An interconnection data transmission control system and method based on a 400G high-speed optical module

Abstract

The application discloses an interconnection data transmission control system and method based on a 400G high-speed optical module, relates to the technical field of transmission control, and comprises the following steps: after physical access and link initialization of the optical module are completed, a link basic state report of a perlane channel impulse response baseline is generated, and link operation data is collected to update the link basic state report; a predefined perturbation pattern is embedded in an Idle gap of a 400G PAM4 data stream, a channel impulse response change rate of each lane is extracted, a future link degradation probability is predicted in combination with a link health index, and a link degradation trend prediction report is generated; when the degradation probability and the perlane channel impulse response change amount in the link degradation trend prediction report satisfy preset trigger conditions, the degraded lane is set as a probation state, and asymmetric interleaving remapping is performed on the remaining active lanes. The application realizes early prediction, active degradation and automatic recovery of link degradation, and improves transmission stability and continuity of the 400G high-speed optical module interconnection link.

Description

Technical Field

[0001] This invention relates to the field of transmission control technology, and in particular to an interconnected data transmission control system and method based on a 400G high-speed optical module. Background Technology

[0002] With the development of cloud computing and high-performance computing, 400G high-speed optical modules are widely used in data centers and core networks. Their multi-lane parallel transmission and PAM4 modulation can significantly improve data throughput. In existing technologies, after completing physical access and link initialization, the optical module assesses the link status through bit error rate monitoring, received optical power, and equalizer parameters, providing a reference for link maintenance. It also uses FEC error correction counting or SNR analysis to predict certain degradation trends.

[0003] Monitoring based on overall bit error rate and equalizer parameters is not sensitive enough to minor lane-level degradation, making it difficult to detect potential risks in a timely manner. Furthermore, the fixed lane allocation mode lacks flexible scheduling capabilities when lane performance fluctuates in a short period of time. By embedding predefined perturbation patterns into the Idle gaps of 400GPAM4 data streams to extract the per-lane channel impulse response change rate, and combining it with the link health index to predict the probability of future degradation, fine-grained lane-level monitoring is achieved. Through asymmetric interleaving remapping and dynamic traffic scheduling, degradation control is performed before lane degradation occurs, and full bandwidth transmission is quickly restored after conditions recover. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides an interconnection data transmission control method based on a 400G high-speed optical module, which solves the problems of difficulty in timely detection of minor lane-level degradation and inflexible dynamic link scheduling.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides an interconnection data transmission control method based on a 400G high-speed optical module, which includes generating a link basic status report of the per-lane channel impulse response baseline after the optical module completes physical access and link initialization, and collecting link operation data to update the link basic status report. Predefined perturbation patterns are embedded in the Idle gaps of the 400GPAM4 data stream. The channel impulse response change rate of each lane is extracted, and the link health index is combined to predict the probability of future link degradation and generate a link degradation trend prediction report. When the degradation probability and per-lane channel impulse response change in the link degradation trend prediction report meet the preset trigger conditions, the degraded lane is set to the PROBATION state, asymmetric interleaving remapping is performed on the remaining active lanes, a degradation execution instruction is generated, and the overall link state is set to ASYMMETRIC_ACTIVE. Adjust the traffic scheduling parameters on the sending side according to the downgrade execution instruction, and generate the adjusted sending parameters; The monitoring link degradation trend prediction report triggers the UNFOLD operation to restore the original lane mapping rules and full bandwidth transmission, generating a recovery completion signal.

[0007] As a preferred embodiment of the interconnect data transmission control method based on a 400G high-speed optical module described in this invention, the updated link basic status report includes: The per-lane channel impulse response baseline is obtained using the link training sequence of the physical medium additional channel, and the per-lane channel impulse response baseline is written into the link basic status report; Using the per-lane channel impulse response baseline in the link basic status report as a reference, the instantaneous value of the per-lane channel impulse response in the link operation data is collected, and the difference between the instantaneous value of the per-lane channel impulse response and the per-lane channel impulse response baseline is obtained to obtain the change in the per-lane channel impulse response. Update the per-lane channel impulse response change to the corresponding per-lane field in the link basic status report, overwriting the original per-lane channel impulse response baseline difference record.

[0008] As a preferred embodiment of the interconnect data transmission control method based on a 400G high-speed optical module described in this invention, the link degradation trend prediction report includes: A predefined perturbation pattern is embedded in the Idle gap of the 400GPAM4 data stream. Frequency domain cross-correlation is performed on the received sampling sequence of the predefined perturbation pattern to obtain the instantaneous value of the per-lane channel impulse response. The instantaneous value of the per-lane channel impulse response is compared with the baseline of the per-lane channel impulse response in the link basic status report by tap phase rotation correction. The amplitude change of each tap is normalized and the change contribution is allocated according to the energy proportion of the tap to obtain the per-lane channel impulse response change rate. Using the per-lane channel impulse response change rate of each of the five consecutive monitoring windows as input, the per-lane channel impulse response change rates from the first to the fifth monitoring windows are denoted as follows: to ,calculate and The difference is used to obtain the first gradient, and the calculation is performed. and The difference between the first and second gradients is used to obtain the second gradient, and the difference between the second gradient and the first gradient is used to obtain the per-lane channel impulse response curvature. The error statistics contained in the link operation data in the link basic status report are mapped to the error impact coefficient, and the equalizer coefficient offset is mapped to the offset impact coefficient. The error impact coefficient and the offset impact coefficient are nonlinearly modulated using the per-lane channel impulse response change curvature to obtain the link health index. Using the per-lane channel impulse response change rate as a time series input, and the link health index as a distribution shape adjustment factor, a non-uniform stretching transformation is performed on the input sample set of the sliding window conditional probability model, and the prediction horizon is set to three consecutive monitoring windows. The first conditional probability value is calculated using the end time of the first monitoring window as the prediction starting point, the second conditional probability value is calculated using the end time of the second monitoring window as the prediction starting point, and the third conditional probability value is calculated using the end time of the third monitoring window as the prediction starting point. The first conditional probability value, the second conditional probability value, and the third conditional probability value are accumulated and merged according to the time decay factor to obtain the future link degradation probability, and a link degradation trend prediction report is generated.

[0009] As a preferred embodiment of the interconnect data transmission control method based on a 400G high-speed optical module described in this invention, the degradation execution instruction includes: The probability of future link degradation is input into the log-likelihood ratio transform unit and nonlinearly compressed to obtain the probability likelihood value; the rate of change of the per-lane channel impulse response is input into the arctangent transform unit and nonlinearly mapped to obtain the stable value of the rate of change. The probability likelihood value is compared with the preset likelihood threshold to obtain the first comparison result, and the stable value of the rate of change is compared with the preset stable threshold to obtain the second comparison result. When both the first comparison result and the second comparison result indicate that they exceed the preset normal range, the triggering condition is determined to be met. Input the determination result of the trigger condition being met into the state machine, retrieve the lane index field corresponding to the trigger condition in the link degradation trend prediction report, decode the lane index field into the physical channel number, and write the physical channel number into the address mapping table of the degradation lane register. Atomic write operations are performed on the state storage unit corresponding to the degraded lane via the state machine control bus to write the predefined PROBATION status code, and a status update instruction for the physical channel number and PROBATION status code is generated. Input the status update instruction with physical channel number and PROBATION status code into the remapping logic, extract the physical channel number field from the status update instruction, match the physical channel number in the parallel search array of the activelane list, and perform bubble sort on the remaining activelane identifiers according to the physical channel index number. Based on predefined asymmetric interleaving rules, the sorted activelane identifiers are reassigned virtual channel numbers to generate an asymmetric interleaving remapping table; The asymmetric interleaving remapping table is converted into a hardware register configuration sequence, and the hardware register configuration sequence is encapsulated into a degraded execution instruction.

[0010] As a preferred embodiment of the interconnect data transmission control method based on a 400G high-speed optical module described in this invention, the step of setting the overall link state to ASYMMETRIC_ACTIVE includes, The downgrade execution instruction is sent to the physical coding sublayer and the system waits for the physical coding sublayer to return a remapping completion signal. The remapping completion signal and the predefined ASYMMETRIC_ACTIVE status code are written to the link global status register, and the output value of the link global status register is updated to ASYMMETRIC_ACTIVE.

[0011] As a preferred embodiment of the interconnect data transmission control method based on a 400G high-speed optical module described in this invention, the adjusted transmission parameters include: The correspondence between virtual channels and physical channels in the asymmetric interleaving remapping table is mapped to virtual channel bandwidth quota parameters. The virtual channel bandwidth quota parameters are then input into the token bucket controller to update the token generation rate of each virtual channel, generating the updated token generation rate configuration. The updated token generation rate configuration is input into the weighted round-robin scheduler. The scheduling priority of the virtual channel is rearranged according to the token generation rate. The rearranged scheduling priority is mapped to the time-division multiplexing time slice allocation table, and a time slice allocation table update instruction is generated. Write the time slice allocation table update instruction into the configuration register of the transmitting side scheduler, and map the output of the configuration register to the adjusted transmitting parameters.

[0012] In a preferred embodiment of the interconnect data transmission control method based on a 400G high-speed optical module described in this invention, the recovery completion signal includes: The probability of future link degradation is input into the inverse log-likelihood ratio transformation unit for nonlinear restoration to obtain the recovery confidence. The recovery confidence is compared with the preset recovery confidence threshold. When the comparison results of multiple consecutive monitoring windows all indicate that the recovery confidence is lower than the preset recovery confidence threshold, the judgment result that the recovery condition is met is obtained. The result of the determination that the recovery condition is met is input into the state machine, the asymmetric interleaving remapping table is loaded into the address generation unit, the address generation unit calls the factory-preset original lane mapping rule lookup table, and writes the data in the original lane mapping rule lookup table into the interleaving address register group of the remapping logic in parallel to generate the UNFOLD operation instruction. The UNFOLD operation instruction is converted into a hardware register reset sequence, which is then sent to the physical coding sublayer to restore the original lane mapping rules and full-bandwidth transmission. The physical coding sublayer returns an operation completion status word, which is then encoded into a recovery completion signal.

[0013] Secondly, the present invention provides an interconnected data transmission control system based on a 400G high-speed optical module, comprising an update module that generates a basic link status report of the per-lane channel impulse response baseline and collects link operation data to update the basic link status report after the optical module completes physical access and link initialization. The prediction module embeds a predefined perturbation pattern in the Idle gap of the 400GPAM4 data stream, extracts the channel impulse response change rate of each lane, and combines it with the link health index to predict the probability of future link degradation and generate a link degradation trend prediction report. The mapping module sets the degraded lane to PROBATION state when the degradation probability and per-lane channel impulse response change in the link degradation trend prediction report meet the preset trigger conditions, performs asymmetric interleaving remapping on the remaining active lanes, generates a degradation execution instruction, and sets the overall link state to ASYMMETRIC_ACTIVE. The adjustment module adjusts the traffic scheduling parameters on the sending side according to the downgrade execution instruction, and generates the adjusted sending parameters; The transmission module monitors the link degradation trend prediction report, triggers the UNFOLD operation to restore the original lane mapping rules and full bandwidth transmission, and generates a recovery completion signal.

[0014] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the interconnect data transmission control method based on a 400G high-speed optical module as described in the first aspect of the present invention.

[0015] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the interconnect data transmission control method based on a 400G high-speed optical module as described in the first aspect of the present invention.

[0016] The beneficial effects of this invention are as follows: By establishing and updating the link basic status report of the per-lane channel impulse response baseline, continuous tracking of the operating status of each lane is achieved; a predefined perturbation pattern is embedded in the idle gap of the 400G PAM4 data stream to extract the per-lane channel impulse response change rate, and combined with the link health index to predict the probability of future link degradation, thus discovering potential degradation trends before the actual link failure occurs; when the preset triggering condition is met, the degraded lane is set to the PROBATION state, and asymmetric interleaving remapping is performed on the remaining active lanes to achieve proactive avoidance of risky channels by service traffic; the transmission-side traffic scheduling parameters are adjusted according to the remapping results to improve the efficiency of link resource utilization; after the link recovers and stabilizes, the original lane mapping rules and full bandwidth transmission are restored through the UNFOLD operation, realizing early prediction, proactive degradation and automatic recovery of link degradation, thereby improving the transmission stability and continuity of the 400G high-speed optical module interconnection link. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the 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.

[0018] Figure 1 This is a flowchart of an interconnect data transmission control method based on a 400G high-speed optical module.

[0019] Figure 2 This is a schematic diagram of an interconnected data transmission control system based on a 400G high-speed optical module. Detailed Implementation

[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0021] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0022] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0023] Reference Figures 1-2 This is one embodiment of the present invention, which provides an interconnect data transmission control method based on a 400G high-speed optical module, comprising the following steps: S1. After the optical module completes physical access and link initialization, a link basic status report of the per-lane channel impulse response baseline is generated, and link operation data is collected to update the link basic status report.

[0024] S1.1 Obtain the per-lane channel impulse response baseline using the link training sequence of the physical medium additional channel, and write the per-lane channel impulse response baseline into the link basic status report.

[0025] Furthermore, the channel impulse response characteristics of each lane are obtained through the link training sequence of the physical medium additional channel, and the collected per-lane channel impulse response baseline is written into the link basic status report. During the acquisition process, the timing and frequency characteristics provided by the link training sequence are used to accurately characterize the main path, front vernier, back vernier, and wake response of each lane, forming a stable baseline reference. This process comprehensively characterizes the signal transmission characteristics of each lane by collecting operating parameters such as equalizer tap coefficients, bit error statistics, and received optical power. A complete link basic status report is generated immediately after link initialization to ensure that the baseline of each lane can reflect the transmission status of the link under normal startup conditions.

[0026] S1.2. Using the per-lane channel impulse response baseline in the link basic status report as a reference, collect the instantaneous value of the per-lane channel impulse response in the link operation data, obtain the difference between the instantaneous value of the per-lane channel impulse response and the per-lane channel impulse response baseline, and obtain the change in the per-lane channel impulse response.

[0027] Furthermore, given that a per-lane channel impulse response baseline has been established in the link basic status report, the instantaneous values ​​of the per-lane channel impulse response in the link operation data are collected and compared with the corresponding per-lane channel impulse response baseline. The difference is then used to form the per-lane channel impulse response change. By continuously sampling the instantaneous response of each lane and normalizing the change by combining tap amplitude distribution and energy proportion, the transmission characteristics of each lane under different time windows can be accurately reflected. This process can capture the micro-degradation or transient fluctuations of lanes in the link in real time and generate the per-lane channel impulse response change, providing the necessary input for the dynamic updating of the link basic status report and ensuring that the link operation status continuously reflects the latest transmission status of each lane.

[0028] S1.3 Update the per-lane channel impulse response change to the corresponding per-lane field in the link basic status report, overwriting the original per-lane channel impulse response baseline difference record.

[0029] Furthermore, during the update process, historical trends are incorporated to maintain the continuity and temporal integrity of changes recorded for each lane, supporting subsequent predictions of link degradation trends. Through dynamic overlay and updates, the basic link status report continuously reflects the current transmission status of each lane, providing reliable input for active scheduling, asymmetric interleaving remapping, and subsequent UNFOLD operations, thus achieving synchronization between link operation data and link status representation.

[0030] S2. Embed a predefined perturbation pattern in the Idle gap of the 400GPAM4 data stream, extract the channel impulse response change rate of each lane, combine it with the link health index to predict the probability of future link degradation, and generate a link degradation trend prediction report.

[0031] S2.1. Embed a predefined perturbation pattern in the Idle gap of the 400GPAM4 data stream, and perform frequency domain cross-correlation operation on the received sampling sequence of the predefined perturbation pattern to obtain the instantaneous value of the per-lane channel impulse response.

[0032] Furthermore, the predefined perturbation pattern can be predetermined during the link initialization phase according to the level transition characteristics of the 400GPAM4 data stream, the length of the Idle gap, the lane correspondence, and the receiving sampling and identification requirements. The predefined perturbation pattern contains a fixed symbol arrangement and phase reference features that are easy to identify in the frequency domain. When the 400GPAM4 data stream enters the Idle gap, the predefined perturbation pattern is embedded in the Idle gap, so that the predefined perturbation pattern is transmitted with each lane. After the receiving side obtains the received sampling sequence of the predefined perturbation pattern, it performs frequency domain cross-correlation operation on the received sampling sequence of the predefined perturbation pattern. By the correlation peak position and frequency domain response difference between the predefined perturbation pattern and the received sampling sequence, the transient response characteristics of each lane after being affected by the channel are separated, and the instantaneous value of the per-lane channel impulse response is obtained.

[0033] Specifically, predefined perturbation patterns are carried within idle intervals, allowing link probing to avoid occupying valid service data locations. The fixed symbol arrangement and phase reference characteristics of the predefined perturbation patterns provide identifiable excitation sources for frequency domain cross-correlation calculations. For example, the predefined perturbation patterns can include alternating high and low level segments to observe high-frequency attenuation, smooth transition segments to observe low-frequency tailing, and lane-distinguishable segments to avoid confusion between different lane responses. Compared to monitoring methods that rely solely on bit error statistics or equalizer parameters, predefined perturbation patterns can actively excite the channel. Frequency domain cross-correlation calculations can extract transient responses closer to the channel itself from the received sampling sequence, giving the per-lane channel impulse response instantaneous values ​​lane-level distinguishability and a basis for trend analysis.

[0034] Frequency-level phase alignment operator: ; in, For each frequency point, phase shift, To receive the sampled sequence at the frequency point phase, For a predefined perturbation pattern at a frequency point phase, For frequency domain indexing.

[0035] Define the amplitude-spectral ratio function: ; in, For amplitude spectrum ratio, To prevent small constants from being divided by zero.

[0036] Spectral shape deviation integral: ; in, For the spectral shape deviation operator, For symbolic functions, This is the logarithm of the amplitude spectrum ratio.

[0037] Improved frequency domain channel characterization: ; in, To improve frequency domain channel characterization, is the base of the natural logarithm. The imaginary unit; The instantaneous value expression for the per-lane channel impulse response is: ; in, The instantaneous value of the per-lane channel impulse response. To improve frequency domain channel characterization, This is the inverse fast Fourier transform.

[0038] S2.2. Perform tap phase rotation correction on the instantaneous value of the per-lane channel impulse response and the baseline of the per-lane channel impulse response in the link basic status report. Normalize the amplitude change of each tap and allocate the change contribution according to the tap energy ratio to obtain the per-lane channel impulse response change rate.

[0039] Furthermore, the amplitude variation of each tap is normalized to unify the amplitude variations of different taps to a comparable scale. At the same time, the contribution of each tap to the amplitude variation is weighted according to its proportion in the total energy, giving greater weight to the variation of high-energy taps and less weight to the variation of low-energy taps. Finally, the per-lane channel impulse response variation rate is calculated by comprehensively calculating the amplitude variation of the taps for each lane, realizing a quantitative representation of the changes in lane transmission characteristics and ensuring that the variation rate can accurately reflect the actual situation of each lane being affected by the channel.

[0040] S2.3. Using the per-lane channel impulse response change rate of each of the five consecutive monitoring windows as input, the per-lane channel impulse response change rates from the first to the fifth monitoring windows are sequentially denoted as follows: to ,calculate and The difference is used to obtain the first gradient, and the calculation is performed. and The difference between the first and second gradients is used to obtain the second gradient, and the difference between the second gradient and the first gradient is used to obtain the per-lane channel impulse response curvature.

[0041] Furthermore, the five consecutive monitoring windows can be pre-determined based on the periodicity of the Idle intervals in the 400GPAM4 data stream, the data update rhythm of the link operation, and the stable observation requirements of the per-lane channel impulse response change rate; the per-lane channel impulse response change rates from the first to the fifth monitoring windows are denoted as follows: to First, the difference between and is calculated to obtain the first gradient, which represents the direction of change of the per-lane channel impulse response rate within the front-end monitoring window. Then, the difference between and is calculated to obtain the second gradient, which represents the direction of change of the per-lane channel impulse response rate within the back-end monitoring window. Finally, the difference between the second gradient and the first gradient is calculated to obtain the per-lane channel impulse response curvature, so that the instantaneous fluctuations within a single monitoring window are absorbed by the trend relationship between continuous monitoring windows, thus obtaining the per-lane channel impulse response curvature.

[0042] Specifically, the direction of change in the trend is expressed by the first gradient, the second gradient, and the curvature of the per-lane channel impulse response, transforming the link status judgment from judging the magnitude of change to judging the acceleration of change. For example, if the per-lane channel impulse response change rate briefly increases and then falls back within different monitoring windows, the first and second gradients show a trend reversal, and the curvature of the per-lane channel impulse response change will not be amplified by a single disturbance. If the per-lane channel impulse response change rate continues to increase and the later increase trend is stronger than the earlier increase trend, the difference between the second and first gradients can reflect the characteristics of accelerated degradation. Through the pre-division of five consecutive monitoring windows and the trend reference of the intermediate monitoring windows, the curvature of the per-lane channel impulse response change focuses on the trend and intensity of change in the channel impulse response process, rather than the instantaneous fluctuation characteristics at a single monitoring moment. Intermittent jitter typically manifests as localized abnormal changes within a short period. While such changes cause instantaneous shifts in the channel impulse response, they are difficult to establish a stable direction of change over a continuous time series. Therefore, the curvature characteristics corresponding to these changes in the curvature calculation process are characterized by short duration and weak continuity. Conversely, a continuous degradation process leads to a consistent direction of change in the channel impulse response over a continuous monitoring period, causing the rate and acceleration of change to gradually accumulate and ultimately reflect as a curvature change with continuous evolution characteristics. By characterizing the curvature of the channel impulse response change process, the impact of short-term random fluctuations on the results can be weakened, while enhancing the expressive power of long-term continuous change characteristics, preserving the continuous degradation trend, and obtaining the target per-lane channel impulse response curvature.

[0043] The first directed gradient expression is: ; in, The first directed gradient, The per-lane channel impulse response change rate for the first monitoring window. This represents the per-lane channel impulse response change rate for the third monitoring window.

[0044] The expression for the second directed gradient is: ; in, The second directed gradient, The per-lane channel impulse response change rate for the fifth monitoring window. This represents the per-lane channel impulse response change rate for the third monitoring window.

[0045] The expression for the curvature of the per-lane channel impulse response is: ; in, The curvature of the impulse response variation in the per-lane channel.

[0046] S2.4 Map the bit error statistics contained in the link operation data in the link basic status report to the bit error impact coefficient, map the equalizer coefficient offset to the offset impact coefficient, and use the per-lane channel impulse response change curvature to perform nonlinear modulation on the bit error impact coefficient and the offset impact coefficient to obtain the link health index.

[0047] Furthermore, the error statistics included in the link operation data are read from the link basic status report, and the error statistics are mapped to error impact coefficients according to the error accumulation trend to characterize the degree of transmission anomalies at the data decision level. At the same time, the equalizer coefficient offset is read, and the equalizer coefficient offset is mapped to offset impact coefficients according to the offset direction and offset magnitude to characterize the degree of transmission mismatch at the channel compensation level. The error impact coefficients and offset impact coefficients are nonlinearly modulated using the per-lane channel impulse response curvature, so that the error impact coefficients and offset impact coefficients are no longer used as isolated indicators for judgment, but are enhanced or suppressed according to the acceleration trend reflected by the per-lane channel impulse response curvature. When the per-lane channel impulse response curvature shows a continuous deterioration trend, the impact of the error impact coefficients and offset impact coefficients on the link status is enhanced; when the per-lane channel impulse response curvature shows a short-term decline or stabilization trend, the impact of the error impact coefficients and offset impact coefficients on the link status is suppressed, thus obtaining the link health index.

[0048] Specifically, bit error statistics, equalizer coefficient offset, and per-lane channel impulse response curvature are hierarchically fused. Bit error statistics reflect the transmission results, equalizer coefficient offset reflects the compensation pressure, and per-lane channel impulse response curvature reflects the acceleration of the degradation trend. For example, when bit error statistics increase but per-lane channel impulse response curvature tends to stabilize, the impact of occasional bit errors on the link health index can be reduced. When equalizer coefficient offset increases and per-lane channel impulse response curvature continues to rise, the judgment weight of channel mismatch evolving into link degradation can be strengthened. By normalizing result anomalies, compensation anomalies, and trend anomalies through nonlinear modulation, a weighted exponential function or a sigmoid function is used to perform a nonlinear transformation on each anomaly index. Then, the index is weighted and superimposed according to preset weights or interaction coefficients to generate the final nonlinear modulated link health index. The exponential or sigmoid function can amplify the impact of high anomalies while weakening the short-term fluctuations of low anomalies. This allows the link health index to not only reflect the current degree of anomaly in value but also naturally reflect the coupling relationship between result anomalies, compensation anomalies, and trend anomalies, thus achieving the fusion of multi-dimensional anomaly information on a single index.

[0049] S2.5. Using the per-lane channel impulse response change rate as a time series input, and the link health index as a distribution shape adjustment factor, a non-uniform stretching transformation is performed on the input sample set of the sliding window conditional probability model, and the prediction horizon is set to three consecutive monitoring windows.

[0050] Furthermore, non-uniform stretching transformation is achieved by applying non-linear scaling functions to the input samples of the sliding window on both the time and amplitude axes. For example, the normalized link health index is used as the stretching coefficient, and the time interval of each sampling point within the window is dynamically expanded or compressed using an exponential function in the time dimension. This allows samples in high-risk time periods to occupy a longer weight in the model input. In the amplitude dimension, a weighted sigmoid function is used to non-linearly amplify or compress the signal amplitude, giving greater weight to samples with high degradation amplitude changes while weakening low-amplitude disturbances. This achieves non-uniform stretching of the sample distribution, enabling the sliding window conditional probability model to capture the continuous change characteristics of the per-lane channel impulse response rate of change. This ensures that the predicted input samples accurately reflect the future link state, providing a reliable basis for calculating the future link degradation probability and achieving quantitative prediction of link degradation trends.

[0051] Specifically, the link health index is used as a distribution pattern adjustment factor. Non-uniform stretching transformation is used to adjust the input samples of the sliding window conditional probability model, allowing the time series to reflect instantaneous changes and also reflecting the adjustment of the link health status's sensitivity to trends. For example, when the per-lane channel impulse response change rate shows a continuous upward trend and the link health index decreases, non-uniform stretching enhances the impact of highly variable samples on the model, making future degradation predictions earlier and more reliable. When the change rate is low or the link health index remains stable, stretching reduces the weight of occasional disturbance samples, avoiding over-sensitivity in prediction. This enables the sliding window conditional probability model to take into account both short-term disturbances and long-term trends, achieving high-precision, lane-level quantifiable predictions of future link degradation probabilities.

[0052] S2.6. Calculate the first conditional probability value using the end time of the first monitoring window as the prediction starting point, calculate the second conditional probability value using the end time of the second monitoring window as the prediction starting point, and calculate the third conditional probability value using the end time of the third monitoring window as the prediction starting point.

[0053] Furthermore, the end time of the first monitoring window is used as the prediction starting point. The per-lane channel impulse response change rate and link health index corresponding to the end time of the first monitoring window are jointly input into the first conditional probability value expression. The first conditional probability value is calculated through a sliding window conditional probability model, reflecting the possibility of the link evolving into a deteriorated state after the current time. Subsequently, the end time of the second monitoring window is used as the prediction starting point. The per-lane channel impulse response change rate and time decay factor corresponding to the end time of the second monitoring window are jointly input into the second conditional probability value expression, calculating the second conditional probability value, reflecting the deterioration trend at the intermediate time position. Finally, the end time of the third monitoring window is used as the prediction starting point. The per-lane channel impulse response change rate corresponding to the end time of the third monitoring window is input into the third conditional probability value expression, calculating the third conditional probability value, reflecting the deterioration trend in the latter part of the prediction horizon. By constructing the conditional probability calculation results for three different prediction starting points, the change relationship over continuous time is preserved in the probability expression process, resulting in the first, second, and third conditional probability values.

[0054] Specifically, a conditional probability chain is constructed using multiple prediction starting points to enable continuous evolutionary characteristics in future link state judgments. For example, when the per-lane channel impulse response change rate shows signs of increase at the end of the first monitoring window, and subsequent monitoring windows maintain the same trend, the first conditional probability value can reflect potential risks in advance. When the change trend further intensifies at the end of the second monitoring window, the second conditional probability value can reflect the degradation evolution process. When the same development direction is maintained at the end of the third monitoring window, the third conditional probability value can reflect the trend continuation characteristic. Compared with a one-time probability calculation based solely on the current state, calculating conditional probability values ​​separately using multiple prediction starting points transforms link state judgment from static to time-continuous judgment. This preserves the evolutionary information of the per-lane channel impulse response change rate at different stages in the probability derivation process, thereby improving the ability to characterize future link degradation trends and obtaining the first, second, and third conditional probability values.

[0055] The expression for the first conditional probability value is: ; in, This is the first conditional probability value. The per-lane channel impulse response change rate at the end of the first monitoring window. For the link health index, This is the Sigmoid activation function.

[0056] The expression for the second conditional probability value is: ; in, This is the second conditional probability value. The per-lane channel impulse response change rate at the end of the second monitoring window. This is the time decay factor.

[0057] The expression for the third conditional probability value is: ; in, This is the third conditional probability value. This represents the rate of change of the per-lane channel impulse response at the end of the third monitoring window.

[0058] The final expression for the probability of future link degradation is: ; in, This represents the probability of future link degradation.

[0059] S2.7. The first conditional probability value, the second conditional probability value and the third conditional probability value are accumulated and merged according to the time decay factor to obtain the future link degradation probability and generate a link degradation trend prediction report.

[0060] Furthermore, degradation trend information from different prediction starting points is merged into a single future link degradation probability. During the accumulation and merging process, the time decay factor can attenuate the influence weight of distant monitoring windows, highlight the decision contribution of recent monitoring windows to the future link status, and retain early trend information to take into account continuity, thus obtaining a quantifiable future link degradation probability. By forming a link degradation trend prediction report from the accumulated and merged future link degradation probabilities, the potential degradation risk of each lane can be quantitatively expressed during continuous monitoring, achieving the operational goal of link degradation trend prediction.

[0061] S3. When the degradation probability and per-lane channel impulse response change in the link degradation trend prediction report meet the preset trigger conditions, the degraded lane is set to the PROBATION state, asymmetric interleaving remapping is performed on the remaining active lanes, a degradation execution instruction is generated, and the overall link state is set to ASYMMETRIC_ACTIVE.

[0062] S3.1. Input the probability of future link degradation into the log-likelihood ratio transformation unit to perform nonlinear compression to obtain the probability likelihood value; input the per-lane channel impulse response change rate into the arctangent transformation unit to perform nonlinear mapping to obtain the stable value of the change rate.

[0063] Furthermore, by using a log-likelihood ratio transformation, the probability of future link degradation is mapped from the probability space to the likelihood space, thus widening the distribution difference between the regions near the normal state and the regions near the degradation state in the probability of future link degradation, resulting in a probability likelihood value. The per-lane channel impulse response change rate is input into the arctangent transformation unit, and the per-lane channel impulse response change rate is nonlinearly mapped through the arctangent transformation, so that the larger rate of change increases gradually converge, while the differences between smaller rates of change are preserved, resulting in a stable rate of change value. The probability likelihood value is used to characterize the risk intensity of the probability of future link degradation at the decision level, and the stable rate of change value is used to characterize the stability of the per-lane channel impulse response change rate at the state level, completing a unified scale transformation between the probability of future link degradation and the per-lane channel impulse response change rate, resulting in the probability likelihood value and the stable rate of change value.

[0064] Specifically, the future link degradation probability and the per-lane channel impulse response change rate originate from different dimensions. The future link degradation probability is a trend prediction result, while the per-lane channel impulse response change rate is a real-time state result. Direct comparison between the two results in scale and sensitivity differences. Therefore, the log-likelihood ratio transformation unit converts the future link degradation probability into a probability likelihood value, which enhances the change characteristics of the future link degradation probability when it approaches the decision boundary. For example, when the future link degradation probability continues to rise but has not yet reached a significant degradation level, the probability likelihood value can reflect the risk accumulation trend in advance. At the same time, the arctangent transformation unit converts the per-lane channel impulse response change rate into a stable change rate value, which suppresses the impact of sudden fluctuations on subsequent judgments. For example, when a single perturbation causes an abnormal increase in the per-lane channel impulse response change rate, the stable change rate value will not be excessively amplified synchronously. Existing methods typically use the probability of future link degradation or the rate of change of per-lane channel impulse response to make threshold judgments. First, nonlinear unified processing of the probability space and state space is completed so that the probability likelihood value reflects the risk accumulation characteristics and the stable value of the rate of change reflects the state stability characteristics.

[0065] S3.2. The probability likelihood value is compared with the preset likelihood threshold to obtain a first comparison result, and the stable value of the rate of change is compared with the preset stable threshold to obtain a second comparison result. When both the first comparison result and the second comparison result indicate that they exceed the preset normal range, the triggering condition is determined to be met.

[0066] Furthermore, the stable value of the rate of change is compared with a preset stability threshold to determine whether the stable value exceeds the allowable range, resulting in a second comparison result. The first and second comparison results are then logically combined for judgment. When both indicate that they exceed the preset normal range, the trigger condition is deemed met, meaning that the current link state is considered to have a potential risk of lane degradation, requiring subsequent active intervention. This process ensures that the dual constraints of the future link degradation probability and the per-lane channel impulse response rate of change are simultaneously assessed, achieving accurate trigger identification of lane degradation.

[0067] Specifically, by comparing the magnitudes of two thresholds, the trend prediction results and real-time status results are jointly determined. This ensures that the triggering condition depends not only on the future link degradation probability but also on the stability of the per-lane channel impulse response change rate. For example, when the future link degradation probability is high but the per-lane change rate is still within a stable range, degradation is not triggered, thus avoiding false alarms. Conversely, when the per-lane change rate increases and the future link degradation probability also exceeds the threshold, potential risks can be accurately captured. By balancing sensitivity and stability through dual constraints, the triggering condition becomes more reliable and accurate, providing a highly credible basis for determining proactive link degradation and asymmetric interleaving remapping, ensuring the effectiveness of lane-level risk identification and intervention.

[0068] S3.3 Input the determination result of the trigger condition being met into the state machine, retrieve the lane index field corresponding to the trigger condition in the link degradation trend prediction report, decode the lane index field into the physical channel number, and write the physical channel number into the address mapping table of the degradation lane register.

[0069] Furthermore, when the trigger condition is met, the determination result of the trigger condition is input into the state machine, which then initiates the degraded lane location process based on the trigger event. The state machine accesses the link degradation trend prediction report and retrieves the lane index field corresponding to the trigger condition from the report. The lane index field is used to identify the target lane that generates anomalies in the probability of future link degradation and per-lane channel impulse response change rate. The lane index field is decoded to convert the logical-level lane identifier into a physical channel number that can directly correspond to the actual transmission resource. After decoding, the physical channel number is written into the address mapping table of the degraded lane register, establishing a correspondence between the physical channel number and subsequent state control operations. The state machine completes the entire process from the trigger condition being met to physical channel location, enabling the risk judgment result in the link degradation trend prediction report to be converted into a specific, operable physical channel object, resulting in the physical channel number written into the address mapping table of the degraded lane register.

[0070] Specifically, the analysis results in the link degradation trend prediction report are transformed into directly controllable physical channel numbers. The mapping relationship between the lane index field and the physical channel number directly links the prediction results to actual transmission resources. For example, when multiple lanes simultaneously exhibit an increased probability of future link degradation, the state machine can quickly determine the corresponding physical channel number based on the lane index field. When the same physical channel continuously exhibits abnormal per-lane channel impulse response rate changes, the physical channel number can be continuously stored in the address mapping table of the degradation lane register. Through the correlation processing between the state machine, lane index field, physical channel number, and address mapping table of the degradation lane register, the direct conversion of risk identification results to physical control objects is achieved. This eliminates the need for re-performing location operations during subsequent PROBATION state writing processes, improving the continuity and consistency of the degradation lane management process.

[0071] S3.4. Perform an atomic write operation to the state storage unit corresponding to the degraded lane through the state machine control bus to write the predefined PROBATION status code, and generate a status update instruction for the physical channel number and PROBATION status code.

[0072] Furthermore, the state machine has already written the physical channel number into the address mapping table of the degraded lane register, and locates the state storage unit of the corresponding degraded lane according to the physical channel number. The state machine initiates a state update request to the corresponding state storage unit through the control bus, writes the predefined PROBATION status code as the target content, and completes the state update process using atomic write operations, so that the state information in the state storage unit is switched in a single write process, avoiding intermediate states during the state update. After the predefined PROBATION status code is written, the state machine reads the physical channel number and the predefined PROBATION status code, combines the physical channel number and the predefined PROBATION status code to form a state update instruction, where the physical channel number is used to identify the target channel being processed, and the PROBATION status code is used to identify the current state level of the target channel, generating a state update instruction for the physical channel number and the PROBATION status code.

[0073] Specifically, when the probability of future link degradation already exhibits risk characteristics but has not yet reached a failure state, the corresponding physical channel number is switched to the controlled state corresponding to the PROBATION status code. Existing methods typically employ a binary switching approach between normal and failed states. When link quality deteriorates, service transmission is directly suspended. An intermediate state hierarchy is constructed using predefined PROBATION status codes, allowing physical channel numbers to retain manageability while preserving subsequent recovery capabilities. For example, if the probability of future link degradation continues to increase but the link still has transmission capacity, the PROBATION status code can pre-mark risky channels; when the link recovers subsequently, the physical channel number can still participate in the recovery process without needing to re-establish the mapping relationship. Simultaneously, atomic write operations are used to complete the state switching, ensuring that the correspondence between the physical channel number and the PROBATION status code is established during a single state update, avoiding inconsistencies in the state storage unit during updates. Through the coordination of the state machine, control bus, state storage unit, atomic write operations, and PROBATION status codes, the link degradation prediction results can be directly converted into executable state control results.

[0074] S3.5 Input the status update instruction with physical channel number and PROBATION status code into the remapping logic, extract the physical channel number field from the status update instruction, match the physical channel number in the parallel search array of the activelane list, and perform bubble sort on the remaining activelane identifiers according to the physical channel index number.

[0075] Furthermore, the status update command of the physical channel number and PROBATION status code is input into the remapping logic. The remapping logic parses the status update command and extracts the physical channel number field. Then, the physical channel number field is input into the parallel search array of the activelane list. A parallel matching operation is performed in the activelane list to determine whether the corresponding physical channel number is currently in the activelane list. After the parallel search array of the activelane list completes the matching, the activelane identifier corresponding to the physical channel number is removed from the activelane list, so that the corresponding physical channel will no longer participate in resource allocation as an effective bearer channel in the subsequent remapping process. After the removal process is completed, the remaining activelane identifiers are bubble sorted according to the physical channel index number. Through continuous comparison and swapping operations, the remaining activelane identifiers are arranged in a unified order to form a continuous and ordered activelane sequence. The remaining activelane identifiers are bubble sorted according to the physical channel index number.

[0076] Specifically, the risk identification results are directly applied to the activelane list, rather than undergoing secondary filtering in the subsequent data scheduling phase. At the remapping logic entry point, the parallel search array of the activelane list is used to locate physical channel numbers, causing physical channels in the PROBATION status code state to exit the activelane resource set early. For example, when multiple physical channel numbers simultaneously enter the PROBATION status code state, the parallel search array of the activelane list can synchronously complete the matching and location, avoiding processing delays caused by individual searches. Furthermore, when subsequent asymmetric interleaving remapping requires reallocation of bearer relationships, the already filtered activelane list retains only available channels, eliminating the need for further channel status checks during the remapping process. A bubble sort method based on physical channel index numbers is used to establish a deterministic order for the remaining activelane identifiers, ensuring consistent channel arrangement results for remapping processes triggered at different times, thus avoiding fluctuations in remapping results due to changes in the activelane arrangement order. Through the coordination of the remapping logic, the parallel search array of the activelane list, and the bubble sort method based on physical channel index numbers, risk channel stripping and ordered processing of available channels are achieved.

[0077] S3.6. Based on the predefined asymmetric interleaving rules, the sorted activelane identifiers are reassigned virtual channel numbers to generate an asymmetric interleaving remapping table.

[0078] Furthermore, virtual channel numbers are reassigned to the sorted active lane identifiers according to predefined asymmetric interleaving rules. These predefined asymmetric interleaving rules are determined in advance during link initialization based on scenarios involving changes in the number of active lanes, virtual channel continuity requirements, and link bandwidth requirements. These rules define the correspondence between virtual channel numbers and active lane identifiers. When the sorted active lane identifiers enter the remapping process, new virtual channel numbers are assigned to each active lane identifier sequentially according to the predefined asymmetric interleaving rules, redistributing the virtual channel mapping relationships originally handled by degraded lanes to the remaining active lanes. Subsequently, a mapping relationship is established between the reassigned virtual channel numbers and their corresponding active lane identifiers, and the mapping results are recorded in order of virtual channel number, forming a complete asymmetric interleaving remapping table. This asymmetric interleaving remapping table records the latest correspondence between each virtual channel number and an active lane identifier, generating the asymmetric interleaving remapping table.

[0079] S3.7. Convert the asymmetric interleaving remapping table into a hardware register configuration sequence, and encapsulate the hardware register configuration sequence into a downgrade execution instruction.

[0080] Furthermore, the correspondence between each virtual channel number and the activelane identifier is converted into corresponding register write entries, and a hardware register configuration sequence is generated according to the register address order. This allows the logical mapping relationship in the asymmetric interleaving remapping table to be converted into directly executable configuration content. After the hardware register configuration sequence is generated, it is combined and encapsulated with the execution identifier field, configuration activation identifier field, and state switching identifier field to form a unified format of degraded execution instructions. The degraded execution instructions carry all the configuration content corresponding to the asymmetric interleaving remapping table and serve as the basis for subsequent link state switching and mapping relationship updates. The hardware register configuration sequence is encapsulated into degraded execution instructions.

[0081] S3.8 Send the downgrade execution instruction to the physical coding sublayer and wait for the physical coding sublayer to return the remapping completion signal. Write the remapping completion signal and the predefined ASYMMETRIC_ACTIVE status code into the link global status register and update the output value of the link global status register to ASYMMETRIC_ACTIVE.

[0082] Furthermore, after the downgrade execution instruction is sent to the physical coding sublayer, the physical coding sublayer performs data distribution path adjustment and interleaving address update operations for the corresponding lane according to the asymmetric interleaving remapping information in the downgrade execution instruction, so that the data flow is reorganized according to the new asymmetric interleaving remapping relationship. During the execution process, the physical coding sublayer continuously detects whether the mapping relationship of each lane has been switched and performs consistency verification on the data transmission path after remapping. When all target lanes have completed remapping and the interleaving relationship has been updated, the physical coding sublayer returns a remapping completion signal. Then, the remapping completion signal and the predefined ASYMMETRIC_ACTIVE status code are written together into the link global status register, so that the link global status register records the current remapping execution result and the current link running status at the same time. After the writing is completed, the link global status register refreshes the current status information and updates the output value of the link global status register to ASYMMETRIC_ACTIVE to indicate that the link has completed asymmetric interleaving remapping and entered the corresponding running status.

[0083] Specifically, the remapping completion signal is used as the trigger condition for state updates, ensuring that state changes are based on the actual completion of remapping. For example, when a downgrade execution instruction has been sent to the physical coding sublayer but some lanes are still in the process of mapping switching, the link global state register maintains its original state information. Alternatively, when all remapping operations are completed, the remapping completion signal directly proves that the current data path is running stably, and the predefined ASYMMETRIC_ACTIVE status code is then written into the link global state register, keeping the state information synchronized with the actual operating state. Through the sequential association between the remapping completion signal, the predefined ASYMMETRIC_ACTIVE status code, and the link global state register, a consistent mapping between state confirmation and execution results is achieved, ensuring that the output value of the link global state register accurately represents the current asymmetric interleaving operating state.

[0084] S4. Adjust the traffic scheduling parameters on the sending side according to the downgrade execution instruction, and generate the adjusted sending parameters.

[0085] S4.1 Map the correspondence between virtual channels and physical channels in the asymmetric interleaving remapping table to virtual channel bandwidth quota parameters, input the virtual channel bandwidth quota parameters into the token bucket controller to update the token generation rate of each virtual channel, and generate the updated token generation rate configuration.

[0086] Furthermore, based on the number of virtual channels currently carried by each physical channel and their corresponding mapping positions, the virtual channel bandwidth quota parameters for each virtual channel are obtained. These parameters characterize the proportion of transmission resources available to each virtual channel under the current asymmetric interleaving remapping condition. The virtual channel bandwidth quota parameters are input to the token bucket controller, which then reconfigures the token generation rate for each virtual channel according to these parameters, ensuring that each virtual channel receives a transmission permission rate that matches the current mapping relationship. When a virtual channel is mapped to a physical channel with higher carrying capacity, the token bucket controller allocates a higher token generation rate. When multiple virtual channels share the same physical channel resource, the token bucket controller rebalances the corresponding token generation rates based on the virtual channel bandwidth quota parameters, resulting in an updated token generation rate configuration.

[0087] S4.2 Input the updated token generation rate configuration into the weighted round-robin scheduler, rearrange the scheduling priority of the virtual channel according to the token generation rate, map the rearranged scheduling priority to the time-division multiplexing time slice allocation table, and generate a time slice allocation table update instruction.

[0088] Furthermore, the updated token generation rate configuration is input into the weighted round-robin scheduler. The weighted round-robin scheduler reads the token generation rate corresponding to each virtual channel and re-acquires the scheduling weight of each virtual channel based on the token generation rate. The scheduling priority of the virtual channels is rearranged according to the scheduling weight, so that virtual channels with higher token generation rates have more scheduling opportunities, and virtual channels with lower token generation rates have relatively fewer scheduling opportunities. After the scheduling priority is rearranged, the scheduling priority is converted into the corresponding time-division multiplexing time slice allocation relationship, and a time-division multiplexing time slice allocation table is generated according to the time order. The time-division multiplexing time slice allocation table is encoded to form a time slice allocation table update instruction.

[0089] S4.3 Write the time slice allocation table update instruction into the configuration register of the transmitting side scheduler, and map the output of the configuration register to the adjusted transmitting parameters.

[0090] Furthermore, the time-division multiplexing time-slice allocation relationship in the time-slice allocation table update instruction is converted into configuration content that can be directly executed by the transmitting-side scheduler. After receiving the time-slice allocation table update instruction, the configuration register overwrites and updates the original scheduling configuration, and reorganizes the transmission order, transmission timing, and resource occupancy relationship of the virtual channels according to the new time-slice allocation relationship. After the configuration register completes the update, the scheduling configuration currently stored in the configuration register is output to the execution path of the transmitting-side scheduler, and the corresponding configuration content is mapped to the adjusted transmission parameters. The adjusted transmission parameters include the current virtual channel transmission order, transmission resource allocation relationship, and time-slice occupancy relationship, generating the adjusted transmission parameters.

[0091] S5 monitors the link degradation trend prediction report, triggers the UNFOLD operation to restore the original lane mapping rules and full bandwidth transmission, and generates a restoration completion signal.

[0092] S5.1 Input the future link degradation probability into the inverse log-likelihood ratio transformation unit to perform nonlinear restoration to obtain the recovery confidence. Compare the recovery confidence with the preset recovery confidence threshold. When the comparison results of multiple consecutive monitoring windows all indicate that the recovery confidence is lower than the preset recovery confidence threshold, the judgment result that the recovery condition is met is obtained.

[0093] Furthermore, the future link degradation probability in the monitoring link degradation trend prediction report is input into the inverse log-likelihood ratio transformation unit for nonlinear restoration processing. This transforms the risk representation, previously compressed by the log-likelihood ratio transformation, back into a recovery confidence level with significance for recovery determination. The recovery confidence level characterizes the degree to which the current link state deviates from the degradation state and reverts to a normal state. Subsequently, the recovery confidence level is compared with a preset recovery confidence threshold to obtain the corresponding comparison results. To avoid misjudgment caused by short-term fluctuations, the comparison results in multiple consecutive monitoring windows are continuously tracked. When multiple consecutive monitoring windows indicate that the recovery confidence level is lower than the preset recovery confidence threshold, it is determined that the link state has broken away from the continuous degradation trend and entered a stable recovery phase, thus obtaining a determination result that the recovery condition has been met.

[0094] S5.2 Input the judgment result of the recovery condition being met into the state machine, load the asymmetric interleaving remapping table into the address generation unit, and have the address generation unit call the factory-preset original lane mapping rule lookup table, write the data in the original lane mapping rule lookup table into the interleaving address register group of the remapping logic in parallel, and generate the UNFOLD operation instruction.

[0095] Furthermore, the judgment result of the recovery condition being met is input into the state machine, which then initiates the mapping recovery process. The state machine reads the currently saved asymmetric interleaving remapping table and loads it into the address generation unit, enabling the address generation unit to obtain the current link mapping status information. Subsequently, the address generation unit calls the factory-preset original lane mapping rule lookup table to obtain the standard mapping relationship corresponding to each lane in the initial state of the link. After obtaining the mapping content in the original lane mapping rule lookup table, the data in the original lane mapping rule lookup table is written in parallel into the interleaving address register group of the remapping logic, so that the current asymmetric interleaving mapping relationship is gradually replaced with the mapping relationship corresponding to the original lane mapping rule. When the interleaving address register group is updated, the state machine generates the UNFOLD operation instruction based on the completed register content.

[0096] S5.3. Convert the UNFOLD operation instruction into a hardware register reset sequence, send the hardware register reset sequence to the physical coding sublayer to restore the original lane mapping rules and full bandwidth transmission, the physical coding sublayer returns the operation completion status word, and encode the operation completion status word into a recovery completion signal.

[0097] Furthermore, the UNFOLD operation instruction is converted into a hardware register reset sequence, enabling the mapping recovery content in the UNFOLD operation instruction to be converted into executable register control information. The hardware register reset sequence is then sent to the physical coding sublayer, which sequentially restores the interleaving address configuration, lane mapping relationship, and corresponding data distribution path according to the hardware register reset sequence, returning the link to the working state corresponding to the original lane mapping rule. After restoring the original lane mapping rule, the physical coding sublayer continues to restore the full-bandwidth transmission state, reopening the transmission resources previously restricted due to the degradation operation. When the entire restoration process is complete, the physical coding sublayer returns an operation completion status word, which is then encoded to generate a restoration completion signal.

[0098] This embodiment also provides an interconnected data transmission control system based on a 400G high-speed optical module, including: an update module, which generates a link basic status report of the per-lane channel impulse response baseline after the optical module completes physical access and link initialization, and collects link operation data to update the link basic status report.

[0099] The prediction module embeds predefined perturbation patterns in the idle gaps of the 400GPAM4 data stream, extracts the channel impulse response change rate of each lane, and combines it with the link health index to predict the probability of future link degradation, generating a link degradation trend prediction report.

[0100] The mapping module sets the degraded lane to PROBATION state when the degradation probability and per-lane channel impulse response change in the link degradation trend prediction report meet the preset trigger conditions. It then performs asymmetric interleaving remapping on the remaining active lanes, generates a degradation execution instruction, and sets the overall link state to ASYMMETRIC_ACTIVE.

[0101] The adjustment module adjusts the traffic scheduling parameters on the sending side according to the downgrade execution instruction, and generates the adjusted sending parameters.

[0102] The transmission module monitors the link degradation trend prediction report, triggers the UNFOLD operation to restore the original lane mapping rules and full bandwidth transmission, and generates a recovery completion signal.

[0103] This embodiment also provides a computer device applicable to the interconnect data transmission control method based on a 400G high-speed optical module, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the interconnect data transmission control method based on a 400G high-speed optical module as proposed in the above embodiment.

[0104] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0105] This embodiment also provides a storage medium storing a computer program. When executed by a processor, the program implements the interconnect data transmission control method based on a 400G high-speed optical module as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0106] In summary, this invention achieves continuous tracking of the operational status of each lane by establishing and updating the link basic status report of the per-lane channel impulse response baseline; it embeds predefined perturbation patterns in the idle gaps of the 400G PAM4 data stream, extracts the per-lane channel impulse response change rate, and combines it with the link health index to predict the probability of future link degradation, thus detecting potential degradation trends before actual link failures occur; when preset triggering conditions are met, the degraded lane is set to the PROBATION state, and asymmetric interleaving remapping is performed on the remaining active lanes to achieve proactive avoidance of risky channels by service traffic; the transmission-side traffic scheduling parameters are adjusted according to the remapping results to improve link resource utilization efficiency; after the link recovers and stabilizes, the original lane mapping rules and full bandwidth transmission are restored through the UNFOLD operation, achieving early prediction, proactive degradation, and automatic recovery of link degradation, thereby improving the transmission stability and continuity of the 400G high-speed optical module interconnection link.

[0107] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for controlling interconnected data transmission based on a 400G high-speed optical module, characterized in that: include, After the optical module completes physical access and link initialization, a basic link status report of the per-lane channel impulse response baseline is generated, and link operation data is collected to update the basic link status report. Predefined perturbation patterns are embedded in the Idle gaps of the 400GPAM4 data stream. The channel impulse response change rate of each lane is extracted, and the link health index is combined to predict the probability of future link degradation and generate a link degradation trend prediction report. When the degradation probability and per-lane channel impulse response change in the link degradation trend prediction report meet the preset trigger conditions, the degraded lane is set to the PROBATION state. Specifically, this includes: inputting the determination result of the trigger condition being met into the state machine, retrieving the lane index field corresponding to the trigger condition in the link degradation trend prediction report, decoding the lane index field into the physical channel number, writing the physical channel number into the address mapping table of the degraded lane register, performing an atomic write operation to the state storage unit corresponding to the degraded lane through the state machine control bus to write the predefined PROBATION state code, and generating a state update instruction for the physical channel number and the PROBATION state code. Perform asymmetric interleaving remapping on the remaining activelane to generate degraded execution instructions; Setting the overall link status to ASYMMETRIC_ACTIVE specifically includes: sending the downgrade execution instruction to the physical coding sublayer and waiting for the physical coding sublayer to return the remapping completion signal; writing the remapping completion signal and the predefined ASYMMETRIC_ACTIVE status code into the link global status register; and updating the output value of the link global status register to ASYMMETRIC_ACTIVE. Adjust the traffic scheduling parameters on the sending side according to the downgrade execution instruction, and generate the adjusted sending parameters; The monitoring link degradation trend prediction report triggers the UNFOLD operation to restore the original lane mapping rules and full bandwidth transmission, generating a recovery completion signal; When the probability of link degradation continues to increase but the link still has transmission capability, the PROBATION status code can mark the risky channel in advance. When the link status recovers later, the physical channel number can still participate in the recovery process without re-establishing the mapping relationship. The state switch is completed by atomic write operation, so that the correspondence between the physical channel number and the PROBATION status code is established in a single state update process, avoiding the problem of state inconsistency in the state storage unit during the update. Through the cooperation between the state machine, control bus, state storage unit, atomic write operation and PROBATION status code, the link degradation prediction result can be directly converted into an executable state control result. After the downgrade execution instruction is sent to the physical coding sublayer, the physical coding sublayer performs data distribution path adjustment and interleaving address update operations for the corresponding lane according to the asymmetric interleaving remapping information in the downgrade execution instruction. This reorganizes the data flow according to the new asymmetric interleaving remapping relationship. During the execution process, the physical coding sublayer continuously monitors whether the mapping relationship of each lane has been switched and performs consistency verification on the remapped data transmission path. When all target lanes have completed remapping and the interleaving relationship has been updated, the physical coding sublayer returns a remapping completion signal. The remapping completion signal and the predefined ASYMMETRIC_ACTIVE status code are written together into the link global status register. This allows the link global status register to record both the current remapping execution result and the current link operating status. After writing, the link global status register refreshes the current status information and updates the output value of the link global status register to ASYMMETRIC_ACTIVE to indicate that the link has completed asymmetric interleaving remapping and entered the corresponding operating state.

2. The interconnection data transmission control method based on a 400G high-speed optical module as described in claim 1, characterized in that: The updated link basic status report includes, The per-lane channel impulse response baseline is obtained using the link training sequence of the physical medium additional channel, and the per-lane channel impulse response baseline is written into the link basic status report; Using the per-lane channel impulse response baseline in the link basic status report as a reference, the instantaneous value of the per-lane channel impulse response in the link operation data is collected, and the difference between the instantaneous value of the per-lane channel impulse response and the per-lane channel impulse response baseline is obtained to obtain the change in the per-lane channel impulse response. Update the per-lane channel impulse response change to the corresponding per-lane field in the link basic status report, overwriting the original per-lane channel impulse response baseline difference record.

3. The interconnection data transmission control method based on a 400G high-speed optical module as described in claim 2, characterized in that: The link degradation trend prediction report includes, A predefined perturbation pattern is embedded in the Idle gap of the 400GPAM4 data stream. Frequency domain cross-correlation is performed on the received sampling sequence of the predefined perturbation pattern to obtain the instantaneous value of the per-lane channel impulse response. The instantaneous value of the per-lane channel impulse response is compared with the baseline of the per-lane channel impulse response in the link basic status report by tap phase rotation correction. The amplitude change of each tap is normalized and the change contribution is allocated according to the energy proportion of the tap to obtain the per-lane channel impulse response change rate. Using the per-lane channel impulse response change rate of each of the five consecutive monitoring windows as input, the per-lane channel impulse response change rates from the first to the fifth monitoring windows are denoted as follows: to ,calculate and The difference is used to obtain the first gradient, and the calculation is performed. and The difference between the first and second gradients is used to obtain the second gradient, and the difference between the second gradient and the first gradient is used to obtain the per-lane channel impulse response curvature. The error statistics contained in the link operation data in the link basic status report are mapped to the error impact coefficient, and the equalizer coefficient offset is mapped to the offset impact coefficient. The error impact coefficient and the offset impact coefficient are nonlinearly modulated using the per-lane channel impulse response change curvature to obtain the link health index. Using the per-lane channel impulse response change rate as a time series input, and the link health index as a distribution shape adjustment factor, a non-uniform stretching transformation is performed on the input sample set of the sliding window conditional probability model, and the prediction horizon is set to three consecutive monitoring windows. The first conditional probability value is calculated using the end time of the first monitoring window as the prediction starting point, the second conditional probability value is calculated using the end time of the second monitoring window as the prediction starting point, and the third conditional probability value is calculated using the end time of the third monitoring window as the prediction starting point. The first conditional probability value, the second conditional probability value, and the third conditional probability value are accumulated and merged according to the time decay factor to obtain the future link degradation probability, and a link degradation trend prediction report is generated.

4. The interconnection data transmission control method based on a 400G high-speed optical module as described in claim 3, characterized in that: The downgrade execution instructions include, The probability of future link degradation is input into the log-likelihood ratio transform unit and nonlinearly compressed to obtain the probability likelihood value. The per-lane channel impulse response change rate is input into the arctangent transform unit and nonlinearly mapped to obtain the stable value of the change rate. The probability likelihood value is compared with the preset likelihood threshold to obtain the first comparison result, and the stable value of the rate of change is compared with the preset stable threshold to obtain the second comparison result. When both the first comparison result and the second comparison result indicate that they exceed the preset normal range, the triggering condition is determined to be met. Input the determination result of the trigger condition being met into the state machine, retrieve the lane index field corresponding to the trigger condition in the link degradation trend prediction report, decode the lane index field into the physical channel number, and write the physical channel number into the address mapping table of the degradation lane register. Atomic write operations are performed on the state storage unit corresponding to the degraded lane via the state machine control bus to write the predefined PROBATION status code, and a status update instruction for the physical channel number and PROBATION status code is generated. Input the status update instruction with physical channel number and PROBATION status code into the remapping logic, extract the physical channel number field from the status update instruction, match the physical channel number in the parallel search array of the activelane list, and perform bubble sort on the remaining activelane identifiers according to the physical channel index number. Based on predefined asymmetric interleaving rules, the sorted activelane identifiers are reassigned virtual channel numbers to generate an asymmetric interleaving remapping table; The asymmetric interleaving remapping table is converted into a hardware register configuration sequence, and the hardware register configuration sequence is encapsulated into a degraded execution instruction.

5. The interconnection data transmission control method based on a 400G high-speed optical module as described in claim 1, characterized in that: The adjusted transmission parameters include, The correspondence between virtual channels and physical channels in the asymmetric interleaving remapping table is mapped to virtual channel bandwidth quota parameters. The virtual channel bandwidth quota parameters are then input into the token bucket controller to update the token generation rate of each virtual channel, generating the updated token generation rate configuration. The updated token generation rate configuration is input into the weighted round-robin scheduler. The scheduling priority of the virtual channel is rearranged according to the token generation rate. The rearranged scheduling priority is mapped to the time-division multiplexing time slice allocation table, and a time slice allocation table update instruction is generated. Write the time slice allocation table update instruction into the configuration register of the transmitting side scheduler, and map the output of the configuration register to the adjusted transmitting parameters.

6. The interconnection data transmission control method based on a 400G high-speed optical module as described in claim 5, characterized in that: The recovery completion signal includes, The probability of future link degradation is input into the inverse log-likelihood ratio transformation unit for nonlinear restoration to obtain the recovery confidence. The recovery confidence is compared with the preset recovery confidence threshold. When the comparison results of multiple consecutive monitoring windows all indicate that the recovery confidence is lower than the preset recovery confidence threshold, the judgment result that the recovery condition is met is obtained. The result of the determination that the recovery condition is met is input into the state machine, the asymmetric interleaving remapping table is loaded into the address generation unit, the address generation unit calls the factory-preset original lane mapping rule lookup table, and writes the data in the original lane mapping rule lookup table into the interleaving address register group of the remapping logic in parallel to generate the UNFOLD operation instruction. The UNFOLD operation instruction is converted into a hardware register reset sequence, which is then sent to the physical coding sublayer to restore the original lane mapping rules and full-bandwidth transmission. The physical coding sublayer returns an operation completion status word, which is then encoded into a recovery completion signal.

7. An interconnect data transmission control system based on a 400G high-speed optical module, based on the interconnect data transmission control method based on a 400G high-speed optical module as described in any one of claims 1 to 6, characterized in that: include, The update module generates a basic link status report of the per-lane channel impulse response baseline after the optical module completes physical access and link initialization, and collects link operation data to update the basic link status report. The prediction module embeds a predefined perturbation pattern in the Idle gap of the 400GPAM4 data stream, extracts the channel impulse response change rate of each lane, and combines it with the link health index to predict the probability of future link degradation and generate a link degradation trend prediction report. The trigger module sets the degraded lane to the PROBATION state when the degradation probability and per-lane channel impulse response change in the link degradation trend prediction report meet the preset trigger conditions. Specifically, it inputs the determination result of the trigger condition into the state machine, retrieves the lane index field corresponding to the trigger condition in the link degradation trend prediction report, decodes the lane index field into the physical channel number, writes the physical channel number into the address mapping table of the degraded lane register, performs an atomic write operation to the state storage unit corresponding to the degraded lane through the state machine control bus to write the predefined PROBATION state code, and generates a state update instruction for the physical channel number and the PROBATION state code. The mapping module performs asymmetric interleaving remapping on the remaining activelanes to generate degraded execution instructions; The execution module sets the overall link status to ASYMMETRIC_ACTIVE, specifically including: sending the downgrade execution instruction to the physical coding sublayer and waiting for the physical coding sublayer to return the remapping completion signal, writing the remapping completion signal and the predefined ASYMMETRIC_ACTIVE status code into the link global status register, and updating the output value of the link global status register to ASYMMETRIC_ACTIVE. The adjustment module adjusts the traffic scheduling parameters on the sending side according to the downgrade execution instruction, and generates the adjusted sending parameters; The transmission module monitors the link degradation trend prediction report, triggers the UNFOLD operation to restore the original lane mapping rules and full bandwidth transmission, and generates a recovery completion signal.

Images