decision set Selection and equalization of output The PAM-4 candidate symbol with the smallest distance is used as the iterative decision symbol at that symbol time: in, Indicates the first In the nth iteration round The iterative decision symbol corresponding to each symbol time; Represents the PAM-4 candidate symbols in the candidate decision set; This represents the distance metric between the balanced output and the candidate symbols.

[0041] By combining the iterative decision symbols at each sign time in sequence, we can obtain the iterative decision symbol sequence for the current iteration round: in, Indicates the first The iterative decision symbol sequence of each iteration round; This indicates the number of PAM-4 symbols processed within the same transport block.

[0042] For the At each symbol time, the decision error can be expressed as: in, Indicates the first In the nth iteration round The decision errors corresponding to each symbol time are combined in order of symbol time to obtain the decision error sequence for the current iteration: in, Indicates the first The decision error sequence for each iteration.

[0043] In some embodiments of the present invention, in the step of calculating the raw bit log-likelihood ratio of each symbol time based on the equalization output corresponding to each symbol time: calculating the conditional probability of each PAM-4 candidate symbol under each equalization output; and calculating the raw bit log-likelihood ratio of each symbol time based on the conditional probability.

[0044] In some embodiments of the present invention, in the step of calculating the conditional probability of each PAM-4 candidate symbol under each equalization output, the conditional probability of each PAM-4 candidate symbol under each equalization output is calculated using the following formula: in, Indicates in PAM-4 candidate symbols Achieve balanced output The conditional probability; Indicates the first Within the iteration rounds, the _th The balanced output corresponding to each symbol time; This represents the noise variance estimated during the initial training phase. The raw bit log-likelihood ratio at each symbol time is calculated using the following formula: in, Indicates the first candidate symbol in PAM-4 The second PAM-4 symbol subset with each original bit valued as 1; Indicates the first candidate symbol in PAM-4 The first subset of PAM-4 symbols with original bits set to 0.

[0045] In some embodiments of the present invention, in the step of generating an XOR domain soft information sequence based on the XOR relationship of two original bits in the PAM-4 Gray map, for any symbol time, based on the conditional probabilities that the two original bits are 0 and 1 respectively, a first XOR probability that the XOR result of the two original bits is 1 and a second XOR probability that the XOR result of the two original bits is 0 are determined; the ratio of the first XOR probability to the second XOR probability is used as the log-likelihood ratio of the XOR domain corresponding to that symbol time; the log-likelihood ratios of the XOR domain corresponding to each symbol time are sequentially combined to obtain the XOR domain soft information sequence.

[0046] In some embodiments of the present invention, in the step based on the conditional probabilities of the two original bits being 0 and 1 respectively, for the first... At symbol time, let there be two original bits. and The log-likelihood ratios are respectively and We can first obtain the conditional probability of each original bit being 0 or 1 based on the log-likelihood ratio: in, Indicates the first In the nth iteration round The first symbol time corresponding to the th symbol time The conditional probability that each of the original bits is 1; Indicates the first In the nth iteration round The first symbol time corresponding to the th symbol time The conditional probability that each original bit is 0; Due to XOR bits Under the approximation that the two original bits are conditionally independent, the conditional probabilities of the XOR bit being 1 and 0 can be expressed as: in, Indicates the first In the nth iteration round The conditional probability that the XOR bit at each symbol time is 1; Indicates the first In the nth iteration round The conditional probability that the XOR bit at each symbol time is 0.

[0047] No. The XOR field soft information corresponding to each symbol time can be represented as: in, Indicates the first In the nth iteration round The XOR field soft information corresponding to each symbol time. Combine the symbols according to the symbol time order. The XOR domain soft information sequence is obtained: in, Indicates the first A sequence of XOR field soft information for each iteration round.

[0048] In some embodiments of the present invention, in the step of inputting the XOR field soft information sequence into an XOR-extended Hamming code decoder to obtain the decoded XOR bit sequence and correction flag sequence for the current iteration, a hard decision is made on the XOR field soft information sequence to obtain an initial XOR codeword; a companion check is performed on the initial XOR codeword based on the parity check matrix of the XOR-extended Hamming code decoder; if the companion check result satisfies the check constraint, the initial XOR codeword is used as the decoded XOR bit sequence; if the companion check result does not satisfy the check constraint, multiple least reliable positions are determined based on the reliability ranking of the XOR field soft information sequence, a finite flip pattern is generated around the multiple least reliable positions, and a candidate XOR codeword set is generated based on the finite flip pattern; a decoded XOR bit sequence is selected from the candidate XOR codeword set according to the soft information metric; a correction flag sequence is generated based on the difference between the decoded XOR bit sequence and the initial XOR codeword.

[0049] In the specific implementation process, for the first In one iteration, the XOR-extended Hamming code decoder first performs a hard decision on the soft information sequence in the XOR field to obtain the initial XOR codeword: in, Indicates the first In the nth iteration round The initial XOR hard decision bits corresponding to each symbol time; This represents the XOR field soft information corresponding to the time of the symbol.

[0050] Each The frames are combined to form the initial XOR codeword. And based on the parity-check matrix of XOR-extended Hamming code. Calculate the syndrome: in, Indicates the first The adjoint expression for each iteration round; The parity-check matrix represents the XOR-extended Hamming code; This represents the modulo operation of a binary field.

[0051] like This indicates that the initial XOR codeword satisfies the check constraint, and can be... As the XOR bit sequence after decoding If the syntactic expression is non-zero, then it is based on the absolute value of the soft information in each XOR field. Select several low-reliability locations from smallest to largest to construct a set of low-reliability locations. Let the number of low-reliability locations be... The maximum flip weight is The test set of flipped patterns can be represented as: in, Indicates the first A set of test flip patterns constructed in each iteration round; This represents a test flip pattern; Indicates the Hamming weight of the test flip pattern; This indicates the maximum allowable flip weight.

[0052] For each test flip pattern, the receiver constructs the corresponding error pattern and flips the initial XOR codeword to obtain a candidate codeword. For candidate codewords obtained through syndrome check or single-error correction... Calculate soft information metrics: in, Indicates candidate codewords Soft information measurement; Indicates the first candidate codeword The XOR bit values ​​at each position are used. The candidate codeword with the largest soft information metric is selected as the XOR bit sequence after decoding in the current iteration round: in, Indicates the first The XOR bit sequence of the decoded output from each iteration round; This represents the set of candidate decoders.

[0053] Correction marks can be defined as: in, Indicates the first In the nth iteration round The correction flag corresponding to each symbol moment; This indicates that the position remains consistent before and after decoding; This indicates that the position was corrected during the decoding process. Arranged in symbol time order. This yields a sequence of correction flags used for selecting the tap update step size in the next iteration.

[0054] In some embodiments of the present invention, in the steps of determining the adaptive update step size corresponding to each symbol time based on the correction flag sequence output from the previous iteration and the decision error sequence of the current iteration, and updating the forward filter coefficients and the feedback filter coefficients of the decision feedback equalizer based on the adaptive update step size corresponding to each symbol time, for the correction flag of the previous iteration representing an uncorrected symbol time, the basic update step size is determined as the adaptive update step size corresponding to that symbol time; for the correction flag of the previous iteration representing a corrected symbol time, the basic update step size is suppressed based on a step size suppression factor less than 1 to obtain the adaptive update step size corresponding to that symbol time; based on the adaptive update step size corresponding to that symbol time, the decision error, the received sampled input vector, and the historical decision symbol vector, the forward filter coefficients and the feedback filter coefficients are updated respectively.

[0055] In the specific implementation process, in the first In the nth iteration round, for the nth At each symbol time, the receiver uses the correction flag output from the previous iteration. Determine the step size weight: in, Indicates the first In the nth iteration round The step size weight corresponding to each symbol time; This indicates the preset step size suppression factor, and ; Indicates the th iteration in the previous iteration. The correction flag corresponding to each symbol time. If the current iteration is the first iteration and the correction flag from the previous iteration has not yet been obtained, then we can let... .

[0056] The adaptive update step size can be expressed as: in, Indicates the first In the nth iteration round The adaptive update step size corresponding to each symbol time; This indicates the baseline update step size.

[0057] In some embodiments of the present invention, in the step of updating the forward filter coefficients and the feedback filter coefficients based on the adaptive update step size, decision error, received sampling input vector, and historical decision symbol vector corresponding to the symbol time, the forward filter coefficients and the feedback filter coefficients are calculated using the following formula: in, Indicates the first Processing within the iteration rounds Forward filter coefficients at each symbol time; Indicates the first The forward filter coefficients are updated at time +1 symbol; This indicates an adaptive update step size; Indicates the error in the decision; Indicates the received sampled input vector; Indicates the first Processing within the iteration rounds Feedback filter coefficients at each symbol time; Indicates the first The feedback filter coefficients are updated at time +1 symbol. This represents the vector of historical decision symbols.

[0058] like Figure 2 As shown, in the steps of this scheme, it is determined whether the current iteration count has reached the preset loop limit. If the loop limit has not been reached, the process returns to continue executing the next round of pseudo-Turbo iteration; if the loop limit has been reached, the final equalization result and the corresponding decision symbol sequence are output, and PAM-4 demapping and bit recovery are completed.

[0059] In the specific implementation process, the receiving end can set the maximum number of iteration rounds. You can also set a threshold for the change in the XOR bit sequence after decoding or an error convergence threshold. If the current iteration round... If the XOR bit sequence still changes after decoding, then proceed to the next iteration round; if The iteration ends if the XOR bit sequences are the same after two consecutive rounds of decoding, or if the average energy of the decision error sequence is less than the preset error threshold.

[0060] The final output may include the iterative decision symbol sequence of the last iteration round, the decoded XOR bit sequence, and the recovered bit sequence. For each final decision symbol, the receiver can perform PAM-4 demapping according to the PAM-4 Gray mapping relationship and recover the original bits by combining the XOR-extended Hamming code decoding result.

[0061] Specifically, to verify the ability of the proposed coding constraint DFE to suppress error propagation, the number of burst errors, average interval, and average length in a received sequence of length 9581 were statistically analyzed. The results are as follows: Figure 3 As shown in (a), (b), and (c), with the received optical power (ROP) increasing from... 2 dBm reduced to At 10 dBm, channel conditions gradually deteriorate, and burst errors increase in all three schemes, but coding-constrained DFE consistently shows a significant advantage. In terms of the number of burst errors, coding-constrained DFE is superior. 2 dBm and At 4dBm, burst errors are completely eliminated, and the number of bursts is significantly lower than that of the classic DFE and the pre-equalization signal under low ROP conditions, indicating that it can effectively suppress error propagation. Meanwhile, the average burst interval of the coding-constrained DFE is significantly larger than other schemes, indicating that this scheme can maintain a longer continuous error-free transmission interval. Furthermore, its average burst length is consistently close to 1, indicating that even if errors occur, they are mostly confined to a single symbol, preventing the error from spreading to subsequent symbols. In summary, the proposed coding-constrained DFE can effectively reduce the frequency of burst errors, increase the error-free transmission interval, and significantly suppress error propagation, thereby improving the overall transmission reliability of the system.

[0062] This invention also provides a decision feedback equalization system based on coding constraints. The system includes a computer device, which includes a processor and a memory. The memory stores computer instructions, and the processor executes the computer instructions stored in the memory. When the computer instructions are executed by the processor, the system implements the steps of the method described above.

[0063] This invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the aforementioned decision feedback equalization method based on coding constraints. The computer-readable storage medium can be a tangible storage medium, such as random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, floppy disks, hard disks, removable storage disks, CD-ROMs, or any other form of storage medium known in the art.

[0064] Those skilled in the art will understand that the exemplary components, systems, and methods described in conjunction with the embodiments disclosed herein can be implemented in hardware, software, or a combination of both. Whether implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this invention. When implemented in hardware, it can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this invention are programs or code segments used to perform the desired tasks. The programs or code segments can be stored in a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried in a carrier wave.

[0065] It should be clarified that the present invention is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of the present invention.

[0066] In this invention, features described and / or illustrated for one embodiment may be used in the same or similar manner in one or more other embodiments, and / or combined with or in place of features of other embodiments.

[0067] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, various modifications and variations of the embodiments of the present invention are possible. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method of decision feedback equalization based on coding constraints, characterized by, The method comprises multiple iterations, each iteration including the following steps: Obtain the received PAM-4 sampling sequence, and extract the received sampling input vector corresponding to each symbol time from the PAM-4 sampling sequence; Each of the received sampled input vectors is input into a decision feedback equalizer, and the decision feedback equalizer outputs an equalized output corresponding to the received sampled input vectors. The candidate decision set for the current iteration round is determined based on the decoded XOR bit sequence corresponding to the current iteration round, and the XOR bit constraint state corresponding to the current iteration round is output by the XOR-extended Hamming code decoder of the previous round. Based on the candidate decision set and the balanced output corresponding to each symbol time, the iterative decision symbol sequence for the current iteration round is determined; The decision error is calculated based on the balanced output corresponding to each symbol time and the iterative decision symbol sequence, and the decision error sequence of the current iteration is determined. The log-likelihood ratio of the original bits at each symbol time is calculated based on the equalization output corresponding to each symbol time, and the XOR domain soft information sequence is generated based on the XOR relationship between the two original bits in the PAM-4 Gray map. The XOR field soft information sequence is input into an XOR-extended Hamming code decoder to obtain the decoded XOR bit sequence and correction flag sequence for the current iteration round. The correction flag sequence includes multiple correction flags corresponding to the symbol time. Based on the correction flag sequence output from the previous iteration and the decision error sequence of the current iteration, the adaptive update step size corresponding to each symbol time is determined, and the forward filter coefficients and the feedback filter coefficients of the decision feedback equalizer are updated based on the adaptive update step size corresponding to each symbol time.

2. The method of claim 1, wherein, In the step of determining the candidate decision set for the current iteration round based on the decoded XOR bit sequence corresponding to the current iteration round: The PAM-4 constellation set is divided into a first PAM-4 symbol subset and a second PAM-4 symbol subset; the first PAM-4 symbol subset includes -3 and +1, and the second PAM-4 symbol subset includes -1 and +3; For any symbol time, if the XOR bit in the decoded XOR bit sequence output from the previous iteration is 0, the first PAM-4 symbol subset is determined as the candidate decision set corresponding to that symbol time. If the XOR bit in the decoded bit sequence output from the previous iteration is 1, then the second PAM-4 symbol subset is determined as the candidate decision set corresponding to that symbol time.

3. The method of claim 1, wherein, In the step of determining the iterative decision symbol sequence of the current iteration round based on the candidate decision set and the equalization output corresponding to each symbol time, for any symbol time, the distance metric between the equalization output corresponding to the symbol time and each PAM-4 candidate symbol in the candidate decision set corresponding to the symbol time is calculated, and the PAM-4 candidate symbol with the smallest distance metric is determined as the iterative decision symbol of the symbol time; the iterative decision symbols of each symbol time are combined in the order of the symbol times to obtain the iterative decision symbol sequence of the current iteration round.

4. The method of claim 1, wherein, In the step of calculating the raw bit log-likelihood ratio for each symbol time based on the equalization output corresponding to each symbol time: the conditional probability of each PAM-4 candidate symbol under each equalization output is calculated; and the raw bit log-likelihood ratio for each symbol time is calculated based on the conditional probability.

5. The method of claim 4, wherein, In the step of calculating the conditional probability of each PAM-4 candidate symbol under each equalization output, the conditional probability of each PAM-4 candidate symbol under each equalization output is calculated using the following formula: wherein, denotes a PAM-4 candidate symbol obtaining an equalized output denotes a conditional probability; denotes an equalized output corresponding to a denotes an equalized output corresponding to a denotes an equalized output corresponding to a denotes a noise variance estimated in an initialization training phase; The raw bit log-likelihood ratio at each symbol time is calculated using the following formula: in, Indicates the first candidate symbol in PAM-4 The second PAM-4 symbol subset with each original bit valued as 1; Indicates the first candidate symbol in PAM-4 The first subset of PAM-4 symbols with original bits set to 0. Indicates the first Within the iteration rounds, the _th The first symbol time corresponding to the th symbol time The log-likelihood ratio of the original bits.

6. The decision feedback equilibrium method based on coding constraints according to claim 1, characterized in that, In the step of generating the XOR domain soft information sequence based on the XOR relationship of two original bits in the PAM-4 Gray map, for any symbol time, based on the conditional probabilities that the two original bits are 0 and 1 respectively, a first XOR probability that the XOR result of the two original bits is 1 and a second XOR probability that the XOR result of the two original bits is 0 are determined; the ratio of the first XOR probability to the second XOR probability is used as the log-likelihood ratio of the XOR domain corresponding to that symbol time; the log-likelihood ratios of the XOR domain corresponding to each symbol time are combined sequentially to obtain the XOR domain soft information sequence.

7. The decision feedback equilibrium method based on coding constraints according to claim 1, characterized in that, In the step of inputting the XOR field soft information sequence into the XOR-extended Hamming code decoder to obtain the decoded XOR bit sequence and correction flag sequence of the current iteration, hard decision is made on the XOR field soft information sequence to obtain the initial XOR codeword. The initial XOR codeword is subjected to a comma-based check using the parity-check matrix of the XOR-extended Hamming code decoder. If the comma-based check result satisfies the check constraint, the initial XOR codeword is used as the decoded XOR bit sequence. If the comma-based check result does not satisfy the check constraint, multiple least reliable positions are determined based on the reliability ranking of the XOR field soft information sequence. A finite flip pattern is generated around the multiple least reliable positions, and a candidate XOR codeword set is generated based on the finite flip pattern. The decoded XOR bit sequence is selected from the candidate XOR codeword set according to the soft information metric. A correction flag sequence is generated based on the difference between the decoded XOR bit sequence and the initial XOR codeword.

8. The decision feedback equilibrium method based on coding constraints according to claim 1, characterized in that, In the steps of determining the adaptive update step size corresponding to each symbol time based on the correction flag sequence output from the previous iteration and the decision error sequence of the current iteration, and updating the forward filter coefficients and the feedback filter coefficients of the decision feedback equalizer based on the adaptive update step size corresponding to each symbol time, for the correction flag of the previous iteration representing the uncorrected symbol time, the basic update step size is determined as the adaptive update step size corresponding to that symbol time. For the correction flag of the previous iteration representing the corrected symbol time, the basic update step size is suppressed based on a step size suppression factor less than 1 to obtain the adaptive update step size corresponding to that symbol time; based on the adaptive update step size corresponding to that symbol time, the decision error, the received sampled input vector, and the historical decision symbol vector, the forward filter coefficients and the feedback filter coefficients are updated respectively.

9. The decision feedback equilibrium method based on coding constraints according to claim 8, characterized in that, In the step of updating the forward filter coefficients and the feedback filter coefficients based on the adaptive update step size, decision error, received sampled input vector, and historical decision symbol vector corresponding to the symbol time, the forward filter coefficients and the feedback filter coefficients are calculated using the following formulas: in, Indicates the first Processing within the iteration rounds Forward filter coefficients at each symbol time; Indicates the first The forward filter coefficients are updated at time +1 symbol; This indicates an adaptive update step size; Indicates the error in the decision; Indicates the received sampled input vector; Indicates the first Processing within the iteration rounds Feedback filter coefficients at each symbol time; Indicates the first The feedback filter coefficients are updated at time +1 symbol. This represents the vector of historical decision symbols.

10. A decision feedback equalization system based on coding constraints, characterized in that, The system includes a computer device, which includes a processor and a memory. The memory stores computer instructions, and the processor executes the computer instructions stored in the memory. When the computer instructions are executed by the processor, the system implements the steps of the method as described in any one of claims 1 to 9.