Method and device for adjusting adaptive weight based on equalizer error, and equalization system

Abstract

The invention discloses a method and a device for adjusting an adaptive weight based on an equalizer error, and an equalization system. The method comprises the following steps: acquiring a decision error output by a decision device in each period; determining the error number of the judgment errors meeting the condition in a preset period number; and configuring an adaptive weight for the equalizer according to the error number, so that the tap coefficient of the equalizer is converged according to the adaptive weight. According to the method, the convergence speed and the steady-state error can be balanced.

Description

Technical Field

[0001] The present invention relates to the field of signal processing technology, and in particular to a method and device for adjusting adaptive weights based on equalizer errors, and an equalization system. Background Art

[0002] In serial transmission systems (such as Ethernet), adaptive equalizer technology, such as forward feedback equalizer and decision feedback equalizer, is usually used to deal with distortion and inter-symbol interference during signal transmission. When data signals are transmitted through cables, the dispersion characteristics of the cables will cause the signals to be deformed, and the degree of deformation will vary depending on the signal spectrum and the length of the transmission medium. To ensure the complete transmission and accurate reception of the signal, the signal received at the receiving end must be properly compensated.

[0003] The adaptive equalizer can flexibly adapt to changes in channel characteristics by measuring channel characteristics in real time and adjusting its coefficients accordingly, thereby effectively eliminating inter-symbol interference. The adaptive equalizer based on the LMS (Least Mean Square) algorithm is a very effective solution. The design goal of the LMS algorithm is to minimize the mean square estimation error. It uses an iterative method to continuously adjust the tap coefficients of the equalizer (i.e., the weight of the filter) to achieve the best compensation for the received signal.

[0004] In the LMS algorithm, the adaptation weight is a crucial parameter that directly affects the convergence speed and steady-state error of the algorithm. In the design of a constant-weight adaptive equalizer, there is a trade-off between the convergence speed and the steady-state error: a larger adaptation weight can accelerate the convergence process, but may increase the steady-state error; while a smaller adaptation weight can reduce the steady-state error, but will slow down the convergence speed. On the contrary, if the adaptation weight is set too large, even if the algorithm is close to convergence, it may cause the system to oscillate or even diverge due to excessive weight updates. If the adaptation weight is set too small, the algorithm may require more iterations to converge, thereby increasing the consumption of computing resources. Summary of the invention

[0005] The present invention provides a method and device for adjusting adaptive weights based on equalizer errors, and an equalization system, which can achieve a balance between convergence speed and steady-state error.

[0006] In one aspect of the present invention, a method for adjusting adaptive weights based on equalizer errors is provided. The method comprises: obtaining decision errors output by a decision device in each cycle; determining the number of decision errors that meet the conditions within a preset number of cycles; and configuring adaptive weights for the equalizer according to the number of errors, so that the tap coefficients of the equalizer are adaptively converged according to the adaptive weights.

[0007] In another aspect of the present invention, a device for adjusting adaptive weights based on equalizer errors is provided. The device comprises: an error acquisition module configured to acquire decision errors output by a decision device in each cycle; a quantity detection module configured to determine the number of decision errors that meet the conditions within a preset number of cycles; and a weight configuration module configured to configure adaptive weights for the equalizer according to the number of errors, so that the tap coefficients of the equalizer are adaptively converged according to the adaptive weights.

[0008] In another aspect of the present invention, an equalization system is provided, which includes the above-mentioned device for adjusting adaptive weights based on equalizer errors, and a decision device and an equalizer electrically coupled to the device.

[0009] According to the technical solution of the present invention, the decision error output by the decision device in each cycle is obtained, and the number of decision errors that meet the conditions within a preset number of cycles is counted to determine the adaptive working state of the current equalizer. The adaptive weight of the equalizer is configured according to the number of decision errors, so that the equalizer adjusts the tap coefficient according to the configured adaptive weight. Compared with the method of fixing the adaptive weight, dynamically adjusting the adaptive weight according to the adaptive working state of the equalizer can avoid the equalizer from causing system oscillation or even divergence due to excessive weight in the convergence state, and at the same time avoid the equalizer from increasing the consumption of computing resources due to the adaptive weight being set too small, thereby achieving a dynamic balance between the stability of the equalizer error and the convergence speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A waveform diagram of a decision error of a decision device at different adaptive stages; Figure 2 A schematic diagram of a decision of a decision device at different adaptive stages; Figure 3 A flowchart of a method for adjusting adaptive weights based on equalizer error according to an embodiment of the present invention; Figure 4 is a schematic diagram of a shift register according to an embodiment of the present invention; Figure 5 A block diagram of a device for adjusting adaptive weights based on equalizer error according to an embodiment of the present invention; Figure 6 FIG. 4 is an architecture diagram of a balancing system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0011] In order to explain the technical content, achieved objectives and effects of the present invention in detail, the following is an explanation in combination with the implementation modes and the accompanying drawings.

[0012] In the prior art, in the LMS algorithm, the adaptive weight is a crucial parameter that directly affects the convergence speed and steady-state error of the algorithm. In the design of a constant-weight adaptive equalizer, there is a trade-off between the convergence speed and the steady-state error: a larger adaptive weight can accelerate the convergence process, but may increase the steady-state error; while a smaller adaptive weight can reduce the steady-state error, but will slow down the convergence speed. On the contrary, if the adaptive weight is set too large, even if the algorithm is close to convergence, it may cause the system to oscillate or even diverge due to excessive weight updates. If the adaptive weight is set too small, the algorithm may require more iterations to converge, thereby increasing the consumption of computing resources.

[0013] In order to solve at least the above technical problems, the present disclosure provides a method for adjusting adaptive weights based on equalizer errors. According to the present disclosure, the decision error output by the decider in each cycle is obtained, and the number of decision errors that meet the conditions within a preset number of cycles is counted to determine the adaptive working state of the current equalizer. The adaptive weight of the equalizer is configured according to the number of decision errors, so that the equalizer adjusts the tap coefficient according to the configured adaptive weight. In this way, compared with the method of fixing the adaptive weight, according to the embodiment of the present disclosure, the adaptive weight can be dynamically adjusted according to the adaptive working state of the equalizer, and the system oscillation or even divergence caused by the excessive weight of the equalizer in the convergence state can be avoided. At the same time, the consumption of computing resources is increased due to the adaptive weight of the equalizer being set too small, thereby achieving a dynamic balance between the stability of the equalizer error and the convergence speed.

[0014] According to an embodiment of the present disclosure, the adaptive process of the adaptive equalizer is divided into an initial adaptive stage, a transition adaptive stage, and a stable adaptive stage according to the size of the decision error. Within a preset number of cycles, the number of errors of the first decision error greater than the first threshold and the number of errors of the second decision error less than the second threshold are respectively counted. If the number of errors of the first decision error is greater than the first numerical threshold, it can be determined that the current adaptive equalizer is in the initial adaptive stage, and the adaptive weight of the adaptive equalizer is configured as the weight value corresponding to the initial adaptive stage. If the number of errors of the second decision error is greater than the second numerical threshold, it can be determined that the current adaptive equalizer is in the stable adaptive stage, and the adaptive weight of the adaptive equalizer is configured as the weight value corresponding to the stable adaptive stage. If the number of errors of the first decision error is not greater than the first numerical threshold or the number of errors of the second decision error is not greater than the second numerical threshold, it can be determined that the current adaptive equalizer is in the transition adaptive stage, and the adaptive weight of the adaptive equalizer is configured as the weight value corresponding to the transition adaptive stage. In this way, different adaptive weights are dynamically switched according to the working state of the adaptive equalizer.

[0015] The adaptive equalizer usually realizes the adaptive adjustment of the tap coefficient based on the LMS algorithm. The specific formula of the LMS algorithm is: W(n+1)=W(n)+μ×e(n)×X(n), where W(n+1) represents the tap coefficient after iterative update, W(n) represents the tap coefficient before iterative update, μ represents the adaptive weight, e(n) represents the decision error, and X(n) represents the input signal. During the adaptive process of the adaptive equalizer, the absolute value of the decision error e changes from large to small. When the absolute value of the decision error approaches 0, the tap coefficient W approaches a stable value, which is considered to be converged. The adaptive weight μ is a step factor used to adjust the speed of tap coefficient update. From the above content, it can be seen that the tap coefficient of the adaptive equalizer needs to be operated with the input signal, and the decision error can only be obtained after the operation result and the decision error output by the decision device are judged.

[0016] Based on this, the present invention uses the feedback path established by the decision error to control the adaptive weight, thereby realizing the dynamic update of the adaptive weight. The LMS algorithm is configured in the adaptive equalizer, and the adaptive weight μ is adjusted by the decision error e, and the μ×e(n) data item in the LMS formula is adjusted, so as to iteratively update the tap coefficient, so that the adaptive equalizer converges.

[0017] The decision error is used to measure the difference between the output signal at the current moment and the expected signal. For example, the output level of device A is 1. After the signal is attenuated and tailed through cable transmission, it is input to the adaptive equalizer of device B. If the adaptive equalizer is in the initial adaptive stage, the equalization amplitude of the signal output by the adaptive equalizer is 0.55. The equalization amplitude 0.55 is compared with the decision reference line. If the decision reference line is ±0.5, the signal will be judged as 1. The decision error at this time = equalization amplitude 0.55-decision amplitude 1=-0.45. If the decision reference line is ±0.6, the signal will be judged as 0. The decision error at this time = equalization amplitude 0.55-decision amplitude 0=0.55.

[0018] Figure 1 1 is a waveform diagram showing a decision error of a decision device at different adaptation stages. Figure 1 The horizontal coordinate in represents the adaptation period, and the vertical coordinate represents the decision error. Figure 1 In the initial adaptive stage, the tap coefficients of the adaptive equalizer are initial values, the system is in an unstable state, and the output decision symbols may be incorrect, that is, the decision error is randomly distributed, and the decision error may be large or small. In the transition adaptive stage, the tap coefficients of the adaptive equalizer slowly stabilize, the output decision symbols are correct, and the decision error gradually decreases. In the stable adaptive stage, the tap coefficients of the adaptive equalizer are in a convergent state, the output decision symbols are correct, and the decision error fluctuates around 0.

[0019] Figure 2 is a schematic diagram showing the decision of a decider at different adaptation stages. Figure 2 The horizontal coordinate in represents the adaptive period, and the vertical coordinate represents the amplitude of the signal input to the decision device (that is, the amplitude of the signal restored after the tap coefficients of the input signal are calculated by the adaptive equalizer). Figure 2 In the initial adaptive stage, the signal recovered by the adaptive equalizer is relatively random, so the decision amplitude output by the decision maker is 1, 0, and -1, and the decision error e is relatively random. In the transition adaptive stage, the signal recovered by the adaptive equalizer gradually stabilizes, so the decision amplitude output by the decision maker is 0 and 1, and the decision error e gradually decreases. In the stable adaptive stage, the signal recovered by the adaptive equalizer is in a convergent state, so the decision amplitude output by the decision maker is 0 and -1, and the decision error e tends to 0. That is, with the adaptive process, the signal recovered by the adaptive equalizer will stabilize near -1, 0, and 1.

[0020] Hereinafter, the technical solution according to the present disclosure will be described with reference to specific embodiments and in conjunction with the accompanying drawings.

[0021] Figure 3 1 is a flow chart showing a method 100 for adjusting adaptive weights based on equalizer error according to an embodiment of the present disclosure. Figure 3 The method 100 includes the following steps 102 to 106.

[0022] In step 102, the decision error output by the decision device in each cycle is obtained.

[0023] In step 104, the number of qualified decision errors within a preset number of cycles is determined.

[0024] In some embodiments, the detection condition includes at least one discrimination criterion. In this embodiment, determining the number of decision errors that meet the conditions within a preset number of cycles includes: respectively counting the number of decision errors detected by the discrimination criterion within the preset number of cycles. In this way, the decision error output by the decider is the difference between the actual signal processed by the equalizer and the expected signal. The data characteristics of the decision errors obtained by the equalizer at different adaptive stages are different. Therefore, by detecting the number of decision errors obtained by the equalizer within the preset number of cycles, the data characteristics of the decision errors can be analyzed, thereby determining the adaptive working state of the equalizer.

[0025] In step 106, adaptive weights are configured for the equalizer according to the number of errors, so that tap coefficients of the equalizer adaptively converge according to the adaptive weights.

[0026] In some embodiments, configuring the adaptive weight for the equalizer according to the number of errors includes: determining the adaptive stage of the equalizer according to the number of errors detected by the discrimination criterion, and configuring the adaptive weight according to the adaptive stage. In this way, since the decision error changes with the change of the tap coefficient of the equalizer, the decision error fluctuates, but the data distribution characteristics of the decision error of the equalizer are different in different adaptive stages, so the adaptive stage of the equalizer can be determined by the number of decision errors. Dynamically configuring the adaptive weight according to the adaptive stage can ensure the convergence speed of the equalizer while avoiding system divergence.

[0027] In some embodiments, determining the adaptive stage of the equalizer according to the number of errors detected by the discrimination criteria includes: if the number of errors detected based on a discrimination criterion is greater than a number threshold corresponding to the discrimination criterion, determining that the equalizer is in the adaptive stage corresponding to the discrimination criterion. In this way, in different adaptive stages of the equalizer, the number of decision errors with data characteristics of the current stage is large, so when the number of errors detected by a certain discrimination criterion is greater than a preset threshold, the adaptive stage of the equalizer can be determined, and the adaptive weight can be updated in time.

[0028] In some embodiments, determining the adaptive stage of the equalizer according to the number of errors detected by the discrimination criteria further includes: if the number of errors detected based on the discrimination criteria is not greater than the number threshold corresponding to the discrimination criteria, then determining that the equalizer is in a transition adaptive stage. In this way, when the equalizer is in a transition adaptive stage, the data distribution of the judgment error does not have a distinct data distribution feature, and therefore cannot be judged by a specific discrimination criterion. Therefore, if the equalizer is not in an adaptive stage determined by any of the above discrimination criteria, the equalizer can be judged as a transition adaptive stage, thereby updating the adaptive weights.

[0029] In some embodiments, the condition includes a first criterion and a second criterion. In this embodiment, respectively counting the number of errors of the decision errors detected by the criterion within the preset number of cycles includes: counting the number of first errors of the first decision errors detected based on the first criterion within the preset number of cycles and the number of second errors of the second decision errors detected based on the second criterion. In this way, the number of errors detected by each criterion is counted, thereby determining the decision error data distribution of the current equalizer.

[0030] In some embodiments, the first criterion is greater than a first threshold, the second criterion is less than a second threshold, and the first threshold is greater than the second threshold. In this embodiment, counting the first error number of the first decision error detected based on the first criterion within the preset number of cycles and the second error number of the second decision error detected based on the second criterion includes: counting the first error number of the first decision error whose absolute value is greater than the first threshold within the preset number of cycles, and counting the second error number of the second decision error whose absolute value is less than the second threshold within the preset number of cycles. In this way, when the equalizer is in the initial adaptive stage, the tap coefficients used by the equalizer to process the signal are all initial values, so the decision error is large, so the first criterion can be used to determine whether the equalizer is in the initial adaptive stage at the current moment. When the equalizer is in the stable adaptive stage, the tap coefficients of the equalizer converge, so the decision error tends to 0, so the second criterion can be used to determine whether the equalizer is in the stable adaptive stage at the current moment. Thus, different adaptive stages can be distinguished by different criteria.

[0031] In some embodiments, if there is a number of errors detected based on a criterion that is greater than a number threshold corresponding to the criterion, determining that the equalizer is in an adaptive stage corresponding to the criterion includes: if the number of first errors corresponding to the first criterion is greater than the first number threshold corresponding to the first criterion, determining that the equalizer is in an initial adaptive stage corresponding to the first criterion. If the number of second errors corresponding to the second criterion is greater than the second number threshold corresponding to the second criterion, determining that the equalizer is in a stable adaptive stage corresponding to the second criterion. In this way, when the equalizer is in the initial adaptive stage, the number of errors detected by the first criterion is large. When the equalizer is in the stable adaptive stage, the number of errors detected by the second criterion is large. Therefore, through the decision error distribution characteristics of the equalizer in different adaptive stages, the working state of the equalizer can be effectively distinguished and determined.

[0032] In a specific application scenario, device A outputs a signal with a level amplitude of 1, which is attenuated and tailed through cable transmission and then input to the adaptive equalizer of device B. Based on this, the first threshold in the first discrimination criterion can be set to 0.4, and the first numerical threshold corresponding to the first discrimination criterion is 64. The second threshold in the second discrimination criterion can be set to 0.1, and the second numerical threshold corresponding to the second discrimination criterion is 128. The maximum value of the first numerical threshold and the second numerical threshold must be less than the preset number of cycles, and the smaller the difference between the numerical threshold and the preset number of cycles, the lower the sensitivity to switching the adaptive weight. The preset number of cycles is set to 256 according to the empirical value. The larger the preset number of cycles, the longer it takes to adjust the adaptive weight. In this scenario, if the number of decision errors greater than 0.4 (the first threshold) detected in 256 cycles exceeds 64 (the first numerical threshold), it can be determined that the adaptive equalizer is in the initial adaptive stage. If within 256 cycles, the number of decision errors less than 0.1 (second threshold) detected exceeds 128 (second numerical threshold), then it can be determined that the adaptive equalizer is in the stable adaptive stage. If within 256 cycles, the number of decision errors greater than 0.4 (first threshold) detected does not exceed 64 (first numerical threshold), and the number of decision errors less than 0.1 (second threshold) does not exceed 128 (second numerical threshold), then it can be determined that the adaptive equalizer is in the transitional adaptive stage.

[0033] In some embodiments, configuring the adaptive weights according to the adaptive stage includes configuring the adaptive weights for the equalizer as standard weights corresponding to the adaptive stage. In this way, the adaptive weights are dynamically configured according to the working state of the equalizer to perform iterative training of tap coefficients.

[0034] In theory, the maximum value of the adaptive weight is determined by the signal power, where the maximum value of the adaptive weight cannot exceed 1 / 2L, where L is the length of the filter. In practical applications, the adaptive weight is usually much smaller than 1 / 2L. In a specific application scenario, the standard weight corresponding to the initial adaptive stage can be set to 2 -4 (i.e. 1 / 16≈0.0625), 2 -4 is a relatively large step size, suitable for scenarios where the signal changes quickly or requires a fast response. The standard weight corresponding to the transition adaptation stage can be set to 2 -6 (i.e. 1 / 64≈0.015625), 2 -6 is a medium-sized step size, which is usually suitable for most application scenarios and provides a good compromise between convergence speed and stability. The standard weight corresponding to the stable adaptive stage can be set to 2 -8 (i.e. 1 / 256≈0.00390625), 2-8 It is a smaller step size, suitable for scenarios where the signal changes slowly or requires very stable behavior. In this case, the algorithm converges more slowly, but the system is less likely to diverge. It should be noted that the adaptive weight has different values ​​in different adaptive stages and is a constant value in the same adaptive stage.

[0035] In some embodiments, making the tap coefficients of the equalizer adaptively converge according to the adaptive weights includes: obtaining an input signal of the equalizer, generating a coefficient adjustment value according to the input signal, the decision error, and the adaptive weights, and controlling the equalizer to iteratively update the tap coefficients based on the coefficient adjustment value. In this way, the coefficient adjustment value of the tap coefficient is determined based on the configured adaptive weights, so that the iterative update of the tap coefficients can be synchronized and adaptive with the working conditions of the equalizer.

[0036] In some embodiments, generating a coefficient adjustment value according to the input signal, the decision error, and the adaptive weight includes: calculating an error compensation value according to the input signal and the decision error, loading the error compensation value into a shift register, performing a shift process on the error compensation value according to the adaptive weight to generate a shift result value, and outputting the shift result value as the coefficient adjustment value. In this way, dynamic configuration of adaptive weights can be achieved based on the shift register, which can effectively improve the processing efficiency of the equalizer.

[0037] Figure 4 is a schematic diagram showing a shift register according to an embodiment of the present disclosure. Figure 4 , the error compensation value A (error compensation value A = X(n) × e(n)) calculated based on the input signal X(n) and the decision error e(n) retains 10 bits and is loaded into the shift register, where the integer part is 1 bit and the decimal part is 9 bits. The value after the final shift processing, that is, the coefficient adjustment value, retains 18 bits, where the integer part is 1 bit and the decimal part is 17 bits. In a specific application scenario, when the adaptive equalizer is in the initial adaptive stage, the configured adaptive weight is 2 -4 At this time, the error compensation value A in the shift register can be shifted 4 bits to the right, and the unfilled integer high 4 bits are filled with the sign bit S, and the unfilled decimal low 4 bits are filled with 0. When the adaptive equalizer is in the transition adaptive stage, the configured adaptive weight is 2 -6 At this time, the error compensation value A in the shift register can be shifted 6 bits to the right, and the unfilled integer high 6 bits are filled with the sign bit S, and the unfilled decimal low 2 bits are filled with 0. When the adaptive equalizer is in the stable adaptive stage, the configured adaptive weight is 2 -8At this time, the error compensation value A in the shift register can be shifted right by 8 bits, and the unfilled high 8 bits of the integer are filled with the sign bit S, and the low bits of the decimal are all filled. In this way, the coefficient adjustment value = adaptive weight μ × error compensation value A = μ × X(n) × e(n) can be obtained. In the adaptive equalizer, LMS operation is performed based on the coefficient adjustment value, thereby realizing iterative update of the tap coefficient.

[0038] According to another aspect of the present invention, Figure 5 2 is a block diagram showing an apparatus 200 for adjusting adaptive weights based on equalizer error according to an embodiment of the present invention. Figure 5 The device 200 includes an error acquisition module 202, a quantity detection module 204, and a weight configuration module 206. The error acquisition module 202 is configured to acquire the decision error output by the decision device in each cycle. The quantity detection module 204 is configured to determine the number of errors of the decision errors that meet the conditions within a preset number of cycles. The weight configuration module 206 is configured to configure adaptive weights for the equalizer according to the number of errors, so that the tap coefficients of the equalizer are adaptively converged according to the adaptive weights.

[0039] According to yet another aspect of the present invention, Figure 6 3 is a diagram showing an architecture of a balancing system 300 according to an embodiment of the present invention. Figure 6 The system 300 includes a decision device 302, an equalizer 304, and the above-mentioned device 200 for adjusting adaptive weights based on equalizer errors. The decision device 302 and the equalizer 304 are both electrically coupled to the device 200.

[0040] The equalizer 304 includes a forward feedback equalizer 3042 and a decision feedback equalizer 3044. The outputs of the forward feedback equalizer 3042 and the decision feedback equalizer 3044 are electrically coupled to the input of the decider 302, and the output of the decider 302 is electrically coupled to the input of the decision feedback equalizer 3044.

[0041] In summary, the present invention provides a method, device and equalization system for adjusting adaptive weights based on equalizer errors, obtains the decision error output by the decision device in each cycle, and respectively counts the number of decision errors greater than or less than the threshold value within a preset number of cycles, thereby determining the adaptive stage of the current equalizer according to the data distribution of the decision error, and configuring different adaptive weights based on different adaptive stages, so that the equalizer can dynamically adjust the update step size of the tap coefficient according to the adaptive stage. Compared with the method of fixing the adaptive weight, dynamically adjusting the adaptive weight according to the adaptive working state of the equalizer can avoid the system oscillation or even divergence caused by excessive weight in the convergence state of the equalizer, and at the same time avoid the increase of computing resource consumption due to the adaptive weight being set too small, thereby achieving a dynamic balance between the stability of the equalizer error and the convergence speed.

[0042] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent transformations made using the contents of the present invention's specification and drawings, or directly or indirectly applied in related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for adjusting adaptive weights based on equalizer error, characterized in that: include: Obtaining the decision error output by the decision device in each cycle; Determine the number of errors of decision errors that meet the conditions within a preset number of cycles; as well as An adaptive weight is configured for the equalizer according to the number of errors, so that the tap coefficients of the equalizer adaptively converge according to the adaptive weight.

2. The method according to claim 1, characterized in that The condition includes at least one criterion, wherein determining the number of errors that meet the condition within a preset number of cycles includes: The number of decision errors detected by the discrimination standard within the preset number of cycles is counted respectively.

3. The method according to claim 2, characterized in that Configuring an adaptive weight for the equalizer according to the number of errors includes: Determining the adaptive stage of the equalizer according to the number of errors detected by the discrimination criterion; and The adaptation weights are configured according to the adaptation stage.

4. The method according to claim 3, characterized in that Determining the adaptive stage of the equalizer according to the number of errors detected by the discrimination criterion includes: If the number of errors detected based on a discrimination criterion is greater than a number threshold corresponding to the discrimination criterion, it is determined that the equalizer is in an adaptive stage corresponding to the discrimination criterion.

5. The method according to claim 4, characterized in that The adaptive stage of determining the equalizer according to the number of errors detected by the discrimination criterion further comprises: If the number of errors detected based on the discrimination criterion is not greater than the number threshold corresponding to the discrimination criterion, it is determined that the equalizer is in a transitional adaptive stage.

6. The method according to claim 3, characterized in that The conditions include a first discrimination criterion and a second discrimination criterion, wherein respectively counting the number of judgment errors detected by the discrimination criterion within the preset number of cycles includes: The number of first errors of the first decision errors detected based on the first decision criterion and the number of second errors of the second decision errors detected based on the second decision criterion within the preset number of cycles are counted.

7. The method according to claim 6, characterized in that The first discrimination criterion is greater than a first threshold, the second discrimination criterion is less than a second threshold, the first threshold is greater than the second threshold, The counting of the first error number of the first decision errors detected based on the first determination criterion within the preset number of cycles and the second error number of the second decision errors detected based on the second determination criterion comprises: Counting the number of first errors of first decision errors whose absolute values ​​are greater than the first threshold within the preset number of cycles; and The number of second errors of the second decision errors whose absolute values ​​are less than the second threshold within the preset number of cycles is counted.

8. The method according to claim 4, characterized in that If the number of errors detected based on a discrimination criterion is greater than a number threshold corresponding to the discrimination criterion, determining that the equalizer is in an adaptive stage corresponding to the discrimination criterion includes: If the first error number corresponding to the first criterion is greater than the first numerical threshold corresponding to the first criterion, determining that the equalizer is in an initial adaptive stage corresponding to the first criterion; and If the second error number corresponding to the second discrimination criterion is greater than the second numerical threshold corresponding to the second discrimination criterion, it is determined that the equalizer is in a stable adaptive stage corresponding to the second discrimination criterion.

9. The method according to claim 3, characterized in that: Configuring the adaptive weight according to the adaptive stage includes: The adaptive weights for the equalizer are configured as standard weights corresponding to the adaptive stage.

10. The method according to claim 1, characterized in that The step of causing the tap coefficients of the equalizer to adaptively converge according to the adaptive weights comprises: obtaining an input signal of the equalizer; and generating a coefficient adjustment value according to the input signal, the decision error and the adaptive weight; and The equalizer is controlled to iteratively update the tap coefficient based on the coefficient adjustment value.

11. The method according to claim 10, characterized in that Generating a coefficient adjustment value according to the input signal, the decision error and the adaptive weight comprises: Calculating an error compensation value according to the input signal and the decision error; Loading the error compensation value into a shift register to perform a shift process on the error compensation value according to the adaptive weight to generate a shift result value; and The shift result value is used as the coefficient adjustment value.

12. A device for adjusting adaptive weights based on equalizer error, characterized in that: include: An error acquisition module is configured to acquire a decision error output by the decision device in each cycle; A quantity detection module is configured to determine the number of errors of qualified decision errors within a preset number of cycles; as well as The weight configuration module is configured to configure adaptive weights for the equalizer according to the number of errors, so that the tap coefficients of the equalizer adaptively converge according to the adaptive weights.

13. A balancing system, characterized in that: include: The device according to claim 12; as well as A decider and an equalizer are electrically coupled to the device.

Images