Feedback decision equalization method, feedback decision equalizer structure and data receiving end

Abstract

The present invention belongs to the field of communication circuits, and proposes a feedback decision equalization method, a feedback decision equalizer structure and a data receiving end. The method comprises: obtaining an equalized signal according to the difference between an input signal and a feedback signal, wherein the feedback signal is a delayed decision signal modulated by a feedback tap coefficient, the decision signal is an equalized signal obtained after threshold decision, the reference signal is a decision signal modulated according to a reference tap coefficient, the difference between the equalized signal and the reference signal is calculated, and the feedback tap coefficient and the reference tap coefficient are adaptively trained and updated until the difference converges. In this case, the traditional method of additionally training the decision threshold is avoided by increasing the reference tap coefficient, thereby obtaining a feedback decision scheme with a simple structure and better performance.

Description

Technical Field

[0001] The invention belongs to the field of communication circuits, and in particular relates to a feedback decision equalization method, a feedback decision equalizer structure and a data receiving end. Background Art

[0002] When data passes through a high-speed channel, ISI (Inter-Symbol Interference) will occur. In order to recover data, the receiving end generally needs modules such as CTLE (Continuous Time Linear Equalizer), FFE (Feed Forward Equalization), and DFE (Decision Feedback Equalizer). The DFE module adopts a feedback decision structure, and the DFE coefficient is adaptively trained through the least mean square (LMS) error algorithm.

[0003] Reference to the working process of traditional feedback decision equalizer Figure 1 , the input signal x in The feedback signal x of the feedback modulation unit fb Subtract and get the equalized signal y eq , where the feedback signal x fb is the DFE coefficient with m taps (w 1 ~w m ) and the delayed decision signal d k-1 ~d k-m The product accumulation value represents the interference of other symbols on the current symbol. The equalized signal y eq Represents the input signal x in Filter out the feedback signal x representing the interference fb The equalized signal y eq After the decision slicer makes a decision, the decision value d is obtained k The gain value of the decision slicer is G th , then the error between the slicer module input and output is e k , record the signal threshold V th =1 / G th , the error value expression is e k =y eq -d k / G th =y eq -d k *V th . The equalized signal y eq , judgment value d k and Vth , calculate the error value e k , and then according to the error value e k Adjust the DFE coefficient in the adaptive training module so that the error of the feedback decision equalizer is gradually reduced. The operation of the feedback decision equalizer depends on the error value, and the error value depends on the signal threshold V th , and the signal threshold V th Usually, additional power consumption and time are required to stabilize, and the complexity is high. Therefore, it is necessary to provide a feedback decision equalization method, a feedback decision equalizer and a data receiving end to solve the above problems existing in the prior art. Summary of the invention

[0004] In order to reduce the dependence of the feedback decision equalizer on the signal threshold and reduce the overall complexity, the present application provides a feedback decision equalization method, a feedback decision equalizer and a data receiving end.

[0005] The present application provides a feedback decision equalization method, a feedback decision equalizer structure and a data receiving end, which adopt the following technical solutions:

[0006] A feedback decision equalization method, comprising:

[0007] The feedback signal is obtained by modulating the delayed decision signal according to the feedback tap coefficient;

[0008] An equalized signal is obtained by calculating the difference between the input signal and the feedback signal;

[0009] Performing threshold decision on the equalized signal to obtain a decision signal;

[0010] Modulating the decision signal according to a reference tap coefficient to obtain a reference signal;

[0011] Calculating a difference between an equalized signal and a reference signal, and adaptively training and updating the feedback tap coefficient and the reference tap coefficient according to the decision signal so that the difference tends to converge;

[0012] The feedback signal is generated by modulating the delayed decision signal by the feedback tap coefficient, and is used to represent the interference of the delayed symbol on the current symbol.

[0013] Through the above technical solution, adaptive training is performed based on the decision signal and the difference to continuously generate a training tap coefficient sequence to update the feedback tap coefficient and the reference tap coefficient, so that the difference gradually tends to converge, wherein the feedback tap coefficient is used to generate a feedback signal that simulates interference from other symbols, and the reference tap coefficient is used to generate a reference signal that equalizes the gain of the decision maker. By increasing the reference tap coefficient, the traditional practice of additional training decision thresholds is avoided, thereby obtaining an equalized signal with a simple structure and better performance.

[0014] Optionally, the feedback signal is obtained by modulating the delayed decision signal according to the feedback tap coefficient, including:

[0015] According to a preset number m of feedback taps, the decision signal is delayed by 1 to m symbol times to obtain a delayed decision signal sequence;

[0016] Modulating the delayed decision signal sequence one by one using the feedback tap coefficients in sequence to obtain a feedback sub-signal sequence, wherein the feedback tap coefficients have initial values;

[0017] The feedback signal is generated according to an aggregation result of the feedback sub-signal sequence.

[0018] By adopting the above technical solution, the decision signal is delayed according to the delay step length to obtain multiple delayed decision signal sequences with different delay timings. The decision signal sequences with different delay timings are modulated using the corresponding feedback tap coefficients to obtain the feedback sub-signals of the delay timings. The feedback sub-signals of all delay timings constitute a feedback sub-signal sequence. Then, all feedback sub-signals in all feedback sub-signal sequences are added to obtain a feedback signal. Among them, the first feedback signal is the initial value of the feedback tap coefficient, which is obtained by modulating the delayed historical decision signal.

[0019] Optionally, performing threshold decision on the equalized signal to obtain a decision signal includes:

[0020] When the threshold is When , the result of threshold judgment is , the decision function is:

[0021] ;

[0022] The decision signal is expressed as,

[0023] ;

[0024] in, is the decision signal, For a balanced signal, is the reference tap coefficient, Indicates a time sequence number.

[0025] Optionally, modulating the decision signal according to a reference tap coefficient to obtain a reference signal includes:

[0026] After threshold decision is performed on the equalized signal to obtain a decision signal, the decision signal is modulated according to a reference tap coefficient to obtain the reference signal, and the reference tap coefficient has an initial value.

[0027] Optionally, calculating the difference between the equalized signal and the reference signal, and adaptively training and updating the feedback tap coefficient and the reference tap coefficient according to the decision signal so that the difference tends to converge, includes:

[0028] Calculating a difference between the equalized signal and the reference signal to obtain an error signal;

[0029] Adaptively training the feedback tap coefficient and the reference tap coefficient according to the decision signal, and updating the feedback tap coefficient and the reference tap coefficient;

[0030] When the error signal meets a convergence criterion, it is determined that the error signal is converged, and the convergence criterion is that the sum of the error signals is less than a preset threshold value within a preset time length.

[0031] By adopting the above technical solution, when the error signal generated by the difference between the equalized signal and the reference signal reaches the convergence standard, it is determined that the equalized signal at this time is the result of the best effect of filtering the interference caused by other symbols of the input signal, so as to ensure that the output equalized signal gradually tends to the optimal.

[0032] The convergence standard is set to the sum of the errors within the preset window time, which is less than the preset threshold. Compared with the instantaneous judgment of the error signal to filter interference, the instantaneous judgment is prone to inaccuracy due to signal fluctuations. In this case, the sum of the errors within the window time is used to fully ensure the stability of the equalization signal during the judgment, reducing the impact of signal fluctuations on the overall accuracy.

[0033] Optionally, calculating a difference between the equalized signal and the reference signal to obtain an error signal includes:

[0034] The error signal is expressed as, ,

[0035] in, For a balanced signal, is the reference signal, represents the reference tap coefficient, is the decision signal, Indicates a time sequence number.

[0036] Optionally, adaptively training the feedback tap coefficient and the reference tap coefficient according to the decision signal includes:

[0037] According to the preset step size, the equalized signal, the reference signal, and the initial training tap coefficient sequence, the feedback tap coefficient and the reference tap coefficient in the modulated training tap coefficient sequence are calculated,

[0038] , ∈[0,m], m is the number of feedback taps;

[0039] The initial training tap coefficient sequence includes initial feedback tap coefficients and initial reference tap coefficients. Indicates the time sequence number. yes The tap coefficients of the corresponding positions after updating; is the error signal, For balanced signal is the preset step length;

[0040] When i=0, represents the reference tap coefficient, represents the initial reference tap coefficient;

[0041] When i>0, represents the feedback tap coefficient, Represents the initial feedback tap coefficient.

[0042] By adopting the above technical solution, is the i-th feedback tap coefficient in the modulated training tap coefficient sequence, is the i-th feedback tap coefficient in the training tap coefficient sequence before modulation, μ is the step length of the training tap coefficient sequence in the initial state, is the error signal, d is the decision signal, is the decision signal that has experienced i delays. The training tap coefficient sequence after adaptive training is updated according to the initial training tap coefficient sequence.

[0043] The present invention also provides a feedback decision equalizer structure, comprising a first subtractor, a decision device, an adaptive training unit, a feedback modulation unit, a reference modulation unit, a second subtractor, a convergence judgment unit and an output unit;

[0044] The first subtractor is connected to the feedback modulation unit and is used to calculate a difference between an input signal from the outside and a feedback signal from the feedback modulation unit to obtain an equalized signal;

[0045] The decision device is connected to the first subtractor and is used to make a decision on the equalized signal according to a threshold to obtain a decision signal;

[0046] The reference modulation unit is connected to the decision device, the adaptive training unit and the second subtractor, and is used to modulate the decision signal according to the reference tap coefficient to obtain a reference signal, and send it to the second subtractor;

[0047] The second subtractor is connected to the first subtractor, the reference modulation unit and the convergence judgment unit, and is used to calculate the difference between the reference signal and the equalized signal, generate an error signal and send it to the convergence judgment unit;

[0048] The feedback modulation unit is connected to the decision device, the adaptive training unit and the first subtractor, and is used to modulate the decision signal according to the feedback tap coefficient to obtain the feedback signal, and send it to the first subtractor;

[0049] The convergence judgment unit is connected to the second subtractor and is used to give a convergence judgment when the error signal meets the convergence standard;

[0050] The output unit is connected to the decision device and is used to output the decision signal;

[0051] The adaptive training unit is connected to the decision device and is used to perform adaptive training according to the decision signal to obtain a training tap coefficient sequence so that the error signal tends to converge. The training tap coefficient sequence includes feedback tap coefficients and reference tap coefficients.

[0052] By adopting the above technical solution, the first subtractor obtains an equalized signal according to the input signal and the feedback signal. The decision device processes the equalized signal to obtain a decision signal, and the decision signal is trained with a tap coefficient sequence after passing through an adaptive training unit, wherein the feedback tap coefficient is used to generate a feedback signal in combination with the decision signal to simulate inter-symbol interference, and the reference tap coefficient is used to align the decision device gain. This case introduces a reference modulation unit, which modulates the decision signal by adjusting the reference tap coefficient to obtain a reference signal, avoiding the traditional practice of additional training of the decision threshold, thereby obtaining an equalized signal with a simple structure and better performance.

[0053] Optionally, the feedback modulation unit includes a third adder and multiple groups of delay multiplication circuits, each group of delay multiplication circuits includes a delay device and a modulation multiplier;

[0054] In each group of delay multiplication circuits, the input end of the delay device is connected to the output end of the decision device, the input end of the modulation multiplier is connected to the output end of the delay device, the delay device is used to perform timing modulation on the decision signal to obtain a delayed decision signal, and the modulation multiplier is used to multiply the feedback tap coefficient with the delayed decision signal to obtain a feedback sub-signal; wherein the delay time length of the timing modulation of the delay device in each group of delay multiplication circuits is increasing;

[0055] The third adder is connected to the modulation multiplier in each group of delay multiplication circuits, and is used to add the feedback sub-signals of all delay multiplication circuits to obtain the feedback signal.

[0056] By adopting the above technical solution, in the adaptive training modulation unit, the delay device of each group of delay multiplication circuits is directly connected to the decision device to delay the decision signal for different time lengths, and then multiply it with the corresponding feedback tap coefficient to obtain a feedback sub-signal. All feedback sub-signals are added together by the third adder to obtain a feedback signal.

[0057] The present invention also provides a data receiving end, including a data recovery module, wherein the data recovery module includes a continuous time linear equalizer, a feedforward equalizer and the above-mentioned feedback decision equalizer structure.

[0058] In summary, the present application includes at least one of the following beneficial technical effects:

[0059] 1. This scheme sets the reference tap coefficient, has a simple structure, saves the power consumption and time of separate signal threshold training, and avoids the coupling of the DFE module and the signal threshold training module.

[0060] 2. This case sets the reference tap coefficient, reuses the traditional DFE feedback coefficient training method, and uses the LMS algorithm to converge along the gradient descent direction, which is faster than the traditional signal threshold convergence.

[0061] 3. In this case, the reference tap coefficient is set and the traditional DFE feedback coefficient training method is reused. Since the LMS algorithm has a small step size, the training results are more accurate and stable than those of the traditional signal threshold, and the DFE performance is better. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 is a structural diagram of a feedback decision equalizer in the background technology;

[0063] Figure 2 It is a flow chart of the feedback decision equalization method provided by the present invention;

[0064] Figure 3 It is a module schematic diagram of the feedback decision equalizer structure provided by the present invention;

[0065] Figure 4 yes Figure 3 A module schematic diagram of a feedback modulation unit;

[0066] Figure 5 yes Figure 4 A schematic diagram of the structure of the feedback modulation unit;

[0067] Figure 6 yes Figure 3 Another module schematic diagram of the feedback modulation unit;

[0068] Figure 7 yes Figure 6 Schematic diagram of the structure of the feedback modulation unit. DETAILED DESCRIPTION

[0069] In order to make the purpose, technical solutions and advantages of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Unless otherwise defined, the technical terms or scientific terms used herein should be understood by people with general skills in the field to which the present invention belongs. "Including" and similar words used in this article mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects.

[0070] The first embodiment of the present invention provides a feedback decision equalization method, referring to Figure 2 As shown, the method includes: obtaining a feedback signal by modulating a delayed decision signal according to a feedback tap coefficient; obtaining an equalized signal according to a difference between an input signal and a feedback signal; performing a threshold decision on the equalized signal to obtain a decision signal; modulating the decision signal according to a reference tap coefficient to obtain a reference signal; calculating the difference between the equalized signal and the reference signal, and adaptively training and updating the feedback tap coefficient and the reference tap coefficient according to the decision signal so that the difference tends to converge; wherein the feedback signal is generated by modulating the delayed decision signal by the feedback tap coefficient, and is used to represent the interference of the delayed symbol on the current symbol.

[0071] In this method, adaptive training is performed based on the decision signal and the difference to continuously generate a training tap coefficient sequence to update the feedback tap coefficient and the reference tap coefficient, so that the difference gradually tends to converge. Among them, the feedback tap coefficient is used to generate a feedback signal that simulates interference from other symbols, and the reference tap coefficient is used to generate a reference signal for the decision gain. By increasing the reference tap coefficient, as the difference gradually tends to converge, the reference tap coefficient also gradually approaches the decision gain, avoiding the traditional practice of additional training decision thresholds, thereby obtaining an equalized signal with a simple structure and better performance.

[0072] The following is a detailed description of the implementation details of a feedback decision equalization method of the invention. The following content is only provided for the convenience of understanding the implementation details. The specific process of this implementation is as follows Figure 2 As shown, the following steps are included:

[0073] Step 101: modulate the delayed decision signal according to the feedback tap coefficient to obtain a feedback signal.

[0074] Specifically, the feedback tap coefficient (i>0) has an initial value, represents the initial feedback tap coefficient. The specific implementation of step 101 includes:

[0075] S1-1, according to the preset number of feedback taps m, the decision signal Perform a delay of 1 to m symbol times, with a delay step of t for each symbol time, and obtain a delayed decision signal sequence { , , … }.

[0076] in, Represents the time sequence number, and the judgment signal Perform the first delay to obtain the delayed decision signal sequence ; The judgment signal Perform a second delay to obtain a delayed decision signal sequence ; The judgment signal Perform the third delay to obtain the delayed decision signal sequence ; The judgment signal Perform the mth delay to obtain the delayed decision signal sequence .

[0077] S1-2, using the training tap coefficient sequence { , , , … The feedback tap coefficients in { , , … }, for the delayed decision signal sequence { , , … } is modulated one by one to obtain the feedback sub-signal sequence, Indicates a time sequence number.

[0078] Specifically, the training tap coefficient sequence { , , … }Delay judgment sequence{ , , … } are modulated one by one, specifically:

[0079] The delay judgment signal obtained for the first delay , using the feedback tap coefficient Modulate and get the feedback sub-signal * ; The delay judgment signal obtained from the second delay , using the feedback tap coefficient Modulate and get the feedback sub-signal * ; The delay judgment signal obtained for the third delay , using the feedback tap coefficient Modulate and get the feedback sub-signal * ; And so on, the delay judgment signal obtained for the mth delay , using the feedback tap coefficient Modulate and get the feedback sub-signal * .

[0080] Among them, when k=0, the training tap coefficient sequence { , , , … } is set as the initial feedback tap coefficient in the initial training tap coefficient sequence { , , … }, using the initial feedback tap coefficient to determine the historical delayed decision signal sequence of the system input { , , … } is modulated one by one, where k ≥ m, to obtain a feedback sub-signal.

[0081] Under the continuous update of step 105, according to the judgment signal From the initial training tap coefficient sequence { , , , … }Start adaptive training and get the updated training tap coefficient sequence{ , , , … }, the updated training tap coefficient sequence { , , , … }Including reference tap coefficients and multiple feedback tap coefficients The training tap coefficient sequence { , , … }. Using the training tap coefficient sequence { , , … }The feedback tap coefficient (0<i≤m) for the delay judgment sequence { , , … } is modulated one by one to obtain the feedback sub-signal.

[0082] S1-3, generating a feedback signal according to the aggregation result of the feedback sub-signal sequence Specifically, the initial training tap coefficient sequence { , , , … }Including the initial feedback tap coefficient and the initial reference tap coefficients .

[0083] In this case, all feedback tap coefficients in the training tap coefficient sequence (i.e., DFE sequence) are (i>0) for the delayed decision signal Modulate and get feedback signal . The judgment signal Perform different delays and determine the decision signal for each delay and the corresponding feedback tap coefficient (i>0) are multiplied to obtain a feedback sub-signal, and then all feedback sub-signals are added to obtain a feedback signal.

[0084] Then the sum of these feedback sub-signals is used as the feedback signal , * + * + * +……+ * . Realize the tap coefficient according to the feedback (0<i≤m) for the decision signal Get feedback signal , the expression is: , .

[0085] Step 102: Calculate and obtain an equalized signal according to the difference between the input signal and the feedback signal.

[0086] Specifically, the data is transmitted to the data receiving end through a high-speed channel. In order to restore the original data, the data receiving end usually needs to use a decision feedback equalizer to filter out the interference of other symbols on the current symbol. The data received by the data receiving end, the feedback signal The quantized signal of other symbols interfering with the current symbol, the equalized signal It represents the output signal after filtering out the feedback signal. Among them, the feedback signal The generation of depends on step 102 and step 103, and k represents the time sequence number.

[0087] And, step 102 continuously outputs the balanced signal In this case, the balanced signal With the feedback tap coefficient The adaptive training update of (i>0) gradually tends to be stable.

[0088] Step 103: Perform threshold decision on the equalized signal to obtain a decision signal.

[0089] This patent includes but is not limited to the following functions. For example, the threshold of the decision slicer is , the corresponding judgment signal results are sorted according to the value size: ,

[0090] The judgment function SlicerJudge is:

[0091] .

[0092] Based on this case, the judgment signal is traditionally It is expressed as,

[0093] ;in, is the decision signal, is the decision function, For an equalized signal, the gain of the decision device is , Indicates a time sequence number.

[0094] In this application, Used to measure the size of Slicer gain and decision signal It is expressed as,

[0095] ;in, Indicates the time sequence number. is the decision signal, For a balanced signal, is the reference tap coefficient. Therefore, in this case, by setting the reference tap coefficient , saving the signal threshold The power consumption and time of separate training are reduced, and the coupling of the DFE module and the signal threshold training module is avoided.

[0096] Step 104: modulate the decision signal according to the reference tap coefficient to obtain a reference signal.

[0097] Specifically, the reference tap coefficient is used The decision signal Modulate to get the reference signal , .

[0098] Among them, when the reference signal is generated for the first time , using the initial reference tap coefficients , for the decision signal Modulation is performed, and then step 105 calculates the training tap coefficient sequence, using the updated reference tap coefficients .

[0099] Step 105, calculating the difference between the equalized signal and the reference signal, adaptively training and updating the feedback tap coefficient and the reference tap coefficient according to the decision signal, so that the difference tends to converge.

[0100] Specifically, the staff sets the initial training tap coefficient sequence and The initial training tap coefficient sequence is adaptively trained to obtain a continuously updated training tap coefficient sequence. Each training tap coefficient sequence includes a reference tap coefficient and the feedback tap coefficient (i>0), where the reference tap coefficient For the decision signal Modulate to get the reference signal ; Feedback tap coefficient (i>0) is used for the decision signal Modulate to obtain the feedback signal applied to step 101 According to the equalized signal and reference signal The error signal generated by the difference between Whether the convergence is the benchmark, the training tap coefficient sequence is continuously updated to make the error signal corresponding to the difference Tends to converge.

[0101] The implementation of this step 105 includes:

[0102] S5-1, calculate the equalized signal With reference signal The difference between the two gives the error signal , .

[0103] S5-2, according to the judgment signal Adaptively train and update feedback tap coefficients (i>0) and the reference tap coefficient , so that the error signal Tends to converge.

[0104] S5-3, judging the error signal Whether the convergence criterion is met, the error signal When the convergence criteria are met, the error signal is determined to be Convergence and feedback.

[0105] In step S5-2, adaptive training is performed according to the decision signal to obtain a training tap coefficient sequence, including: , balanced signal , reference signal , the initial training tap coefficient sequence { , , , … }, calculate the modulated training tap coefficient sequence { , , , … },

[0106] ,

[0107] ,

[0108] Among them, the initial training tap coefficient sequence includes the initial feedback tap coefficient and the initial reference tap coefficients , yes The tap coefficients of the corresponding positions after updating; is the error signal, For a balanced signal, is the reference signal, is the preset step size, k represents the time sequence number; when i=0, represents the reference tap coefficient, Represents the initial reference tap coefficient; when 0<i≤m, represents the feedback tap coefficient, represents the initial feedback tap coefficient, and m is the number of feedback taps.

[0109] The error signal When the convergence criteria are met, that is, when the input signal filters out interference caused by other symbols to the best effect, Established, among which, is the training tap coefficient sequence after modulation, is the training tap coefficient sequence before modulation, is the step size of the training tap coefficient sequence, is the error signal, is the decision signal, is the decision signal that has experienced i delays. The training tap coefficient sequence after adaptive training is updated according to the initial training tap coefficient sequence.

[0110] In step S5-3, the error signal is determined Whether the convergence standard is met, the convergence standard is that the error signal The sum of the error signals is less than the preset threshold. Instability during transmission, the error signal The convergence standard is set to the preset time window, and the sum of the errors is less than the preset threshold. In this case, the sum of the errors in the window time is used to fully ensure the balanced signal during the judgment period. The stability of the signal is improved, reducing the impact of signal fluctuations on the overall accuracy. Equalize the signal when converging Approximately , so that the balanced signal There is basically no inter-symbol interference (ISI).

[0111] It is worth mentioning that the balanced signal ; Error signal ; According to the LMS algorithm, along the gradient descent - In the direction of , adaptive training (i.e., DFE training) converges fastest, where i=0,1…m, m is a positive integer; , training tap coefficient sequence ~ Preset step size between , then is the DFE training step size, according to Direction Update When, then,

[0112] ,

[0113] in, yes The tap coefficients of the corresponding positions after updating. After simplification, when 0≤i≤m,

[0114] ;

[0115] .

[0116] Therefore, this case reuses the traditional DFE feedback coefficient training method and uses the LMS algorithm to converge along the gradient descent direction. This is faster and more stable than the traditional solution that requires additional training to obtain the signal threshold result, and the DFE performance is better.

[0117] The step division of the above-mentioned methods is only for the purpose of clear description. When implemented, they can be combined into one step or some steps can be split and decomposed into multiple steps. As long as they include the same logical relationship, they are all within the scope of protection of this patent. Adding insignificant modifications or introducing insignificant designs to the algorithm or process without changing the core design of the algorithm and process are all within the scope of protection of this patent.

[0118] The second embodiment of the present invention provides a feedback decision equalizer structure, such as Figure 3 As shown, including:

[0119] The first subtractor 301 is connected to the feedback modulation unit 304 and is used to calculate the input signal With feedback signal The difference between ;

[0120] The decision device 302 is connected to the first subtractor 301 and is used to Make a decision and get a decision signal ;

[0121] The reference modulation unit 305 is connected to the decision device 302, the adaptive training unit 303 and the second subtractor 306, and is used to modulate the reference tap coefficients according to the reference tap coefficients. The decision signal Modulate and get the reference signal and sent to the second subtractor 306;

[0122] The second subtractor 306 is connected to the first subtractor 301, the reference modulation unit 305 and the convergence judgment unit 307, and is used to calculate the reference signal With equalized signal The difference between ;

[0123] The feedback modulation unit 304 is connected to the decision device 302, the adaptive training unit 303, and the first subtractor 301, and is used to feedback the tap coefficients (0<i≤m, m is the number of taps), for the decision signal Modulate to get feedback signal and sent to the first subtractor 301;

[0124] The convergence judgment unit 307 is connected to the second subtractor 306 and is used to When the convergence criteria are met, a convergence judgment is given;

[0125] Output unit 308, connected to decision device 302, for outputting decision signal ;

[0126] The adaptive training unit 303 is connected to the decision maker 302, the reference modulation unit 305, and the feedback modulation unit 304, and is used to Adaptive training is performed to obtain a training tap coefficient sequence so that the error signal Towards convergence, the training tap coefficient sequence includes the feedback tap coefficient and the reference tap coefficients ; Where 0<i≤m, m is the number of taps.

[0127] In this case, the first subtractor 301 is based on the input signal and feedback signals used to simulate interference caused by other symbols , and get the balanced signal The decision maker 302 determines the equalized signal Modulation to get the decision signal , the decision signal After passing through the feedback modulation unit 304, a feedback signal is generated The adaptive training unit 303 will gradually adjust its own training tap coefficient sequence and update the feedback tap coefficients. (0<i≤m, m is the number of taps) and reference tap coefficient .

[0128] The present invention introduces a reference modulation unit 305, by adjusting the reference tap coefficient The decision signal Modulate to get the reference signal . The reference signal is then calculated and equalized signal The error signal formed by the difference , and then adjust the training tap sequence so that the equalized signal tends to perform best. In some examples, such as Figure 4 and Figure 5 As shown, the feedback modulation unit 304 includes a third adder 3041 and multiple groups of delay multiplication circuits, each group of delay multiplication circuits includes a delay 3042 and a modulation multiplier 3043;

[0129] In each group of delay multiplication circuits, the input end of the delay device 3042 is connected to the output end of the decision device 302, and the input end of the modulation multiplier 3043 is connected to the output end of the delay device 3032. The delay device 3042 is used to Perform timing modulation to obtain a delayed decision signal (0<i≤m, m is the number of taps), the modulation multiplier 3043 is used to feedback the tap coefficient With delayed decision signal Multiply to get the feedback sub-signal ; Among them, the delay time of the timing modulation of the delay device in each group of delay multiplication circuits is increasing, and the feedback tap coefficient The training tap coefficient sequence of the preset step size { , , , … }, i is any integer between 1 and m, m is the number of taps; the third adder 3041 is connected to the modulation multiplier 3042 in each group of delay multiplication circuits, and is used to add the feedback sub-signals of all delay multiplication circuits to obtain a feedback signal .

[0130] In the feedback modulation unit 304 of this example, the delay device 3042 of each delay multiplication circuit is directly connected to the decision device 302 to determine the decision signal. The delay of different lengths is performed, and then multiplied by the corresponding feedback tap coefficient to obtain a feedback sub-signal, and all the feedback sub-signals are added together by the third adder 3041 to obtain a feedback signal .

[0131] In other examples, such as Figure 6 and Figure 7 As shown, the feedback modulation unit 304 includes a third adder 3041 and multiple groups of delay multiplication circuits, each group of delay multiplication circuits includes a delay 3042 and a modulation multiplier 3043;

[0132] The input end of the delay device 3042 in the first group of delay multiplication circuits is connected to the output end of the decision device 302, and the input end of the modulation multiplier 3043 is connected to the output end of the delay device 3042;

[0133] The input end of the delay device 3042 in the latter delay multiplication circuit is connected to the output end of the delay device 3042 in the former delay multiplication circuit, and the input end of the modulation multiplier 3043 is connected to the output end of the delay device 3042. The delay device 3042 in the first position is used to Perform timing modulation to obtain a delayed decision signal The next delay device 3042 is used to perform timing modulation on the delayed decision signal input by the previous delay device 3042 to obtain a new delayed decision signal and update it; the modulation multiplier 3043 is used to feedback the tap coefficient With delayed decision signal The feedback sub-signal is obtained by multiplying the delay 3042 in each group of delay multiplication circuits. The timing modulation is the same, and the feedback tap coefficient is (i>0 and i≤m) belongs to the training tap coefficient sequence of the preset step size { , , , … The third adder 3041 is connected to the modulation multiplier 3032 in each group of delay multiplication circuits, and is used to add the feedback sub-signals of all delay multiplication circuits to obtain a feedback signal .

[0134] In the adaptive training modulation unit 303 of this example, the delay 3042 of the delay multiplication circuit in the first place is connected to the decision device 302, and the decision signal The delay is then multiplied by the corresponding feedback tap coefficient to obtain a feedback sub-signal; the delay device 3042 of the delay multiplication circuit of the latter bit is connected to the delay device 3042 of the delay multiplication circuit of the previous bit, and another delay is performed on the basis of the delay of the previous delay device 3042, and then multiplied by the corresponding feedback tap coefficient to obtain a feedback sub-signal. All feedback sub-signals are added together by the third adder 3041 to obtain a feedback signal .

[0135] In addition, other implementation details and working methods of the feedback decision equalizer structure disclosed in the present application are the same or similar to the feedback decision equalization method described above and will not be repeated here.

[0136] The third embodiment of the present invention also discloses a data receiving end. The data receiving end includes a data recovery module, and the data recovery module includes a continuous time linear equalizer, a feedforward equalizer and the above-mentioned feedback decision equalizer structure.

[0137] Although the embodiments of the present invention are described in detail above, it is obvious to those skilled in the art that various modifications and variations can be made to these embodiments. However, it should be understood that such modifications and variations are within the scope and spirit of the present invention as described in the claims. Moreover, the present invention described herein may have other embodiments and may be implemented or realized in a variety of ways.

Claims

1. A feedback decision equalization method, characterized in that: include: The feedback signal is obtained by modulating the delayed decision signal according to the feedback tap coefficient; An equalized signal is obtained by calculating the difference between the input signal and the feedback signal; Performing threshold decision on the equalized signal to obtain a decision signal; Modulating the decision signal according to a reference tap coefficient to obtain a reference signal; Calculating a difference between an equalized signal and a reference signal, and adaptively training and updating the feedback tap coefficient and the reference tap coefficient according to the decision signal so that the difference tends to converge; The feedback signal is generated by modulating the delayed decision signal by the feedback tap coefficient, and is used to represent the interference of the delayed symbol on the current symbol.

2. The feedback decision equalization method according to claim 1, characterized in that: The method of obtaining a feedback signal by modulating the delayed decision signal according to the feedback tap coefficient includes: According to a preset number m of feedback taps, the decision signal is delayed by 1 to m symbol times to obtain a delayed decision signal sequence; Modulating the delayed decision signal sequence one by one using the feedback tap coefficients in sequence to obtain a feedback sub-signal sequence, wherein the feedback tap coefficients have initial values; The feedback signal is generated according to an aggregation result of the feedback sub-signal sequence.

3. The feedback decision equalization method according to claim 1, characterized in that: The step of performing threshold decision on the equalized signal to obtain a decision signal comprises: When the threshold is When , the result of threshold judgment is , the decision function is: ； The decision signal is expressed as, ； in, is the decision signal, For a balanced signal, is the reference tap coefficient, Indicates a time sequence number.

4. The feedback decision equalization method according to claim 1, characterized in that: Modulating the decision signal according to the reference tap coefficient to obtain a reference signal includes: After threshold decision is performed on the equalized signal to obtain a decision signal, the decision signal is modulated according to a reference tap coefficient to obtain the reference signal, and the reference tap coefficient has an initial value.

5. The feedback decision equalization method according to claim 1, characterized in that: The calculating the difference between the equalized signal and the reference signal, and adaptively training and updating the feedback tap coefficient and the reference tap coefficient according to the decision signal so that the difference tends to converge, comprises: Calculating a difference between the equalized signal and the reference signal to obtain an error signal; Adaptively training the feedback tap coefficient and the reference tap coefficient according to the decision signal, and updating the feedback tap coefficient and the reference tap coefficient; When the error signal meets a convergence criterion, it is determined that the error signal is converged, and the convergence criterion is that the sum of the error signals is less than a preset threshold value within a preset time length.

6. The feedback decision equalization method according to claim 5, characterized in that: Calculating the difference between the equalized signal and the reference signal to obtain an error signal includes: The error signal is expressed as, , in, For a balanced signal, is the reference signal, represents the reference tap coefficient, is the decision signal, Indicates a time sequence number.

7. The feedback decision equalization method according to claim 5, characterized in that: Adaptively training the feedback tap coefficient and the reference tap coefficient according to the decision signal, comprising: According to a preset step size, the equalized signal, the reference signal, and an initial training tap coefficient sequence, the feedback tap coefficient and the reference tap coefficient in the training tap coefficient sequence are calculated, , ∈[0,m], m is the number of feedback taps; The initial training tap coefficient sequence includes initial feedback tap coefficients and initial reference tap coefficients. Indicates the time sequence number. yes The tap coefficients of the corresponding positions after updating; is the error signal, is the preset step length; When i=0, represents the reference tap coefficient, represents the initial reference tap coefficient, The decision signal ; When i>0, represents the feedback tap coefficient, represents the initial feedback tap coefficient, Indicates delayed decision signal, meaning decision signal Conduct the The signal obtained by delaying.

8. A feedback decision equalizer, characterized in that: It includes a first subtractor, a decision device, an adaptive training unit, a feedback modulation unit, a reference modulation unit, a second subtractor, a convergence judgment unit and an output unit; The first subtractor is connected to the feedback modulation unit and is used to calculate a difference between an input signal from the outside and a feedback signal from the feedback modulation unit to obtain an equalized signal; The decision device is connected to the first subtractor and is used to make a decision on the equalized signal according to a threshold to obtain a decision signal; The reference modulation unit is connected to the decision device, the adaptive training unit and the second subtractor, and is used to modulate the decision signal according to the reference tap coefficient to obtain a reference signal, and send it to the second subtractor; The second subtractor is connected to the first subtractor, the reference modulation unit and the convergence judgment unit, and is used to calculate the difference between the reference signal and the equalized signal, generate an error signal and send it to the convergence judgment unit; The feedback modulation unit is connected to the decision device, the adaptive training unit and the first subtractor, and is used to modulate the decision signal according to the feedback tap coefficient to obtain the feedback signal, and send it to the first subtractor; The convergence judgment unit is connected to the second subtractor and is used to give a convergence judgment when the error signal meets the convergence standard; The output unit is connected to the decision device and is used to output the decision signal; The adaptive training unit is connected to the decision device, the reference modulation unit, and the feedback modulation unit, and is used to perform adaptive training according to the decision signal to obtain a training tap coefficient sequence so that the error signal tends to converge. The training tap coefficient sequence includes feedback tap coefficients and reference tap coefficients.

9. The feedback decision equalizer structure according to claim 8, characterized in that: The feedback modulation unit includes a third adder and a plurality of groups of delay multiplication circuits, each group of delay multiplication circuits includes a delay device and a modulation multiplier; In each group of delay multiplication circuits, the input end of the delay device is connected to the output end of the decision device, the input end of the modulation multiplier is connected to the output end of the delay device, the delay device is used to perform timing modulation on the decision signal to obtain a delayed decision signal, and the modulation multiplier is used to multiply the feedback tap coefficient with the delayed decision signal to obtain a feedback sub-signal; wherein the delay time length of the timing modulation of the delay device in each group of delay multiplication circuits is increasing; The third adder is connected to the modulation multiplier in each group of delay multiplication circuits, and is used to add the feedback sub-signals of all delay multiplication circuits to obtain the feedback signal.

10. A data receiving end, characterized in that: It comprises a data recovery module, which comprises a continuous time linear equalizer, a feedforward equalizer and the feedback decision equalizer structure as claimed in claim 8.

Images