Polarization code serial cancellation list flipping decoding method and device

Abstract

The invention provides a polar code serial cancellation list flipping decoding method and device, and the method comprises the steps: obtaining a flipping set corresponding to a code tree, and the flipping set comprises at least one target layer; for each target layer, determining an overturning skipping result corresponding to the current target layer according to an overturning skipping rule; if the overturning skipping result is skipping, updating the current target layer into a next target layer of the current target layer in the overturning set, and executing the overturning skipping result corresponding to the current target layer determined according to the overturning skipping rule and the following steps; and if the flipping skipping result is no skipping, performing path flipping on the current target layer on the code tree and performing serial offset list decoding. According to the method, the target layer is evaluated through the path flipping skipping rule, and whether one flipping attempt is necessary is dynamically determined, so that the decoding attempt times and complexity of SCLF are reduced, and the time delay efficiency of the polarization code in a wireless communication system is improved.

Description

Technical Field

[0001] This invention relates to the field of polar code decoding technology in the field of communication technology, and particularly to a method and apparatus for polar code serial cancellation list flipping decoding. Background Technology

[0002] Successive Cancellation List Flip (SCLF) decoding improves error correction performance by continuously attempting to flip error-prone bits.

[0003] However, the SCLF decoding process involves decoding the initial Successive Cancellation List (SCL) and subsequent path flipping and re-decoding, resulting in high computational complexity. In low signal-to-noise ratio environments, this decoding method requires many more attempts to find the correct path, significantly increasing the decoding delay in the worst case and severely impacting decoding efficiency. Therefore, an effective solution is urgently needed to address these issues. Summary of the Invention

[0004] To address the aforementioned problems, this invention provides a polar code serial cancellation list flipping decoding method and apparatus.

[0005] This invention provides a polar code serial cancellation list flipping decoding method, comprising: Obtain the flipped set corresponding to the code tree, wherein the flipped set includes at least one target layer; For each target layer, the flip-skip result corresponding to the current target layer is determined according to the flip-skip rule; If the flip skip result is skipped, then the current target layer is updated to the next target layer of the current target layer in the flip set, and the steps of determining the flip skip result corresponding to the current target layer according to the flip skip rule and the subsequent steps are executed. If the flip-skip result is not to skip, then the current target layer on the code tree is flipped and serial cancellation list decoding is performed.

[0006] According to the present invention, a polar code serial cancellation list reversal decoding method is provided, wherein the path reversal and serial cancellation list decoding of the current target layer on the code tree are performed include: Perform path flipping on the current target layer in the code tree; Determine whether the target path in each layer of the code tree meets the abnormal path condition, wherein the target path is the path with the maximum path metric; If not, then perform serial cancellation list decoding on the code tree after path reversal.

[0007] According to the present invention, a polar code serial cancellation list flipping decoding method, after determining whether the target path in each layer of the code tree meets the abnormal path condition, further includes: If so, then stop serial cancellation list decoding of the code tree after path reversal and determine that the current target layer decoding has failed; Update the current target layer to the next target layer in the flip set, and execute the steps of determining the flip skip result corresponding to the current target layer according to the flip skip rule and the subsequent steps.

[0008] According to the present invention, a polar code serial cancellation list flipping decoding method is provided, wherein the abnormal path condition is: The path metric of the target path is less than the path threshold corresponding to the current target layer; The path threshold is determined based on the termination threshold and the minimum decoding path metric corresponding to the current target layer.

[0009] According to a polar code serial cancellation list flipping decoding method provided by the present invention, the path threshold is the difference between the initial decoding minimum path metric and a set value; The set value is the product of the termination threshold and the target standard deviation, or the set value is the product of the termination threshold and the path metric range value; The target standard deviation is the standard deviation of each path metric corresponding to the current target layer in the first round of serial cancellation list decoding.

[0010] According to the present invention, a polar code serial cancellation list flipping decoding method is provided, wherein the flipping skipping rule is as follows: If the number of decoding failures from the first specified layer to the second specified layer in the code tree reaches a threshold, then skip the path flipping of the current target layer. Wherein, the first designated layer number is the difference between the layer number of the current target layer and the set parameter; the second designated layer is the layer above the current target layer on the code tree.

[0011] According to the present invention, a polar code serial cancellation list flipping decoding method is provided, the method further comprising: The set parameters, the number of iterations threshold, and the termination threshold are determined using a deep Q-network or a fractional programming solver.

[0012] According to the present invention, a polar code serial cancellation list flipping decoding method is provided, wherein determining the set parameters, the number of attempts threshold, and the termination threshold through a deep Q-network includes: Based on the set parameters, the number of attempts threshold, and the termination threshold, a state is constructed, and based on the step size corresponding to the set parameters, the number of attempts threshold, and the termination threshold, an action set is determined. Based on the state and the set of actions, multiple rounds of decoding attempts are performed, and the reward for each round of decoding attempts is determined. Based on each of the aforementioned rewards, a target state is determined, which includes the final determined set parameters, the number of attempts threshold, and the termination threshold.

[0013] A polar code serial cancellation list flipping decoding method provided by the present invention further includes: Segmented cyclic redundancy check (CRC) codes are inserted at predefined split points. These segmented CRC codes are used for verification during the middle and end stages of the decoding process.

[0014] The present invention also provides a polar code serial cancellation list flipping decoding apparatus, comprising: The acquisition module is configured to acquire the flipped set corresponding to the code tree, the flipped set including at least one target layer; The first determining module is configured to determine the flip-skip result corresponding to the current target layer according to the flip-skip rule for each target layer. The update module is configured to update the current target layer to the next target layer in the flip set if the flip skip result is skipped, and execute the steps of determining the flip skip result corresponding to the current target layer according to the flip skip rule and the subsequent steps. The decoding module is configured to perform path reversal and serial cancellation list decoding on the current target layer of the code tree if the flip-skip result is not skipped.

[0015] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the polar code serial cancellation list flipping decoding method as described above.

[0016] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the polar code serial cancellation list flipping decoding method as described above.

[0017] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the polar code serial cancellation list flipping decoding method as described above.

[0018] The present invention provides a polar code serial cancellation list (SCLF) reversal decoding method and apparatus. The method involves obtaining a reversal set corresponding to a code tree, wherein the reversal set includes at least one target layer. For each target layer, a reversal skipping result is determined according to a reversal skipping rule. If the reversal skipping result is skipped, the current target layer is updated to the next target layer in the reversal set, and the steps of determining the reversal skipping result and subsequent steps according to the reversal skipping rule are executed. If the reversal skipping result is not skipped, a path reversal is performed on the current target layer in the code tree, and serial cancellation list decoding is performed. By using the path reversal skipping rule, the target layers in the reversal set are evaluated, and the necessity of a reversal attempt is dynamically determined, thereby reducing the number and complexity of SCLF decoding attempts and improving the latency efficiency of polar codes in wireless communication systems. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating the polar code serial cancellation list flipping decoding method provided by the present invention.

[0021] Figure 2 This is a probability density diagram of the items involved in this invention.

[0022] Figure 3 This is a schematic diagram of the structure of the polar code serial cancellation list flipping decoding device provided by the present invention.

[0023] Figure 4 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0025] First, a brief description of the relevant content involved in this invention will be given.

[0026] Polar codes have become the standard scheme for the control channel of 5G (5th Generation Mobile Communication Technology) because they have attracted widespread attention because when the code length approaches infinity, the successive cancellation (SC) decoding can approximate the channel capacity.

[0027] Successive Cancellation List (SCL) decoding is the baseline decoding algorithm for polar codes. To enhance error correction performance, researchers have proposed a Cyclic Redundancy Check-Aided SCL (CA-SCL) decoding algorithm. Simultaneously, another type of SC algorithm—Successive Cancellation Flip (SCF)—has also been proposed. When SC decoding fails the Cyclic Redundancy Check (CRC) check, it improves performance by identifying and flipping error-prone bits. Its computational complexity is comparable to SC decoding at high signal-to-noise ratios.

[0028] Inspired by the SCF decoding algorithm, the SCLF decoding technique was proposed. Subsequently, a path-reversal-based SCLF decoding algorithm and its performance-enhanced version emerged, which can more accurately identify and reverse the paths of error-prone layers and restart SCL decoding.

[0029] To reduce the complexity and improve the efficiency of SCLF decoding, existing technologies have introduced an early termination mechanism based on the Relative Partial Path Metric (RPPM), which significantly reduces decoding latency. However, the RPPM-based early termination mechanism has certain limitations in performance optimization.

[0030] In summary, existing polar code SCLF decoding has problems in terms of error correction performance, computational complexity, and decoding delay, as detailed below: SCLF decoding suffers from high complexity and long decoding delays: SCLF decoding improves error correction performance by continuously attempting to flip error-prone bits, but this process involves initial SCL decoding and multiple subsequent path flips and re-decoding, resulting in high computational complexity. In low signal-to-noise ratio environments, this decoding method requires even more attempts to find the correct path, significantly increasing the decoding delay in the worst case and severely impacting decoding efficiency.

[0031] To address the aforementioned problems, this invention provides a polar code serial cancellation list flipping decoding method and apparatus.

[0032] The following is combined Figures 1-4The present invention describes a polar code serial cancellation list flipping decoding method and apparatus.

[0033] Figure 1 This is a flowchart illustrating the polar code serial cancellation list flipping decoding method provided by the present invention, as shown below. Figure 1 As shown, the method includes the following: Step 101: Obtain the flipped set corresponding to the code tree, wherein the flipped set includes at least one target layer; Step 102: For each target layer, determine the flip-skip result corresponding to the current target layer according to the flip-skip rule; Step 103: If the flip skip result is skipped, then update the current target layer to the next target layer of the current target layer in the flip set, and execute the steps of determining the flip skip result corresponding to the current target layer according to the flip skip rule and the subsequent steps. Step 104: If the flip skip result is no skip, then perform path flipping on the current target layer on the code tree and perform serial cancellation list decoding.

[0034] Specifically, the polar code serial cancellation list reversal decoding is the polar code SCLF decoding.

[0035] Specifically, in the case of failure of the initial SCL decoding in SCLF decoding, a code tree can be constructed based on the dependency relationship between the decisions of each bit in SCL decoding of the polar code. The polar code SCL decoding code tree is actually a full binary tree. Therefore, the decoding process is to find a suitable path on the full binary tree.

[0036] In practical applications, the flip set on the polar code tree can be obtained. The flip set contains at least one layer on the code tree that may cause decoding failure, i.e., the target layer.

[0037] Specifically, the process of obtaining the flipped set can be as follows: obtain the reliability of each layer on the code tree, add each layer to the flipped set in order of reliability from low to high, or add layers with reliability below the reliability threshold to the flipped set in order of reliability from low to high.

[0038] Furthermore, it is necessary to perform polar code serial cancellation list flip decoding on each target layer one by one, according to the order of each target layer in the flip set.

[0039] Specifically, the process of polar code serial cancellation list reversal decoding for each target layer is as follows: determine whether the current target layer meets the reversal skipping condition corresponding to the reversal skipping rule; if it meets the condition, i.e. the reversal skipping result is skipped, then the current target layer is not reversed and decoded, and then the next target layer in the reversal set is reversed and decoded; if it does not meet the condition, i.e. the reversal skipping result is not skipped, then the current target layer on the code tree is reversed and decoded using the serial cancellation list.

[0040] It should be noted that before performing a path flip on the current target layer in the code tree, the path that was last flipped in the code tree can be reset, i.e., a reverse flip can be performed.

[0041] Furthermore, for skipped target layers, since no flipping attempt was performed, it is uncertain whether the flipping attempt (decoding) was successful or failed. Therefore, the result of the flipping attempt (decoding) does not need to be recorded.

[0042] The present invention provides a polar code serial cancellation list (SCLF) reversal decoding method. This method obtains a reversal set corresponding to a code tree, where the reversal set includes at least one target layer. For each target layer, a reversal skipping result is determined according to a reversal skipping rule. If the reversal skipping result is skipped, the current target layer is updated to the next target layer in the reversal set, and the steps of determining the reversal skipping result according to the reversal skipping rule and subsequent steps are executed. If the reversal skipping result is not skipped, a path reversal is performed on the current target layer in the code tree, and serial cancellation list decoding is performed. Based on the path reversal skipping rule, the target layers in the reversal set are evaluated, and the necessity of a reversal attempt is dynamically determined, thereby reducing the number and complexity of SCLF decoding attempts and improving the latency efficiency of polar codes in wireless communication systems.

[0043] In one or more optional embodiments of the present invention, the step of performing path flipping and serial cancellation list decoding on the current target layer of the code tree includes: Perform path flipping on the current target layer in the code tree; Determine whether the target path in each layer of the code tree meets the abnormal path condition, wherein the target path is the path with the maximum path metric; If not, then perform serial cancellation list decoding on the code tree after path reversal.

[0044] In practical applications, the SCL decoder can be restarted by SCL Re-Enable (SCL-RE) and adjusted in conjunction with the flip set, that is, the path of the current target layer on the code tree is flipped.

[0045] After flipping, it is determined whether the target path with the maximum path metric in each layer of the code tree meets the abnormal path criteria, i.e., whether it meets the abnormal path conditions. If it does not meet the criteria, it means that the target path in the current target layer is not an abnormal path. Therefore, the decoder needs to perform serial cancellation list decoding on the code tree after the path is flipped to complete the current decoding.

[0046] For example, decoding can be performed while determining the path after path inversion: first, the path of the current target layer in the code tree is inverted, and then starting from the first layer of the code tree, it is determined whether the target path of the first layer meets the abnormal path condition. If it does, the serial cancellation list decoding is stopped; if it does not, the serial cancellation list decoding is performed on the first layer. Then, it is determined whether the target path of the second layer meets the abnormal path condition, and so on.

[0047] For example, after path flipping, a holistic assessment can be performed before deciding whether to decode: first, the path at the current target layer of the code tree is flipped; then, it is determined whether the target path in each layer of the code tree meets the abnormal path condition. If a target path meets the abnormal path condition, serial cancellation list decoding is stopped; otherwise, serial cancellation list decoding is performed on the code tree. In this way, decoding of the code tree can be avoided even when a target path meets the abnormal path condition, thereby reducing the amount of data processing.

[0048] In this embodiment of the invention, abnormal paths can be avoided by using abnormal path conditions, thereby performing serial cancellation list decoding on non-existent abnormal path code trees, reducing the number of decoding operations and improving the decoding success rate.

[0049] In one or more optional embodiments of the present invention, after determining whether the target path in each layer of the code tree meets the abnormal path condition, the method further includes: If so, then stop serial cancellation list decoding of the code tree after path reversal and determine that the current target layer decoding has failed; Update the current target layer to the next target layer in the flip set, and execute the steps of determining the flip skip result corresponding to the current target layer according to the flip skip rule and the subsequent steps.

[0050] In practical applications, after reversing the path of the current target layer on the code tree, it is determined whether the target path with the maximum path metric in each layer of the code tree meets the abnormal path criteria, i.e., whether it meets the abnormal path conditions.

[0051] If a target path matching the abnormal path label exists, it indicates that the target path is an abnormal path. Therefore, there is no need to perform serial cancellation list decoding on the code tree after the path is flipped; that is, the path flipping is skipped. In this case, it indicates that the decoding of the current target layer has failed, i.e., the flipping attempt has failed.

[0052] Then, polar code serial cancellation list flip decoding is performed on the next target layer of the current target layer in the flip set.

[0053] In this embodiment of the invention, abnormal paths can be filtered out by abnormal path conditions, thereby performing path reversal and skipping, reducing the number of decoding operations and the amount of data processing.

[0054] In one or more optional embodiments of the present invention, the abnormal path condition is: The path metric of the target path is less than the path threshold corresponding to the current target layer; The path threshold is determined based on the termination threshold and the minimum decoding path metric corresponding to the current target layer.

[0055] Specifically, the path threshold is the difference between the initial decoding minimum path metric and a set value.

[0056] Specifically, the set value can be the product of the termination threshold and the target standard deviation, or the set value can be the product of the termination threshold and the path metric range value; wherein, the target standard deviation is the standard deviation of each path metric corresponding to the current target layer in the first round of serial cancellation list decoding.

[0057] Specifically, the initial decoding minimum path metric refers to the minimum value among all path metrics of the current target layer during initial SCL decoding.

[0058] In SCLF decoding, continuously restarting the SCL decoder after each path flip does not fully utilize the information provided by the initial SCL. Therefore, this invention utilizes the intermediate information of the initial SCL decoder to guide the behavior of the reset SCL decoder, thereby improving the efficiency of SCLF decoding: it uses the path metric (PM) of the initial SCL decoder to assist the reset SCL decoder.

[0059] The path metrics of all paths in each layer of the SCL decoder are assumed to follow a Gaussian distribution. Therefore, the Gaussian distribution... Inspired by the principle, the path metric deviates from the mean by more than Paths that exceed one standard deviation are considered outliers.

[0060] Based on the above settings, for the restarted SCL decoding process, it can be checked whether the path with the maximum path metric in each layer meets the abnormal path condition. If it does, the decoding is terminated early, meaning the flip attempt (decoding) of the current target layer fails. Afterwards, polar code serial cancellation list flip decoding begins for the next target layer in the flip set.

[0061] Specifically, for the first In the second bit-flipping SCL attempt, the... Maximum path metric of layer Decoding will terminate prematurely if the following abnormal path conditions are met: in, Indicates the first Next bit flip; Represents the first in the code tree layer; It is the first round of SCL decoding. Standard deviation of layer path metric; 0 indicates initial SCL decoding; It is the termination threshold, which is a configurable early termination threshold parameter; This represents the minimum path metric of the l-th layer in the initial SCL decoding; This represents the path metric for all 2L paths in the l-th layer of the initial SCL decoding, where L is the scalable path in the l-th layer.

[0062] Since the metric path is generally negative, The smaller the threshold, the easier it is to terminate decoding prematurely. At the same time, it is crucial to set a reasonable threshold without affecting the decoder's error correction performance.

[0063] In addition, calculate the standard deviation. This would impose a significant complexity burden on the SCLF decoder, so the range value of the path metric can be used instead of the variance. Since calculating the range value for each layer is straightforward, as the maximum and minimum path metrics are readily available, the abnormal path condition can also be expressed as: in, Indicates the first Next bit flip; Represents the first in the code tree Layer; 0 indicates initial SCL decoding; It is the termination threshold, and L is the scalable path of the l-th layer. This represents the maximum path metric of layer 1 in the t-th round of SCL decoding. This represents the minimum path metric for layer 1 in the initial SCL decoding. This represents the path metric of the (2L-1)th path in the l-th layer of the initial SCL decoding. This represents the path metric for the 0th path in the l-th layer of the initial SCL decoding.

[0064] For example, the SCL-RE (SCL termination) algorithm can be as follows: Input: Received signal Flipped set Minimum value of path metric in the first round Threshold .

[0065] Output: Estimated bits Early termination sign .

[0066] 1. ; / / Early termination flag 2. For arrive Perform the following operations: - Update the metric for each path. / / Calculate the metric for each path - Forked into Path and Sort; / / L is the search width - If the abnormal path condition is true or no path passes the segmented CRC check, then: - Returns to the empty set; - Termination condition judgment - when or At that time, retain path metrics that are smaller / larger Path; - ; 3. End the loop 4. Return , .

[0067] Among them, path metric depends on signal N is The number of target layers. A value of 0 indicates no stopping, while a value of 1 indicates stopping.

[0068] In this embodiment of the invention, redundant path flipping attempts are dynamically skipped by analyzing historical failure patterns, and SCL is terminated early by using a path metric threshold inspired by Gaussian distribution, which effectively reduces the number of decoding attempts.

[0069] Furthermore, by employing the proposed termination threshold-based outlier condition, the decoder does not require a total ordering network. Instead, a low-complexity extractor can be used to obtain range values. This can reduce latency.

[0070] In one or more optional embodiments of the present invention, the flip-skip rule is as follows: If the number of decoding failures from the first specified layer to the second specified layer in the code tree reaches a threshold, then skip the path flipping of the current target layer. Wherein, the first designated layer number is the difference between the layer number of the current target layer and the set parameter; the second designated layer is the layer above the current target layer on the code tree.

[0071] Specifically, SCLF decoding involves path reversal and restarting SCL decoding. Therefore, restarting SCL decoding can refine the search space for a specific layer, thereby increasing the probability of preserving the correct path.

[0072] Based on this, the present invention designs a flip skip rule: for a given layer If from the first layer to the first At least one layer If the second flip attempt (decoding) fails, then the layer will be... Mark as unreliable and skip the current layer. The attempt to flip. Among them, To set parameters, The threshold is the number of occurrences.

[0073] Specifically, the goal of the flip skip rule can be: in, It is the flip of the first The layer represents an event that enables successful decoding; From the first layer to the first At least one layer The event that the flipping attempt fails; P is the probability, representing... In the event of this, The probability of occurrence.

[0074] Then, according to Bayes' theorem, the above equation can be transformed into: in, It is a definite item, not subject to adjustment criteria. The impact of this. To achieve the goal of the flip skip rule, it is necessary to... Smaller while making Larger.

[0075] Figure 2 This is a probability density diagram of the items involved in this invention. See also... Figure 2 As a simple example, Figure 2 The middle gives and The probability density. From Figure 2 It can be seen from this that and In each of its independent variables and The above are all monotonous. However, directly using mathematical methods to analyze simultaneously makes Maximize and Minimized Combinatorial processing is complex. On the other hand, considering the diminishing nature of these functions, an agent-based stepwise adjustment method can be employed. Therefore, agent-based reinforcement learning techniques can be used for fine-tuning. and .

[0076] For example, the FLIPSKIP algorithm (path reversal skip rule) can be as follows: Algorithm 1 FLIPSKIP: Path Reversal Skip Criterion Input; Flipped set Current flip round ,parameter Output: Skip flag 1. Initialize the counter If the value is 0, skip the flag. It is 0.

[0077] 2. Calculate within the interval Number of path flipping attempts within the range: - For from 0 to Each Perform the following operations: - if Then the counter Add 1.

[0078] 3. If the counter ,but .

[0079] 4. Return to skip flag .

[0080] in, It is the i-th layer in S that needs to be flipped. A value of 1 indicates skipping, and a value of 0 indicates not skipping.

[0081] In this embodiment of the invention, by summarizing the path flip skipping rules, evaluating the flip set S and considering the historical performance of previous flip rounds, the algorithm dynamically determines whether a flip attempt is necessary. Therefore, it can reduce the number of decoding attempts and complexity of the SCLF decoder.

[0082] In recent years, deep learning (DL) technology has been applied to SCLF decoding, which has effectively reduced computational complexity and improved error correction performance. However, the embedding of neural networks has increased the difficulty of practical application.

[0083] While integrating deep learning (DL) technology into SCLF decoding can improve error correction performance and reduce computational complexity to some extent, it presents numerous challenges in practical applications due to the need to embed neural networks into the decoder. For example, neural networks require significant computational resources and place high demands on hardware; furthermore, their complex model structure increases the difficulty of decoder design and implementation, hindering widespread adoption.

[0084] Furthermore, traditional optimization methods face challenges in parameter tuning: when optimizing decoding algorithms, the threshold parameter significantly impacts decoding performance. However, due to the vast parameter search space, traditional mathematical methods struggle to finely adjust the threshold parameters at each layer to achieve a balance between minimizing decoding attempts and optimizing error correction performance.

[0085] To address the above problems, in one or more optional embodiments of the present invention, the method further includes: determining the set parameters, the number threshold, and the termination threshold using a deep Q-network or a fractional programming solver.

[0086] In practical applications, the parameters, number of attempts, and termination threshold can be adjusted to better balance complexity and error rate. However, due to the enormous parameter search space, fine-tuning the threshold parameters of each layer to obtain the minimum number of decoding attempts is very complex. Therefore, Deep Q-Networks (DQNs) can be used as an optimization tool, with the agent learning the thresholds.

[0087] In addition, DQN can be replaced with any fractional programming solver to determine the set parameters, number of iterations, and termination threshold, such as genetic algorithms or differential evolution algorithms.

[0088] In this embodiment, the traditional deep learning (DL) technology is abandoned, and a deep Q-network or fractional programming solver is used to adjust the threshold parameters. There is no need to embed the neural network into the decoder, which reduces the requirements for hardware devices and the difficulty of decoder design and implementation in the model structure. This facilitates widespread application. Moreover, without a huge parameter search space, each threshold parameter can be finely adjusted to obtain the minimum number of decoding attempts and the best balance between error correction performance.

[0089] In one or more optional embodiments of the present invention, determining the set parameters, the number of iterations threshold, and the termination threshold using a deep Q-network includes: Based on the set parameters, the number of attempts threshold, and the termination threshold, a state is constructed, and based on the step size corresponding to the set parameters, the number of attempts threshold, and the termination threshold, an action set is determined. Based on the state and the set of actions, multiple rounds of decoding attempts are performed, and the reward for each round of decoding attempts is determined. Based on each of the aforementioned rewards, a target state is determined, which includes the final determined set parameters, the number of attempts threshold, and the termination threshold.

[0090] Reinforcement learning is often used in scenarios that require interaction with the environment. Given a state of the environment, the program selects a corresponding action based on a certain policy. After executing this action, the environment changes, and the state is transformed into a new state S'. After each action is executed, the program receives a reward. The program then adjusts its policy based on the size of the reward to maximize the sum of rewards obtained when all steps are completed and the state reaches the terminal state.

[0091] Specifically, the state: the state consists of a parameter containing three threshold parameters. The vector representation of , where To set parameters, The threshold number of times, The termination threshold; Action: The action space consists of possible adjustments to the threshold parameter.

[0092] For each threshold parameter, the agent can choose to increase or decrease a threshold parameter by a constant step size.

[0093] Specifically, action sets are Optionally, the constant step size is: , , .

[0094] Action sets enable agents to find near-optimal values. This combination reduces the number of decoding attempts while maintaining the Frame Error Rate (FER).

[0095] The reward R comes from the average number of decoding attempts and error correction performance, as detailed below: in, and These represent the average number of decoding attempts for the current state and the previous state, respectively.

[0096] Specifically, an agent receives a positive reward for reducing complexity without reducing the error rate, and a negative reward for reducing error performance.

[0097] It should be noted that the training loss function for the combination of DQN and the agent can be the mean squared error (MSE) between the predicted value and the target value.

[0098] In this embodiment of the invention, by combining DQN and an agent, DQN can effectively achieve threshold learning, thereby enabling the SCLF decoder to reduce latency while maintaining performance.

[0099] In one or more optional embodiments of the present invention, the method further includes: Segmented cyclic redundancy check (CRC) codes are inserted at predefined split points. These segmented CRC codes are used for verification during the middle and end stages of the decoding process.

[0100] Specifically, starting with SCL decoding, segmented CRC can be used to reduce SCLF decoding latency. Segmented CRC codes are inserted at predefined split points and can be checked during and at the end of the decoding process. If an estimated bit fails the CRC check, the decoder generates a flip set S. Unlike CRC-assisted SCLF, if no path passes the first segmented CRC check during the initial SCL decoding, it indicates that path flipping is needed before the split point. In this case, all target layers in the flip set S must be limited to indices smaller than the split point, and the core optimization occurs in a loop of at most T iterations, i.e., the flip set S contains T target layers. One round (iteration) is a serial cancellation list flip decoding of one target layer.

[0101] The polar code serial cancellation list flipping decoding method provided by the present invention will be further described below.

[0102] SCL decoding is a tree search-based algorithm that includes steps such as path metric calculation, path sorting, and path expansion. SCLF decoding improves error correction performance by sequentially performing path reversals and restarting the SCL decoding process. However, it suffers from high complexity and long decoding latency in the worst case, especially under low signal-to-noise ratio conditions. To reduce the decoding latency of SCLF, this invention designs a stopping criterion for SCL decoding and path reversals to reduce the number of decoding attempts.

[0103] In each iteration, the FLIPSKIP algorithm determines whether to skip the current flip attempt based on the path flip skip criterion. If it indicates that a flip should not be performed (i.e., ...), the algorithm skips the current flip attempt. If the condition is met (i.e., the path reversal attempt is successful), the FLIPSKIP algorithm proceeds to the next iteration without requiring additional path reversal attempts. Otherwise, the SCL decoder is restarted using the SCL-RE algorithm, and adjustments are made in conjunction with the reversal set. When the condition is met (i.e., ... Decoding can be restarted prematurely (decoding failure); when the conditions are not met (i.e. Once the current decoding is complete, decoding ends (decoding successful) if the estimated value passes the CRC checkpoint; otherwise, decoding continues for the next target layer.

[0104] Specifically, the ThresholdStudy SCLF (TS–SCLF) algorithm based on threshold learning can be described as follows: Input: Size of the flipped set Receive signal Output: Estimated bits 1. Initialize the flipped set , logo .

[0105] 2. Perform SCL decoding to obtain the estimated bits. logo and the minimum value of the first round of path metric .

[0106] 3. If estimating bits If the CRC check passes, the estimated bits are returned. .

[0107] 4. Generate a flipped set .

[0108] 5. Perform the most Next iteration: - Call the FLIPSKIP function (Algorithm 1), passing in the flipped set. Current iteration round and threshold parameters Get the skip flag .

[0109] - If skip flag =0 (i.e., do not skip this flip attempt): - Call the SCL-RE function (Algorithm 2) and pass in the received signal. Current flip set elements and the minimum value of the first round of path metric Update estimated bits and early termination sign .

[0110] - If the estimated bits If the CRC check passes, the loop will exit.

[0111] 6. Return the estimated bits .

[0112] The polar code serial cancellation list (SCL) reversal decoding method provided by this invention employs a dual-threshold strategy. By analyzing historical failure patterns, redundant path reversal attempts are dynamically skipped. A path metric threshold inspired by Gaussian distribution is used to achieve early termination of the SCL. Segmented CRC is also incorporated into predefined segmentation points for intermediate verification, effectively reducing decoding latency. In terms of threshold optimization, a DQN framework is introduced, which can dynamically adjust the threshold parameters according to the decoding situation, achieving a good balance between complexity and error performance.

[0113] Experimental results demonstrate significant advantages. For (1024, 512) polar codes, under low signal-to-noise ratio (SNR) conditions, compared with existing advanced methods, the polar code serial cancellation list reversal decoding method provided in this invention can reduce the average number of decoding attempts by up to 55.3%, and in the worst case, by up to 41.5%, while maintaining essentially unchanged error correction performance. This represents a substantial improvement in decoding efficiency and demonstrates its potential for practical communication system applications. In other words, the polar code serial cancellation list reversal decoding method provided in this invention can reduce the complexity and decoding delay of SCLF decoding, especially under low SNR conditions, by reducing the average number of decoding attempts and the worst-case decoding delay while ensuring error correction performance, thereby improving the decoding efficiency of polar codes.

[0114] The polar code serial cancellation list reversal decoding apparatus provided by the present invention is described below. The polar code serial cancellation list reversal decoding apparatus described below and the polar code serial cancellation list reversal decoding method described above can be referred to in correspondence.

[0115] Figure 3 This is a schematic diagram of the structure of the polar code serial cancellation list flipping decoding device provided by the present invention, as shown below. Figure 3 As shown, the device includes the following: The acquisition module 301 is configured to acquire the flip set corresponding to the code tree, wherein the flip set includes at least one target layer; The first determining module 302 is configured to determine the flip-skip result corresponding to the current target layer according to the flip-skip rule for each target layer. The update module 303 is configured to update the current target layer to the next target layer of the current target layer in the flip set if the flip skip result is skipped, and execute the steps of determining the flip skip result corresponding to the current target layer according to the flip skip rule and the subsequent steps. The decoding module 304 is configured to perform path reversal and serial cancellation list decoding on the current target layer of the code tree if the flip skip result is not skipped.

[0116] The polar code serial cancellation list (SCLF) reversal decoding apparatus provided by this invention obtains a reversal set corresponding to a code tree, the reversal set including at least one target layer; for each target layer, a reversal skipping result corresponding to the current target layer is determined according to a reversal skipping rule; if the reversal skipping result is skipped, the current target layer is updated to the next target layer in the reversal set, and the steps of determining the reversal skipping result corresponding to the current target layer according to the reversal skipping rule and subsequent steps are executed; if the reversal skipping result is not skipped, the current target layer on the code tree is path-reversed and serial cancellation list decoding is performed. By using the path-reversal skipping rule, the target layers in the reversal set are evaluated, and the necessity of a reversal attempt is dynamically determined, thereby reducing the number and complexity of SCLF decoding attempts and improving the latency efficiency of polar codes in wireless communication systems.

[0117] In one or more optional embodiments of the present invention, the decoding module 304 is specifically configured as follows: Perform path flipping on the current target layer in the code tree; Determine whether the target path in each layer of the code tree meets the abnormal path condition, wherein the target path is the path with the maximum path metric; If not, then perform serial cancellation list decoding on the code tree after path reversal.

[0118] In one or more optional embodiments of the present invention, the decoding module 304 is further configured to: If so, then stop serial cancellation list decoding of the code tree after path reversal and determine that the current target layer decoding has failed; Update the current target layer to the next target layer in the flip set, and execute the steps of determining the flip skip result corresponding to the current target layer according to the flip skip rule and the subsequent steps.

[0119] In one or more optional embodiments of the present invention, the abnormal path condition is: The path metric of the target path is less than the path threshold corresponding to the current target layer; The path threshold is determined based on the termination threshold and the minimum decoding path metric corresponding to the current target layer.

[0120] In one or more optional embodiments of the present invention, the path threshold is the difference between the initial decoding minimum path metric and a set value; The set value is the product of the termination threshold and the target standard deviation, or the set value is the product of the termination threshold and the path metric range value; The target standard deviation is the standard deviation of each path metric corresponding to the current target layer in the first round of serial cancellation list decoding.

[0121] In one or more optional embodiments of the present invention, the flip-skip rule is as follows: If the number of decoding failures from the first specified layer to the second specified layer in the code tree reaches a threshold, then skip the path flipping of the current target layer. Wherein, the first designated layer number is the difference between the layer number of the current target layer and the set parameter; the second designated layer is the layer above the current target layer on the code tree.

[0122] In one or more optional embodiments of the present invention, the apparatus further includes a second determining module configured to: The set parameters, the number of iterations threshold, and the termination threshold are determined using a deep Q-network or a fractional programming solver.

[0123] In one or more optional embodiments of the present invention, the second determining module is specifically configured as follows: Based on the set parameters, the number of attempts threshold, and the termination threshold, a state is constructed, and based on the step size corresponding to the set parameters, the number of attempts threshold, and the termination threshold, an action set is determined. Based on the state and the set of actions, multiple rounds of decoding attempts are performed, and the reward for each round of decoding attempts is determined. Based on each of the aforementioned rewards, a target state is determined, which includes the final determined setting parameters, the number of attempts threshold, and the termination threshold.

[0124] In one or more optional embodiments of the invention, the device further includes an insertion module configured to: Segmented cyclic redundancy check (CRC) codes are inserted at predefined split points. These segmented CRC codes are used for verification during the middle and end stages of the decoding process.

[0125] Figure 4 This is a schematic diagram of the structure of the electronic device provided by the present invention, such as... Figure 4As shown, the electronic device may include: a processor 410, a communications interface 420, a memory 430, and a communication bus 440, wherein the processor 410, the communications interface 420, and the memory 430 communicate with each other via the communication bus 440. The processor 410 can call logical instructions in the memory 430 to execute a polar code serial cancellation list reversal decoding method, which includes: Obtain the flipped set corresponding to the code tree, wherein the flipped set includes at least one target layer; For each target layer, the flip-skip result corresponding to the current target layer is determined according to the flip-skip rule; If the flip skip result is skipped, then the current target layer is updated to the next target layer of the current target layer in the flip set, and the steps of determining the flip skip result corresponding to the current target layer according to the flip skip rule and the subsequent steps are executed. If the flip-skip result is not to skip, then the current target layer on the code tree is flipped and serial cancellation list decoding is performed.

[0126] Furthermore, the logical instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0127] On the other hand, the present invention also provides a computer program product, the computer program product comprising a computer program that can be stored on a non-transitory computer-readable storage medium, wherein when the computer program is executed by a processor, the computer is capable of executing the polar code serial cancellation list reversal decoding method provided by the above methods, the method comprising: Obtain the flipped set corresponding to the code tree, wherein the flipped set includes at least one target layer; For each target layer, the flip-skip result corresponding to the current target layer is determined according to the flip-skip rule; If the flip skip result is skipped, then the current target layer is updated to the next target layer of the current target layer in the flip set, and the steps of determining the flip skip result corresponding to the current target layer according to the flip skip rule and the subsequent steps are executed. If the flip-skip result is not to skip, then the current target layer on the code tree is flipped and serial cancellation list decoding is performed.

[0128] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the polar code serial cancellation list reversal decoding method provided by the methods described above, the method comprising: Obtain the flipped set corresponding to the code tree, wherein the flipped set includes at least one target layer; For each target layer, the flip-skip result corresponding to the current target layer is determined according to the flip-skip rule; If the flip skip result is skipped, then the current target layer is updated to the next target layer of the current target layer in the flip set, and the steps of determining the flip skip result corresponding to the current target layer according to the flip skip rule and the subsequent steps are executed. If the flip-skip result is not to skip, then the current target layer on the code tree is flipped and serial cancellation list decoding is performed.

[0129] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0130] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0131] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A polar code serial cancellation list reversal decoding method, characterized in that, include: Obtain the flipped set corresponding to the code tree, wherein the flipped set includes at least one target layer; For each target layer, the flip-skip result corresponding to the current target layer is determined according to the flip-skip rule; If the flip skip result is skipped, then the current target layer is updated to the next target layer of the current target layer in the flip set, and the steps of determining the flip skip result corresponding to the current target layer according to the flip skip rule and the subsequent steps are executed. If the flip-skip result is not to skip, then the current target layer on the code tree is flipped and serial cancellation list decoding is performed.

2. The polar code serial cancellation list reversal decoding method according to claim 1, characterized in that, The step of performing path reversal and serial cancellation list decoding on the current target layer of the code tree includes: Perform path flipping on the current target layer in the code tree; Determine whether the target path in each layer of the code tree meets the abnormal path condition, wherein the target path is the path with the maximum path metric; If not, then perform serial cancellation list decoding on the code tree after path reversal.

3. The polar code serial cancellation list reversal decoding method according to claim 2, characterized in that, After determining whether the target path in each layer of the code tree meets the abnormal path condition, the method further includes: If so, then stop serial cancellation list decoding of the code tree after path reversal and determine that the current target layer decoding has failed; Update the current target layer to the next target layer in the flip set, and execute the steps of determining the flip skip result corresponding to the current target layer according to the flip skip rule and the subsequent steps.

4. The polar code serial cancellation list reversal decoding method according to claim 2, characterized in that, The abnormal path conditions are as follows: The path metric of the target path is less than the path threshold corresponding to the current target layer; The path threshold is determined based on the termination threshold and the minimum decoding path metric corresponding to the current target layer.

5. The polar code serial cancellation list reversal decoding method according to claim 4, characterized in that, The path threshold is the difference between the initial decoding minimum path metric and a set value; The set value is the product of the termination threshold and the target standard deviation, or the set value is the product of the termination threshold and the path metric range value; The target standard deviation is the standard deviation of each path metric corresponding to the current target layer in the first round of serial cancellation list decoding.

6. The polar code serial cancellation list reversal decoding method according to claim 1, characterized in that, The flip skip rule is as follows: If the number of decoding failures from the first specified layer to the second specified layer in the code tree reaches a threshold, then skip the path flipping of the current target layer. Wherein, the first designated layer number is the difference between the layer number of the current target layer and the set parameter; the second designated layer is the layer above the current target layer on the code tree.

7. The polar code serial cancellation list reversal decoding method according to claim 6, characterized in that, The method further includes: The set parameters, the number of iterations threshold, and the termination threshold are determined using a deep Q-network or a fractional programming solver.

8. The polar code serial cancellation list reversal decoding method according to claim 7, characterized in that, The step of determining the set parameters, the number of attempts threshold, and the termination threshold using a deep Q-network includes: Based on the set parameters, the number of attempts threshold, and the termination threshold, a state is constructed, and based on the step size corresponding to the set parameters, the number of attempts threshold, and the termination threshold, an action set is determined. Based on the state and the set of actions, multiple rounds of decoding attempts are performed, and the reward for each round of decoding attempts is determined. Based on each of the aforementioned rewards, a target state is determined, which includes the final determined set parameters, the number of attempts threshold, and the termination threshold.

9. The polar code serial cancellation list reversal decoding method according to any one of claims 1-8, characterized in that, The method further includes: Segmented cyclic redundancy check (CRC) codes are inserted at predefined split points. These segmented CRC codes are used for verification during the middle and end stages of the decoding process.

10. A polar code serial cancellation list reversal decoding device, characterized in that, include: The acquisition module is configured to acquire the flipped set corresponding to the code tree, the flipped set including at least one target layer; The first determining module is configured to determine the flip-skip result corresponding to the current target layer according to the flip-skip rule for each target layer. The update module is configured to update the current target layer to the next target layer in the flip set if the flip skip result is skipped, and execute the steps of determining the flip skip result corresponding to the current target layer according to the flip skip rule and the subsequent steps. The decoding module is configured to perform path reversal and serial cancellation list decoding on the current target layer of the code tree if the flip-skip result is not skipped.

11. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the polar code serial cancellation list flipping decoding method as described in any one of claims 1 to 9.

12. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the polar code serial cancellation list flipping decoding method as described in any one of claims 1 to 9.

13. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the polar code serial cancellation list flipping decoding method as described in any one of claims 1 to 9.

Images