LDPC error correction method and system based on adaptive adjustment

Abstract

The invention relates to an LDPC error correction method and system based on adaptive adjustment, and belongs to the technical field of data processing.The method comprises the steps that performance indexes reflecting the error rate of current data are obtained, and the performance indexes at least comprise the bit error rate; the performance index is compared with a preset threshold value, a target code word scheme is selected from a plurality of pre-stored LDPC code word schemes according to a comparison result, and the plurality of LDPC code word schemes have different error correction capabilities; and switching the current LDPC codeword scheme into the target codeword scheme, and executing the target codeword scheme to perform error correction on the current data, thereby reducing the error rate of the current data. The LDPC error correction method and the LDPC error correction device have the effect of considering both the resource efficiency under the normal condition and the high-reliability LDPC error correction under the abnormal condition.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to an adaptive adjustment-based LDPC error correction method and system. Background Technology

[0002] In the field of data storage and transmission, especially in applications with extremely high requirements for data reliability such as aerospace, deep space exploration, and high-performance computing, error correction coding (ECC) technology is crucial for ensuring data integrity. Low-density parity-check codes (LDPC codes), as an advanced error correction coding system with performance approaching the Shannon limit, are widely used to address random and burst errors in channels.

[0003] Traditional LDPC code applications typically involve determining a fixed code rate (e.g., 1 / 2, 2 / 3) and corresponding parity-check matrix during the design phase, based on the expected maximum error rate or worst-case environmental conditions. The system uses this fixed configuration for encoding and decoding throughout its entire lifecycle. However, to cope with rare but extremely severe situations (such as strong solar particle events or performance degradation at the end of the storage medium's lifespan), a powerful LDPC code with a low code rate (high redundancy) must be employed. This means that for most of the normal operating time, valuable storage space and bandwidth are constantly occupied, and computational resources are used for unnecessary strong error correction operations, resulting in significant resource waste. Therefore, improvements are needed. Summary of the Invention

[0004] In order to balance resource efficiency under normal conditions with high reliability of LDPC error correction under abnormal conditions, this application provides an adaptive adjustment-based LDPC error correction method and system.

[0005] Firstly, this application provides an adaptive LDPC error correction method, employing the following technical solution: Obtain a performance metric that reflects the error rate of the current data, wherein the performance metric includes at least the bit error rate; The performance index is compared with a preset threshold, and a target codeword scheme is selected from a number of pre-stored LDPC codeword schemes based on the comparison results. The number of LDPC codeword schemes have different error correction capabilities. The current LDPC codeword scheme is switched to the target codeword scheme, and the target codeword scheme is executed to correct errors in the current data, thereby reducing the error rate of the current data.

[0006] By adopting the above technical solution, this approach dynamically monitors performance metrics reflecting the data error rate and adaptively switches the LDPC codeword scheme to match the current error rate. This allows the system to operate in a high code rate, low redundancy mode under most normal operating conditions (e.g., when the error rate is low), significantly reducing the storage space occupied by verification data and decreasing computational latency and power consumption due to reduced encoding / decoding computational complexity. Under a few adverse operating conditions (e.g., when the error rate is high), the system can seamlessly switch to a strong error correction mode to ensure data integrity. Thus, it achieves global optimization of system resource utilization from a time perspective.

[0007] Optionally, switching the current LDPC codeword scheme to the target codeword scheme and executing the target codeword scheme to correct the current data includes: When determining the target codeword scheme, if the determined target codeword scheme is different from the currently executed LDPC codeword scheme, a transition time range is determined, wherein the transition time range covers several data frames being processed before and after the current moment and those that are about to be processed. For all data within the transition time range, the two LDPC codeword schemes before and after the switch are used for parallel error correction, and the correct data result is selected to complete the error correction.

[0008] By adopting the above technical solution, a brief "overlapping decoding window" is introduced. During this period, the old and new error correction schemes work in parallel. Through the three-stage process of time overlap, parallel decoding, and result arbitration, it is ensured that each frame of data has two opportunities to be corrected during the critical period of switching codeword schemes, thereby achieving a truly gradual switching, that is, a smooth transition of the switching process.

[0009] Optionally, the step of performing parallel error correction using two LDPC codeword schemes before and after the switch for all data within the transition time range includes: The current LDPC codeword scheme before the switch is taken as the original LDPC codeword scheme, and the switching direction from the original LDPC codeword scheme to the target codeword scheme is determined; wherein, if the error correction capability of the original LDPC codeword scheme is lower than that of the target codeword scheme, the corresponding switching direction is an upgrade switch; otherwise, the corresponding switching direction is a downgrade switch. If it is an upgrade switch, within the transition time range, the original LDPC codeword scheme is used as the main decoding scheme. When the main decoding scheme fails to process or the output confidence is lower than the preset confidence, the target codeword scheme is used as the slave decoding scheme to perform error correction processing on the corresponding failed data. When the startup frequency and success rate of the slave decoding scheme exceed the preset first threshold, the target decoding scheme is used as the main decoding scheme to complete the switch. If it is a downgrade switch, within the transition time range, the original LDPC codeword scheme is used as the primary decoding scheme, a portion of the data frames within the transition time range are extracted, and the target codeword scheme is used as the secondary decoding scheme to process the extracted data frames. When the success rate of the secondary decoding scheme exceeds a preset second threshold, the target codeword scheme is used as the primary decoding scheme to complete the switch.

[0010] By adopting the above technical solution, the main decoding scheme is the LDPC codeword scheme used for data error correction in the current environment. During switching, the aim is to switch the target codeword scheme to the main decoding scheme. The above solution mainly discloses the specific method of "when to switch the target codeword scheme to the main decoding scheme" within the transition time range. That is, for all data within the transition time range, not all data is processed by the two codeword schemes before and after the switch. Instead, the data is processed separately while the original codeword scheme is the main decoding scheme. Only when specific conditions are met will the target decoding scheme be triggered to process the same data a second time as the secondary decoding scheme. This reduces the amount of data that is processed twice and reduces unnecessary waste of processing resources while ensuring data reliability under any circumstances.

[0011] Optionally, extracting a portion of the data frames within the transition time range includes: If it is a downgrade switch, the internal features are extracted from the current data frame decoded by the main decoding scheme. Based on the internal features, the failure probability of the current data frame being processed by the target codeword scheme is predicted, and data frames with a predicted failure probability higher than a preset probability value are extracted.

[0012] By adopting the above technical solution, precise sampling inspection is achieved, and only frames with a high probability of failure are verified. The highest verification value is obtained with the least computational overhead. Even after the main channel is successfully decoded, high-risk data frames can still be identified and verified in a focused manner before the downgrade switch, which greatly improves the security of the downgrade decision.

[0013] Optionally, the performance indicators may also include external environment data and equipment health status data; The method further includes: Real-time recording of performance index data and the adopted LDPC codeword scheme; analysis of the intrinsic relationship and evolution law between performance indicators in historical periods; identification of key critical points regarding error correction requirements in the equipment life cycle; and division of the equipment life cycle into two main stages based on the key critical points; wherein the main stages include a first stable stage of stable use of LDPC codeword scheme with low error correction capability, and a second stable stage of stable use of LDPC codeword scheme with high error correction capability. Before the current time reaches the critical critical point, the LDPC codeword scheme that is stably used corresponding to the critical critical point is pre-selected as the target codeword scheme.

[0014] By adopting the above technical solution, and by dividing the device lifecycle into stages and preloading the codeword scheme before critical critical points, a paradigm shift from "reactive switching" to "proactive preparation" has been achieved for the first time in the field of adaptive error correction. This solution completely eliminates the performance window caused by codeword scheme loading delays, achieving truly seamless, zero-latency switching; by pre-positioning and distributing high-load tasks, it smooths system power consumption peaks and optimizes computing resource allocation; ultimately, it significantly improves the reliability, stability, and determinism of the system at critical nodes in its lifecycle, providing crucial protection for highly reliable application scenarios.

[0015] Optionally, the method further includes: Calculate the correlation coefficient between bit error rate and external environment data in historical performance index data; Based on the correlation coefficient and the periodicity of the external environment data, the sub-periods in the first stable stage and the second stable stage are predicted respectively. The sub-periods refer to the periods when the code word scheme may be frequently switched due to changes in the external environment. When the current time enters any sub-time period, and during the downgrade switching, the extracted data frames are added to a preset queue for processing by reading data frames from the preset queue from the decoding scheme.

[0016] By adopting the above technical solution, the queue plays a buffering role, effectively dealing with a sudden surge in sampling requests and preventing system congestion; by controlling the size of the queue, simple flow control can be achieved.

[0017] Optionally, obtaining the performance metrics reflecting the error rate of the current data includes: If the current moment is in the first stable phase, then the performance index reflecting the error rate of the current data is obtained according to the preset first frequency; If the current moment is in the second stable phase, the performance index reflecting the error rate of the current data is obtained according to the preset second frequency; wherein, the preset first frequency is higher than the preset second frequency.

[0018] By adopting the above technical solution, the monitoring of performance indicators is triggered at a higher frequency during the first stabilization phase, so as to avoid the LDPC codeword scheme with low error correction capability from failing to effectively perform error correction processing on the data.

[0019] Secondly, this application provides an adaptive adjustment-based LDPC error correction system, including, A performance indicator monitoring module is used to acquire performance indicators that reflect the error rate of the current data, and the performance indicators include at least the bit error rate. The error correction capability verification module is used to compare the performance index with a preset threshold, and select a target codeword scheme from a number of pre-stored LDPC codeword schemes based on the comparison result, wherein the number of LDPC codeword schemes have different error correction capabilities. The error correction scheme switching module is used to switch the current LDPC codeword scheme to the target codeword scheme and execute the target codeword scheme to correct the current data, thereby reducing the error rate of the current data.

[0020] Thirdly, this application provides an adaptive adjustment-based LDPC error correction device, including a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed as described in any of the first aspects. Fourthly, this application provides a computer-readable storage medium storing a computer program that can be loaded by a processor and executed as described in any of the first aspects.

[0021] In summary, this application includes the following beneficial technical effects: In this application, the LDPC codeword scheme is adaptively switched to match the current data error rate by dynamically monitoring performance indicators that reflect the data error rate. This allows the system to operate in a high code rate, low redundancy mode under most normal operating conditions (such as when the error rate is low), significantly reducing the storage space occupied by verification data and reducing computational latency and system power consumption due to the reduced computational complexity of encoding and decoding. Under a few severe operating conditions (such as when the error rate is high), the system can seamlessly switch to a strong error correction mode to ensure data integrity. Thus, global optimization of system resource utilization is achieved from a time perspective. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating the adaptive adjustment-based LDPC error correction method disclosed in an embodiment of this application.

[0024] Figure 2 This is a structural block diagram of the adaptive adjustment-based LDPC error correction system disclosed in the embodiments of this application.

[0025] Explanation of reference numerals in the attached diagram: 201, Performance indicator monitoring module; 202, Error correction capability verification module; 203, Error correction scheme switching module. Detailed Implementation

[0026] The following is in conjunction with the appendix Figure 1-2 This application will be described in further detail.

[0027] This application discloses an adaptive adjustment-based LDPC error correction method (hereinafter referred to as the error correction method), the execution subject of which is an adaptive adjustment-based LDPC error correction system (hereinafter referred to as the error correction system). The following will be combined with... Figure 1 The specific steps of the error correction system in implementing the error correction method are described in detail.

[0028] S101, obtain performance metrics that reflect the error rate of the current data, including at least the bit error rate.

[0029] S102, compare the performance indicators with the preset threshold, and select the target codeword scheme from several pre-stored LDPC codeword schemes based on the comparison results. Among them, several LDPC codeword schemes have different error correction capabilities.

[0030] S103, switch the current LDPC codeword scheme to the target codeword scheme, execute the target codeword scheme to correct errors in the current data, thereby reducing the error rate of the current data.

[0031] In practice, the error correction system is used to acquire performance metrics that reflect the data error rate in real time or periodically. These performance metrics include bit error rate (i.e., bit error rate), external environmental sensor data, and the internal health status of the device.

[0032] Specifically, during data decoding, the current channel bit error rate (BER) can be estimated in real time by calculating the difference in soft information before and after decoding or the number of check equations that fail to be satisfied. By accessing environmental sensors (such as particle detectors and radiation dosimeters) to acquire data on space radiation intensity, and considering that high-energy particle bombardment is a major cause of memory errors in space electronic devices (such as single-event flips), environmental data can serve as an advanced predictor of an impending increase in the error rate. Additionally, data reflecting the health of storage devices (such as flash memory), such as block erase / write cycles (P / ECycles), read distortion rate, and the number of bad blocks, can be obtained by reading smart status data, such as the number of erase / write cycles (P / ECycles), read distortion rate, and bad block growth. Deterioration in these data indicates a natural increase in the original bit error rate.

[0033] If the received performance metrics (such as bit error rate, radiation intensity, and bad block growth) consistently exceed a preset threshold for a specified duration, the environment is considered severe, requiring an increase in the error correction level to select a more powerful LDPC codeword scheme as the target codeword scheme (as described in Scheme A below). Conversely, if the received metrics do not exceed the preset threshold, the environment is considered normal, and the error correction level can be reduced to save resources, selecting a less powerful LDPC codeword scheme as the target codeword scheme (as described in Scheme B below). It should be noted that since there are multiple performance metrics mentioned above, this application proposes pre-assigning weights to each performance metric. The product of the expected weights for each performance metric is calculated, and the sum of the products of all performance metrics (i.e., weighted summation) is used to compare against a preset threshold. If this result is higher than the preset threshold, the environment is considered severe; otherwise, it is considered normal.

[0034] Correspondingly, several LDPC codeword schemes are pre-stored, such as: Option A: Higher bit rate, shorter parity bit, corresponding error correction capability ≤72bit / 2kb, suitable for normal environments; Option B: Lower bit rate, longer parity bits, corresponding error correction capability ≤144bit / 2kb, suitable for harsh environments.

[0035] In other embodiments, more LDPC codeword schemes may be included, and a preset correspondence table may be set to store the correspondence between the comparison deviation range between the performance index and the preset threshold and the LDPC codeword scheme, so as to determine the corresponding target codeword scheme by looking up the correspondence table.

[0036] Next, using a pre-configured LDPC codec (which is a reconfigurable hardware IP core (or software program)), the parity check matrix and generator matrix of the target codeword scheme are loaded into its internal working register or memory. Thereafter, all encoding and decoding operations on the data will be performed based on the newly loaded parity check matrix and generator matrix, thus realizing the loading and updating of the LDPC codec and the switching of the LDPC codeword scheme.

[0037] Optionally, S103 specifically includes the following sub-steps: S1031, when determining the target codeword scheme, if the determined target codeword scheme is different from the currently executed LDPC codeword scheme, then a transition time range is determined, wherein the transition time range covers several data frames being processed before and after the current moment and those that are about to be processed. S1032: For all data within the transition time range, use the two LDPC codeword schemes before and after the switch to perform parallel error correction, and select the data result with correct output verification to complete the error correction.

[0038] In implementation, whenever the target codeword scheme is determined (taking this time point as the current time t), a transition time range centered on the current time t is determined. This range covers several data frames that are being executed and those that are about to be executed before and after the current time, such as from the Nth frame of data that is currently being processed to the end of the N+Mth frame of data. The size of M depends on the depth of the system buffer and the processing latency.

[0039] During the aforementioned transition period, the configurable LDPC codec within the error correction system will be temporarily configured to a "full-duplex mode," or there will be two logically independent decoding channels within the system (which can be a time-division multiplexed physical core or two physical cores). This will allow decoding channel one to continue using the LDPC codeword scheme before the switch during the transition period, while decoding channel two will use the currently determined target LDPC codeword scheme for decoding. As a result, each frame of data received during the transition period will be processed once by each of the two different LDPC schemes, producing two different decoding results. In communication and storage systems, after a data block (frame) is decoded by LDPC, a final validity verification step is usually performed. The most common and simplest method is CRC check. Therefore, this application proposes that for data frames that have been processed twice and produce different decoding results, a CRC value is calculated for the corresponding data frame before LDPC decoding, and the CPC value is recalculated for the decoded data during decoding to achieve consistency comparison of the CRC before and after decoding. If the comparison is consistent, the verification is correct; otherwise, the decoding may have failed and the decoding result is unreliable. Finally, the decoding result that has been successfully verified is selected. If the decoding results of both decoding channels pass the CRC check, the decoding result corresponding to the current target codeword scheme is selected. If the decoding results of both decoding channels fail the CRC check, the decoding result with higher confidence (LDPC software decoding algorithms usually output a confidence value, such as LLR Sum) is selected, and the error correction operation within the transition time range is finally completed.

[0040] Optionally, the step "for all data within the transition time range, perform parallel error correction using the two LDPC codeword schemes before and after the switch" in S1032 specifically includes the following sub-steps: The current LDPC codeword scheme before the switch is taken as the original LDPC codeword scheme, and the switching direction from the original LDPC codeword scheme to the target codeword scheme is determined. If the error correction capability of the original LDPC codeword scheme is lower than that of the target codeword scheme, the corresponding switching direction is an upgrade switch; otherwise, the corresponding switching direction is a downgrade switch. If it is an upgrade switch, during the transition period, the original LDPC codeword scheme will be used as the primary decoding scheme. When the primary decoding scheme fails to process or the output confidence level is lower than the preset confidence level, the target codeword scheme will be used as the secondary decoding scheme to perform error correction processing on the corresponding failed data. When the activation frequency and success rate of the secondary decoding scheme exceed the preset first threshold, the target decoding scheme will be used as the primary decoding scheme to complete the switch. If it is a downgrade handover, within the transition time range, the original LDPC codeword scheme is used as the primary decoding scheme. Internal features are extracted from the current data frame decoded by the primary decoding scheme. Based on the internal features, the failure probability of the current data frame being processed by the target codeword scheme is predicted. Data frames with a predicted failure probability higher than a preset probability value are extracted. The target codeword scheme is then used as the secondary decoding scheme to process the extracted data frames. When the success rate of the secondary decoding scheme exceeds a preset second threshold, the target codeword scheme is used as the primary decoding scheme, and the handover is completed.

[0041] In implementation, optionally, this application defaults to using the previous LDPC codeword scheme (i.e., the original LDPC codeword scheme) as the primary decoding scheme, which is used as the sole codeword scheme for data error correction processing before the next switch. When a switch is required (i.e., when the target codeword scheme is used as the primary decoding scheme), as mentioned above, this application will generate a transition time range. However, within this transition time range, this application will prioritize using the primary decoding scheme as the sole LDPC codeword scheme to process the data within the transition time range and determine the switching direction (upgrade switching / downgrade switching).

[0042] For upgrade switching, the target codeword scheme will be used as the slave decoding scheme to re-process the data that the current primary decoding scheme (i.e., LDPC codeword scheme) failed to process (i.e., CRC check failed) or whose output confidence (i.e., the LDPC software decoding algorithm usually outputs a confidence value, such as LLR Sum) is lower than the preset confidence. The frequency of the slave decoding scheme being activated and the probability of successful processing (i.e., CRC check successful) are recorded in real time. When both the activation frequency and the success rate exceed the corresponding preset first threshold, the target decoding scheme is used as the primary decoding scheme, thus achieving the switch.

[0043] If it is a downgrade switch, during the period when the primary decoding scheme is used as the only LDPC codeword scheme to process the data within the transition time range, some data frames are extracted, and the target codeword scheme is started as the secondary decoding scheme to process the extracted data frames. When the success rate of the secondary decoding scheme exceeds the preset second threshold, the target codeword scheme is used as the primary decoding scheme to achieve the switch.

[0044] The following scheme is used for extracting data frames: Within the transition time range, whenever a newly acquired data frame is decoded using the main decoding scheme, the internal features of that data frame are extracted to form a feature vector, wherein the internal features include any of the following data: The number of iterations (i.e., the number of iterations required for the main decoder to successfully decode the data frame; the more iterations, the more difficult it is to correct the error and the higher the probability of failure), the average absolute confidence (i.e., the average of the absolute values ​​of the log-likelihood ratio (LLR) of all bits after decoding; the lower the average absolute confidence, the lower the overall confidence of the data frame and the higher the probability of failure), the minimum absolute confidence (the minimum value among the absolute values ​​of the LLR of all bits), and the sum of checksums (i.e., the sum of the number of checksums that are not satisfied (i.e., the result is 1) after iterative decoding).

[0045] The feature vector is input into a pre-defined prediction model, and the failure probability P_fail (a number between 0 and 1) is calculated in real time. This failure probability is then compared to a pre-defined probability value. If the failure probability is greater than the pre-defined probability value, the corresponding data frame is extracted and processed by the decoding scheme. The prediction model is either logistic regression or a small decision tree, and it is pre-trained. Its input is the feature vector, and its output is the failure probability. Specifically, the model training process is as follows: In a laboratory environment, various channel conditions are simulated, and a large number of data frames are collected. For each frame, a low-error-correction LDPC codeword scheme (i.e., a secondary decoding scheme in degraded handover) is used for decoding, and its success or failure result is recorded as a label. Then, using the collected data pairs consisting of the feature vector and the label, a logistic regression model is trained on the server. This model will learn the following function: P_fail = σ(preset weight 1 * number of iterations + preset weight 2 * absolute value of average confidence + preset weight 3 * absolute value of minimum confidence + preset bias); where σ is the Sigmoid function.

[0046] Optionally, the error correction method may also include the following steps: Real-time recording of performance index data and the adopted LDPC codeword scheme; analysis of the intrinsic relationship and evolution pattern between performance indicators in historical periods; identification of key critical points regarding error correction requirements in the equipment lifecycle; and division of the equipment lifecycle into two main stages based on the key critical points. The main stages include the first stable stage of stable use of LDPC codeword scheme with low error correction capability, and the second stable stage of stable use of LDPC codeword scheme with high error correction capability. Before the current time reaches the critical critical point, the LDPC codeword scheme that is stably used at the critical critical point is pre-selected as the target codeword scheme. Calculate the correlation coefficient between bit error rate and external environment data in historical performance index data; Based on the correlation coefficient and the periodicity of external environmental data, the sub-periods in the first and second stable stages are predicted respectively. The sub-periods refer to the periods when the code word scheme may be frequently switched due to changes in the external environment. When the current time enters any sub-time period, and during the downgrade handover, the extracted data frames are added to a preset queue for processing by reading data frames from the preset queue from the decoding scheme. The corresponding S101 further includes the following steps: If the current moment is in the first stable phase, then the performance index reflecting the error rate of the current data is obtained according to the preset first frequency; If the current moment is in the second stable phase, the performance index reflecting the error rate of the current data is obtained according to the preset second frequency; wherein, the preset first frequency is higher than the preset second frequency.

[0047] In implementation, the acquired performance index data and corresponding LDPC codeword schemes are collected in real time. It should be noted that the LDPC codeword scheme refers to the target codeword scheme determined after comparing the corresponding performance index data with a preset threshold. The correspondence between performance indices and LDPC codeword schemes is established and stored. Next, the error correction system periodically selects time-series data of equipment health status indicators from the stored performance indices and applies filtering algorithms (such as Kalman filtering) for smoothing to remove short-term fluctuations. Then, a sliding window analysis of variance or a Bayesian change point detection algorithm is used to traverse the aforementioned time-series data, calculate the abrupt change locations (i.e., abrupt changes in the time series) of its statistical characteristics (such as mean and variance), and mark the corresponding abrupt change locations as candidate critical points.

[0048] For each candidate critical point, the Bit Error Rate (BER) data within a time period before and after it (e.g., 30 days before and after) is analyzed. The Mann-Whitney U test (a non-parametric statistical test method) is used to determine whether the BER baseline level after the candidate critical point is statistically significantly higher than the level before (the candidate critical point). If so, the corresponding candidate critical point is identified as a critical point. It should be noted that the final determined critical point is unique (i.e., when selecting a critical point, the candidate critical point with the most significant increase in the BER baseline level is selected). Correspondingly, the time period before the critical point is taken as the first stable phase, and the time period after the critical point is taken as the second stable phase.

[0049] Furthermore, Fourier transform or autocorrelation analysis is performed on the external environmental data (such as solar radiation intensity) in the historical performance indicators to extract their periodic patterns (such as the 11-year solar activity cycle, the 1-year orbital period, and the 1-day rotation period). Based on the BER and external environmental data in each performance indicator data in the historical storage, the Pearson correlation coefficient or Spearman rank correlation coefficient between the BER and the external environmental data is calculated. If the absolute value of the correlation coefficient exceeds a preset threshold (such as 0.7), the two are considered to be strongly correlated. Then, time series periodic analysis (such as using Fourier transform FFT or periodogram) is performed on the external environmental data (such as daily radiation intensity) to find one or more significant main periods (such as 1 year, 27 days, and 1 day). For each period found, the mean (μ) and standard deviation (σ) of the environmental data at the time points within that period in the entire historical record are calculated.

[0050] Analyze whether there are periods in the first and second stable phases where external environmental data periodically exceeds a high threshold (e.g., μ + nσ, n=2) and historically, BER and switching frequency also spike during the same period. For example, taking a 1-year cycle, align the external environmental data for the "same day" in all historical years, and for each time point within the cycle (e.g., July 15th of each year), statistically analyze the following two points: Does the environmental data at this point in time exceed a high threshold (e.g., μ+nσ, n=2)? At this point in time for these years, are the average BER and switching frequency significantly higher than the annual average (e.g., use hypothesis testing to see if the p-value is <0.05)? If it exists, the corresponding time period will be marked as a period of frequent codeword switching (i.e., a sub-period), such as a few specific months of the year (peak solar activity) or a few specific days of the month (near-Earth radiation zone).

[0051] Finally, the current time is compared with the aforementioned time periods. If the current time is in the first stable phase, the device will default to loading a low-error-correction LDPC codeword scheme when it starts up. If the difference between the current time and the start time of the second stable phase is less than a preset difference, the high-error-correction LDPC codeword scheme will be used as the target codeword scheme in advance to achieve a gradual transition in advance (i.e., using the high-error-correction LDPC codeword scheme and the low-error-correction LDPC codeword scheme to process data in parallel). When the current time enters any sub-time period, a preset queue is called in advance so that when a downgrade switch occurs in the sub-time period, the extracted data frame is added to the preset queue (e.g., data enters at the tail of the queue and exits at the head of the queue), and the corresponding data frame is read from the preset queue for reading from the decoding scheme.

[0052] Furthermore, regarding the content of "obtaining a performance index reflecting the error rate of the current data" in S101, this application proposes that: if the current time is in the first stable stage, the performance index reflecting the error rate of the current data is obtained according to a preset first frequency; if the current time is in the second stable stage, the performance index reflecting the error rate of the current data is obtained according to a preset second frequency, and the preset first frequency is higher than the preset second frequency.

[0053] This application also discloses an LDPC error correction system based on adaptive adjustment. (Refer to...) Figure 2 ,include: The performance indicator monitoring module 201 is used to acquire performance indicators that reflect the error rate of the current data. The performance indicators include at least the bit error rate. The error correction capability verification module 202 is used to compare the performance indicators with the preset thresholds and select the target codeword scheme from a number of pre-stored LDPC codeword schemes based on the comparison results. The number of LDPC codeword schemes have different error correction capabilities. The error correction scheme switching module 203 is used to switch the current LDPC codeword scheme to the target codeword scheme and execute the target codeword scheme to correct the current data, thereby reducing the error rate of the current data.

[0054] Optionally, the error correction scheme switching module 203 is also used to determine a transition time range when determining the target codeword scheme. If the determined target codeword scheme is different from the currently executed LDPC codeword scheme, the transition time range covers several data frames being processed before and after the current moment and the upcoming data frames. For all data within the transition time range, the two LDPC codeword schemes before and after the switch are used for parallel error correction, and the correct data results are selected to complete the error correction.

[0055] Optionally, the error correction scheme switching module 203 is also used to take the current LDPC codeword scheme before the switch as the original LDPC codeword scheme and determine the switching direction from the original LDPC codeword scheme to the target codeword scheme; wherein, if the error correction capability of the original LDPC codeword scheme is lower than that of the target codeword scheme, the corresponding switching direction is an upgrade switch; otherwise, the corresponding switching direction is a downgrade switch. If it is an upgrade switch, during the transition period, the original LDPC codeword scheme will be used as the primary decoding scheme. When the primary decoding scheme fails to process or the output confidence level is lower than the preset confidence level, the target codeword scheme will be used as the secondary decoding scheme to perform error correction processing on the corresponding failed data. When the activation frequency and success rate of the secondary decoding scheme exceed the preset first threshold, the target decoding scheme will be used as the primary decoding scheme to complete the switch. If it is a downgrade switchover, within the transition time range, the original LDPC codeword scheme is used as the primary decoding scheme, a portion of the data frames within the transition time range are extracted, and the target codeword scheme is started as the secondary decoding scheme to process the extracted data frames. When the success rate of the secondary decoding scheme exceeds the preset second threshold, the target codeword scheme is used as the primary decoding scheme to complete the switchover.

[0056] Optionally, the error correction scheme switching module 203 is also used to extract internal features from the current data frame decoded by the main decoding scheme if it is a downgrade switching, predict the failure probability of the current data frame being processed by the target codeword scheme based on the internal features, and extract data frames whose predicted failure probability is higher than the preset probability value.

[0057] Optionally, a scheme preloading module is also included, which is used to record performance index data and the adopted LDPC codeword scheme in real time, analyze the inherent relationship and evolution law between performance indicators in historical periods, identify the key critical points of error correction requirements in the equipment life cycle, and divide the equipment life cycle into two main stages based on the key critical points. The main stages include a first stable stage of stable use of LDPC codeword scheme with low error correction capability, and a second stable stage of stable use of LDPC codeword scheme with high error correction capability. Before the current time reaches the key critical point, the LDPC codeword scheme that is stably used at the key critical point is pre-selected as the target codeword scheme.

[0058] Optionally, the scheme preloading module is also used to calculate the correlation coefficient between the bit error rate and the external environment data in the historical performance index data; based on the correlation coefficient and the periodicity of the external environment data, it predicts the sub-periods in the first stable stage and the second stable stage respectively. The sub-period refers to the period when the codeword scheme may frequently switch due to changes in the external environment; when the current time enters any sub-period, and during the downgrade switch, the extracted data frame is added to the preset queue for processing by reading the data frame in the preset queue from the decoding scheme.

[0059] Optionally, the performance indicator monitoring module 201 is further configured to acquire a performance indicator reflecting the error rate of the current data at a preset first frequency if the current time is in the first stable phase; and to acquire a performance indicator reflecting the error rate of the current data at a preset second frequency if the current time is in the second stable phase; wherein the preset first frequency is higher than the preset second frequency.

[0060] This application also discloses an adaptive adjustment-based LDPC error correction device, which includes a memory and a processor. The memory stores a computer program that can be loaded by the processor and executed as described above for the adaptive adjustment-based LDPC error correction method.

[0061] This application also discloses a computer-readable storage medium that stores a computer program that can be loaded by a processor and executed as described above in the adaptive adjustment-based LDPC error correction method. The computer-readable storage medium includes, for example, various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0062] It should be noted that in this paper, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0063] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit the scope of protection of the application. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on these embodiments, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

Claims

1. An LDPC error correction method based on adaptive adjustment, characterized in that, include: Obtain a performance metric that reflects the error rate of the current data, wherein the performance metric includes at least the bit error rate; The performance index is compared with a preset threshold, and a target codeword scheme is selected from a number of pre-stored LDPC codeword schemes based on the comparison results. The number of LDPC codeword schemes have different error correction capabilities. The current LDPC codeword scheme is switched to the target codeword scheme, and the target codeword scheme is executed to correct errors in the current data, thereby reducing the error rate of the current data.

2. The LDPC error correction method based on adaptive adjustment according to claim 1, characterized in that, The step of switching the current LDPC codeword scheme to the target codeword scheme and executing the target codeword scheme to correct the current data includes: When determining the target codeword scheme, if the determined target codeword scheme is different from the currently executed LDPC codeword scheme, a transition time range is determined, wherein the transition time range covers several data frames being processed before and after the current moment and those that are about to be processed. For all data within the transition time range, the two LDPC codeword schemes before and after the switch are used for parallel error correction, and the correct data result is selected to complete the error correction.

3. The LDPC error correction method based on adaptive adjustment according to claim 2, characterized in that, The step of performing parallel error correction for all data within the transition time range using two LDPC codeword schemes before and after the switch includes: The current LDPC codeword scheme before the switch is taken as the original LDPC codeword scheme, and the switching direction from the original LDPC codeword scheme to the target codeword scheme is determined; wherein, if the error correction capability of the original LDPC codeword scheme is lower than that of the target codeword scheme, the corresponding switching direction is an upgrade switch; otherwise, the corresponding switching direction is a downgrade switch. If it is an upgrade switch, within the transition time range, the original LDPC codeword scheme is used as the main decoding scheme. When the main decoding scheme fails to process or the output confidence is lower than the preset confidence, the target codeword scheme is used as the slave decoding scheme to perform error correction processing on the corresponding failed data. When the startup frequency and success rate of the slave decoding scheme exceed the preset first threshold, the target decoding scheme is used as the main decoding scheme to complete the switch. If it is a downgrade switch, within the transition time range, the original LDPC codeword scheme is used as the primary decoding scheme, a portion of the data frames within the transition time range are extracted, and the target codeword scheme is used as the secondary decoding scheme to process the extracted data frames. When the success rate of the secondary decoding scheme exceeds a preset second threshold, the target codeword scheme is used as the primary decoding scheme to complete the switch.

4. The LDPC error correction method based on adaptive adjustment according to claim 3, characterized in that, The extraction of a portion of data frames within the transition time range includes: If it is a downgrade switch, the internal features are extracted from the current data frame decoded by the main decoding scheme. Based on the internal features, the failure probability of the current data frame being processed by the target codeword scheme is predicted, and data frames with a predicted failure probability higher than a preset probability value are extracted.

5. The LDPC error correction method based on adaptive adjustment according to claim 3, characterized in that, The performance indicators also include external environment data and equipment health status data; The method further includes: Real-time recording of performance index data and the adopted LDPC codeword scheme; analysis of the intrinsic relationship and evolution law between performance indicators in historical periods; identification of key critical points regarding error correction requirements in the equipment life cycle; and division of the equipment life cycle into two main stages based on the key critical points; wherein the main stages include a first stable stage of stable use of LDPC codeword scheme with low error correction capability, and a second stable stage of stable use of LDPC codeword scheme with high error correction capability. Before the current time reaches the critical critical point, the LDPC codeword scheme that is stably used corresponding to the critical critical point is pre-selected as the target codeword scheme.

6. The LDPC error correction method based on adaptive adjustment according to claim 5, characterized in that, The method further includes: Calculate the correlation coefficient between bit error rate and external environment data in historical performance index data; Based on the correlation coefficient and the periodicity of the external environment data, the sub-periods in the first stable stage and the second stable stage are predicted respectively. The sub-periods refer to the periods when the code word scheme may be frequently switched due to changes in the external environment. When the current time enters any sub-time period, and during the downgrade switching, the extracted data frames are added to a preset queue for processing by reading data frames from the preset queue from the decoding scheme.

7. The LDPC error correction method based on adaptive adjustment according to claim 5, characterized in that, The acquisition of performance metrics reflecting the error rate of the current data includes: If the current moment is in the first stable phase, then the performance index reflecting the error rate of the current data is obtained according to the preset first frequency; If the current moment is in the second stable phase, the performance index reflecting the error rate of the current data is obtained according to the preset second frequency; wherein, the preset first frequency is higher than the preset second frequency.

8. An LDPC error correction system based on adaptive adjustment, characterized in that, include, A performance indicator monitoring module (201) is used to obtain performance indicators that reflect the error rate of the current data, wherein the performance indicators include at least the bit error rate; The error correction capability verification module (202) is used to compare the performance index with a preset threshold and select a target codeword scheme from a number of pre-stored LDPC codeword schemes based on the comparison result, wherein the number of LDPC codeword schemes have different error correction capabilities. The error correction scheme switching module (203) is used to switch the current LDPC codeword scheme to the target codeword scheme and execute the target codeword scheme to correct the current data, thereby reducing the error rate of the current data.

9. An LDPC error correction device based on adaptive adjustment, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer program is stored that can be loaded by a processor and executed as described in any one of claims 1 to 7.

Images