Baseline drift correction circuit, baseline drift correction method and communication system

Abstract

The invention provides a baseline drift correction circuit, a baseline drift correction method and a communication system, and relates to the technical field of communication, the baseline drift correction circuit comprises a compensator, a decision device, an adaptive updater, a low pass filter and a multiplier; the compensator outputs a compensated signal according to the acquired current receiving signal and a previous compensation signal output by the multiplier; the decision device outputs a current decision signal corresponding to the compensated signal; the self-adaptive updater performs self-adaptive updating of the gain and the time constant through a gradient descent algorithm according to the compensated signal and the current judgment signal to obtain the current gain and the current time constant; the low-pass filter outputs a current low-pass filtering signal according to the current time constant and the current judgment signal; and the multiplier outputs a current compensation signal according to the current gain and the current low-pass filtering signal. According to the invention, a gradient descent algorithm is adopted, two-dimensional optimization of gain and cut-off frequency is realized, and the accuracy of offset compensation is improved.

Description

Technical Field

[0001] The present invention relates to the field of communication technologies, and in particular, to a baseline drift correction circuit, a baseline drift correction method, and a communication system. Background Art

[0002] High-speed data communication technologies are widely used in communication networks. In recent years, with the rapid development of emerging industries such as the new energy vehicle industry, industrial robots, and artificial intelligence, the demand for data transmission speed in each emerging industry is getting higher and higher. Taking the intelligent vehicle industry as an example, intelligent vehicles transmit various data collected by various sensors in real time to a central processing chip and a dedicated high-computing power chip for data fusion and powerful neural network calculations, thereby achieving powerful perception capabilities. With the explosive growth of the amount of sensing data (such as high-definition cameras covering the vehicle body, lidar, millimeter-wave radar, etc.), how to transmit these data stably at high speed has become a very crucial link in vehicle intelligence. On-vehicle SerDes (Serializer / Deserializer) and on-vehicle Ethernet technologies play important roles in this. However, both SerDes and Ethernet will face the problem of baseline drift.

[0003] In wired communication systems such as SerDes and Ethernet, baseline drift is often caused by a transformer (i.e., a transformer) that realizes the coupling and separation of full-duplex communication, which is used to isolate the signal transmission path and can effectively separate the signals on the sending and receiving paths. When a waveform passes through the transformer, a phenomenon of DC droop (waveform distortion) will occur because the transformer often exhibits a high-pass characteristic and can be approximated as a high-pass filter.

[0004] Existing baseline drift compensation schemes generally assume that the high-pass characteristic of the transformer is known, that is, the cut-off frequency of the high-pass filter is known, and try in a limited frequency selection range, and then adaptively iterate the appropriate signal compensation gain based on the front and back errors of the decision maker. However, this scheme still has the problem that the offset compensation is not accurate enough, which easily causes the performance of the communication system to be unstable. Summary of the Invention

[0005] The purpose of the present invention is to provide a baseline drift correction circuit, a baseline drift correction method, and a communication system to improve the accuracy of offset compensation, thereby improving the stability of the performance of the communication system.

[0006] In a first aspect, the present invention provides a baseline drift correction circuit, which is applied to the receiving end of a communication system. The baseline drift correction circuit includes a compensator, a decision maker, an adaptive updater, a low-pass filter, and a multiplier; the input end of the compensator is connected to the output end of the multiplier, and the output end of the compensator is respectively connected to the input end of the decision maker and the input end of the adaptive updater. The output end of the decision maker is respectively connected to the input end of the adaptive updater and the input end of the low-pass filter. The output end of the adaptive updater is respectively connected to the input end of the multiplier and the input end of the low-pass filter. The output end of the low-pass filter is connected to the input end of the multiplier;

[0007] The compensator is configured to output a compensated signal according to the acquired current received signal and the previous compensation signal output by the multiplier;

[0008] The decision maker is configured to output a current decision signal corresponding to the compensated signal;

[0009] The adaptive updater is configured to adaptively update the gain and the time constant through a gradient descent algorithm according to the compensated signal and the current decision signal to obtain the current gain and the current time constant;

[0010] The low-pass filter is configured to output a current low-pass filtered signal according to the current time constant and the current decision signal;

[0011] The multiplier is configured to output a current compensation signal according to the current gain and the current low-pass filtered signal, and the current compensation signal is used for compensating the next received signal.

[0012] In an optional embodiment, the adaptive updater is specifically configured to: calculate a current error according to the compensated signal and the current decision signal; calculate a gain partial derivative and a time constant partial derivative according to the current decision signal, the previous gain, and the previous time constant; wherein, the gain partial derivative is the partial derivative of the output of the low-pass filter with respect to the gain, and the time constant partial derivative is the partial derivative of the output of the low-pass filter with respect to the time constant; perform an adaptive update of gradient descent on the previous gain and the previous time constant according to the gain partial derivative, the time constant partial derivative, and the current error to obtain the current gain and the current time constant.

[0013] In an optional embodiment, the gain partial derivative is calculated by the following formula:

[0014]

[0015] The time constant partial derivative is calculated by the following formula:

[0016]

[0017] Wherein, represents the gain partial derivative, Denote the partial derivative of the time constant, T s Denote the sampling period, RC(n) denote the previous time constant, g(n) denote the previous gain, and X(n + 1) denote the current decision signal.

[0018] In an alternative embodiment, the current gain is calculated by the following formula:

[0019]

[0020] The current time constant is calculated by the following formula:

[0021]

[0022] where g(n + 1) denote the current gain, g(n) denote the previous gain, μ g denote the preset gain learning rate, e(n + 1) denote the current error, denote the partial derivative of the gain, RC(n + 1) denote the current time constant, RC(n) denote the previous time constant, μ RC denote the preset time constant learning rate, denote the partial derivative of the time constant.

[0023] In an alternative embodiment, the current low - pass filtered signal is calculated by the following formula:

[0024]

[0025] where B pre (n + 1) denote the current low - pass filtered signal, B pre (n) denote the previous low - pass filtered signal, g(n) denote the previous gain, T s denote the sampling period, RC(n + 1) denote the current time constant, X(n + 1) denote the current decision signal.

[0026] In an alternative embodiment, the current compensation signal is calculated by the following formula:

[0027] BW(n + 1) = g(n + 1)*B pre (n + 1);

[0028] where BW(n + 1) denote the current compensation signal, g(n + 1) denote the current gain, B pre (n + 1) denote the current low - pass filtered signal.

[0029] In an alternative embodiment, the compensated signal is calculated by the following formula:

[0030] C(n + 1) = HP(n + 1)+BW(n);

[0031] Among them, C(n + 1) represents the compensated signal, HP(n + 1) represents the current received signal, and BW(n) represents the previous compensated signal.

[0032] In a second aspect, the present invention provides a baseline drift correction method, which is applied to the baseline drift correction circuit in any one of the foregoing embodiments; the baseline drift correction method includes:

[0033] Calculating a compensated signal based on the obtained current received signal and the previous compensated signal;

[0034] Determining a current decision signal corresponding to the compensated signal;

[0035] Performing adaptive update of the gain and the time constant through the gradient descent algorithm according to the compensated signal and the current decision signal to obtain the current gain and the current time constant;

[0036] Determining a current low-pass filtered signal according to the current time constant and the current decision signal;

[0037] Determining a current compensated signal according to the current gain and the current low-pass filtered signal, and the current compensated signal is used for compensating the next received signal.

[0038] In an optional embodiment, after performing adaptive update of the gain and the time constant through the gradient descent algorithm according to the current error and the current decision signal to obtain the current gain and the current time constant, the above baseline drift correction method further includes:

[0039] Calculating a current cut-off frequency according to the current time constant.

[0040] In a third aspect, the present invention provides a communication system, including a receiving end, and the receiving end includes the baseline drift correction circuit in any one of the foregoing embodiments.

[0041] In the baseline drift correction circuit, baseline drift correction method, and communication system provided by the present invention, the baseline drift correction circuit applied to the receiving end of the communication system includes a compensator, a decision maker, an adaptive updater, a low-pass filter, and a multiplier; the input end of the compensator is connected to the output end of the multiplier, the output end of the compensator is respectively connected to the input end of the decision maker and the input end of the adaptive updater, the output end of the decision maker is respectively connected to the input end of the adaptive updater and the input end of the low-pass filter, the output end of the adaptive updater is respectively connected to the input end of the multiplier and the input end of the low-pass filter, and the output end of the low-pass filter is connected to the input end of the multiplier; the compensator is configured to output a compensated signal according to the acquired current received signal and the previous compensation signal output by the multiplier; the decision maker is configured to output a current decision signal corresponding to the compensated signal; the adaptive updater is configured to perform adaptive updating of the gain and time constant through the gradient descent algorithm according to the compensated signal and the current decision signal to obtain the current gain and the current time constant; the low-pass filter is configured to output a current low-pass filtered signal according to the current time constant and the current decision signal; the multiplier is configured to output a current compensation signal according to the current gain and the current low-pass filtered signal, and the current compensation signal is used for compensating the next received signal. In this way, the adaptive updating of the gain and time constant is performed by using the gradient descent algorithm, realizing the two-dimensional self-optimization of the gain and cut-off frequency, improving the accuracy of offset compensation, and thus improving the stability of the performance of the communication system. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for use in the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0043] Figure 1 FIG. is a schematic structural diagram of a baseline drift correction circuit provided by an embodiment of the present invention;

[0044] Figure 2 FIG. is a schematic signal flow diagram of a baseline drift correction circuit provided by an embodiment of the present invention;

[0045] Figure 3 FIG. is a schematic curve diagram of gain adaptive optimization provided by an embodiment of the present invention;

[0046] Figure 4 FIG. is a schematic curve diagram of time constant adaptive optimization provided by an embodiment of the present invention;

[0047] Figure 5 FIG. is a schematic flow diagram of a baseline drift correction method provided by an embodiment of the present invention.

[0048] Icons: 101 - compensator; 102 - decision maker; 103 - adaptive updater; 104 - low-pass filter; 105 - multiplier. Specific embodiments

[0049] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0050] SerDes is a technology for high-speed data transmission that can convert the information on multiple parallel data lines into one or more high-speed serial lines, thereby reducing the number of pins required for connection and the number of signal paths. Both SerDes and Ethernet face the problem of baseline drift. Existing baseline drift compensation schemes assume a known high-pass characteristic of the transformer, but in an actual communication system, it is often impossible to model with a single high-pass filter characteristic. Moreover, existing baseline drift compensation schemes only focus on how to adaptively obtain the gain of the compensation signal and do not care about the cut-off frequency of the compensation signal, which results in inaccurate baseline offset compensation values and missing compensation for some frequency points, causing unstable system performance. Based on this, a baseline drift correction circuit, a baseline drift correction method, and a communication system provided by the embodiments of the present invention use the gradient descent algorithm to find the amplitude gain value g and the cut-off frequency fc of the high-pass characteristic (including but not limited to the high-pass characteristic of the transformer) of the signal waveform of the communication system. Based on the adaptive algorithm of the decision error, a real-time feedback network is used to adjust the baseline drift compensation value, which can achieve the correction of baseline drift occurring in the field of data communication.

[0051] For ease of understanding of this embodiment, a baseline drift correction circuit disclosed in the embodiments of the present invention will be introduced in detail first.

[0052] The embodiments of the present invention provide a baseline drift correction circuit, which is applied to the receiving end of a communication system. The communication system can be, but is not limited to, a wired communication system such as SerDes or Ethernet. As Figure 1As shown in the figure, the baseline drift correction circuit includes a compensator 101, a discriminator 102, an adaptive updater 103, a low-pass filter 104, and a multiplier 105. The input end of the compensator 101 is connected to the output end of the multiplier 105. The output end of the compensator 101 is respectively connected to the input end of the discriminator 102 and the input end of the adaptive updater 103. The output end of the discriminator 102 is respectively connected to the input end of the adaptive updater 103 and the input end of the low-pass filter 104. The output end of the adaptive updater 103 is respectively connected to the input end of the multiplier 105 and the input end of the low-pass filter 104. The output end of the low-pass filter 104 is connected to the input end of the multiplier 105.

[0053] Specifically, the above-mentioned compensator 101 is used to output a compensated signal according to the currently received signal obtained and the previous compensation signal output by the multiplier 105. The discriminator 102 is used to output a current decision signal corresponding to the compensated signal. The adaptive updater 103 is used to perform adaptive updates of the gain and time constant through the gradient descent algorithm according to the compensated signal and the current decision signal to obtain the current gain and the current time constant. The low-pass filter 104 is used to output a current low-pass filtered signal according to the current time constant and the current decision signal. The multiplier 105 is used to output a current compensation signal according to the current gain and the current low-pass filtered signal, and the current compensation signal is used for the compensation of the next received signal.

[0054] The baseline drift correction circuit provided by the embodiment of the present invention uses the gradient descent algorithm to perform adaptive updates of the gain and time constant, and the cut-off frequency is determined by the time constant. Therefore, two-dimensional optimization of the gain and cut-off frequency is achieved, the accuracy of offset compensation is improved, and thus the stability of the performance of the communication system is improved.

[0055] Optionally, the above-mentioned adaptive updater 103 is specifically used to: calculate a current error according to the compensated signal and the current decision signal; calculate a gain partial derivative and a time constant partial derivative according to the current decision signal, the previous gain, and the previous time constant, where the gain partial derivative is the partial derivative of the output of the low-pass filter 104 with respect to the gain, and the time constant partial derivative is the partial derivative of the output of the low-pass filter 104 with respect to the time constant; perform adaptive updates of the previous gain and the previous time constant by gradient descent according to the gain partial derivative, the time constant partial derivative, and the current error to obtain the current gain and the current time constant.

[0056] Optionally, the output formula of the low-pass filter 104 can be determined according to the impulse response of the low-pass filter 104 and the circuit structure of the baseline drift correction, and then the calculation formulas corresponding to the gain partial derivative and the time constant partial derivative can be determined.

[0057] Specifically, taking the low-pass filter 104 as a first-order RC filter as an example, its impulse response formula is as follows:

[0058]

[0059] where h LPF (t) represents the impulse response of the low-pass filter 104; RC represents the time constant; T s represents the sampling period, F s represents the sampling frequency; u(t) represents the unit step function, ensuring that the response is zero when t < 0, and t is the time variable.

[0060] Based on the above impulse response formula, taking the low-pass filter 104 as an example of a first-order IIR LPF (Infinite Impulse Response Low-Pass Filter), its output is:

[0061]

[0062] where B pre (n) represents the output of the low-pass filter 104 at time n, B pre (n - 1) represents the output of the low-pass filter 104 at time n - 1, g represents the gain, RC represents the time constant, X(n) represents the decision signal at time n, T s represents the sampling period.

[0063] Based on this, taking time n + 1 as the current time and time n as the previous time as an example, the above gain partial derivative is calculated through the following formula:

[0064]

[0065] The time constant partial derivative is calculated through the following formula:

[0066]

[0067] where, represents the gain partial derivative, represents the time constant partial derivative, T s represents the sampling period, RC(n) represents the previous time constant, g(n) represents the previous gain, and X(n + 1) represents the current decision signal.

[0068] Optionally, the above current gain is calculated through the following formula:

[0069]

[0070] The current time constant is calculated through the following formula:

[0071]

[0072] Among them, g(n + 1) represents the current gain, g(n) represents the previous gain, μg represents the preset gain learning rate, and e(n + 1) represents the current error. represents the gain partial derivative, RC(n + 1) represents the current time constant, RC(n) represents the previous time constant, and μ RC represents the preset time constant learning rate. The gain learning rate and the time constant learning rate are both preset empirical values. For example, the gain learning rate and the time constant learning rate are 10 -3 and 10 -4 .

[0073] Optionally, taking the first-order IIR LPF of the low-pass filter 104 as an example, the above-mentioned current low-pass filtered signal is calculated by the following formula:

[0074]

[0075] Among them, B pre (n + 1) represents the current low-pass filtered signal, B pre (n) represents the previous low-pass filtered signal, g(n) represents the previous gain, and T s represents the sampling period, RC(n + 1) represents the current time constant, and X(n + 1) represents the current decision signal.

[0076] Optionally, the above-mentioned multiplier 105 is specifically used for: multiplying the current gain and the current low-pass filtered signal to obtain the current compensation signal. Based on this, the above-mentioned current compensation signal is calculated by the following formula:

[0077] BW(n + 1) = g(n + 1) * B pre (n + 1);

[0078] Among them, BW(n + 1) represents the current compensation signal, g(n + 1) represents the current gain, and B pre (n + 1) represents the current low-pass filtered signal.

[0079] Optionally, the above-mentioned compensator 101 can be an adder. Based on this, the above-mentioned compensated signal is calculated by the following formula:

[0080] C(n + 1) = HP(n + 1) + BW(n);

[0081] Among them, C(n + 1) represents the compensated signal, HP(n + 1) represents the current received signal, and BW(n) represents the previous compensation signal.

[0082] In addition, the current cut-off frequency Fc of the above-mentioned low-pass filter 104 can be calculated based on the current time constant, and the calculation formula is as follows: The calculated cut-off frequency can be used to monitor the system state and determine whether the gradient descent algorithm converges. Specifically, based on the fluctuation of Fc, it can be determined whether Fc is stable, and then whether the gradient descent algorithm converges. In addition, the Fc range of the transformer is known, and it is possible to observe whether the gradient descent algorithm is about to converge based on the calculated Fc.

[0083] For the sake of easy understanding, taking the low-pass filter 104 as a first-order IIR LPF as an example, refer to Figure 2 to introduce the above-mentioned baseline drift correction circuit in detail.

[0084] As Figure 2 shown, the transformer in the communication system is modeled as a first-order RC HPF (Resistor-Capacitor High-Pass Filter). The transmitted signal X passes through the first-order RC HPF and outputs HP. After HP and the compensation signal BW pass through the adder, the compensated signal C is output. After C passes through the decision maker, it outputs The difference between C and is the error signal e. Based on e and

[0085] gain g self-adaptation and time constant RC self-adaptation are performed to realize the update of g and RC. After the updated RC and are input into the first-order RC LPF, the updated B pre is output. The updated g and the updated B pre pass through the multiplier and then output the updated compensation signal BW = g * B pre .

[0086] Specifically, assume that the transmitted signal X(n + 1) at time n + 1 passes through the first-order RC HPF and outputs HP(n + 1);

[0087] Compensate HP(n + 1) according to the compensation signal BW(n) output at time n to obtain the compensated signal C(n + 1), C(n + 1) = HP(n + 1) + BW(n);

[0088] The compensated signal C(n + 1) is used as the input of the decision maker. After the decision maker makes a decision, it outputs the signal

[0089] The decision error (i.e., the error signal) is or

[0090] Take the partial derivative of the output of the first-order RC LPF:

[0091] For g:

[0092] For RC:

[0093] Iteratively update the signal gain g according to: ;

[0094] Iteratively update the time constant RC of the first-order RC LPF according to: ;

[0095] The output signal B of the first-order RC LPF at the (n + 1)th moment pre (n + 1) is:

[0096]

[0097] The baseline drift compensation signal output at the (n + 1)th moment is: BW(n + 1) = g(n + 1) × B pre (n + 1).

[0098] The following provides an example of digital or analog implementation:

[0099] 1. Initialize parameters:

[0100] 1) Sampling frequency Fs, the duration T of one data symbol, and the time vector t; taking a 100 Mb bandwidth as an example, the duration T = 1 / 100 Mb;

[0101] 2) The cut-off frequency Fc of the high-pass filter (HPF) HPF and the cut-off frequency Fc of the low-pass filter (LPF) LPF ;

[0102] 3) The coefficients of HPF (numHPF, denHPF) and LPF (numLPF, denLPF); where numHPF and denHPF are the coefficients of the numerator and denominator polynomials in the HPF transfer function respectively, and numLPF and denLPF are the coefficients of the numerator and denominator polynomials in the LPF transfer function respectively;

[0103] 4) LMS (Least Mean Squares) adaptive parameters (such as the step size μ g , μ RC );

[0104] 5) Initialize signals and buffers (X, HP, B pre , e, etc.);

[0105] 6) Set the initial values of the filter parameters (such as g, RC LPF ).

[0106] 2. Main processing loop (for each time point n):

[0107] 1) High-pass filtering:

[0108] a. Update the HPF input buffer (RegBHPF);

[0109] b. Calculate the HPF output (hfTmp) using the difference equation;

[0110] c. Update the HPF state buffer (RegAHPF);

[0111] d. Save the HPF output as HP(n);

[0112] 2) Construct the combined signal:

[0113] a. Add the baseline drift estimate (BW) to the HP output to form C(n);

[0114] 3) Slicer module:

[0115] a. Pass C(n) through the slicer to obtain

[0116] 4) Error signal calculation:

[0117] a. Calculate the error

[0118] 5) Update the LPF parameters using LMS:

[0119] a. Calculate the gradient with respect to g: grad g ;

[0120] b. Calculate the gradient with respect to RC LPF : grad RC ;

[0121] c. Update g and RC using the LMS formula LPF :

[0122] g(n) = g(n - 1) - μ g × e(n) × grad g ;

[0123] RC LPF (n) = RC LPF (n - 1) - μ RC × e(n) × grad RC ;

[0124] d. Ensure RC LPFKeep it positive; if it is not positive, set RC LPF to a preset positive value close to 0;

[0125] e. Recalculate the LPF coefficient according to the updated RC LPF ;

[0126] 6) Low-pass filtering:

[0127] a. Update the LPF input buffer (RegBLPF);

[0128] b. Calculate the LPF output (lpTmp) using the difference equation;

[0129] c. Update the LPF status buffer (RegALPF);

[0130] d. Save the LPF output as B pre (n);

[0131] 3. End of loop.

[0132] To verify the baseline drift correction effect of the above baseline drift correction circuit, the real-time results of this embodiment are given below, Figure 3 and Figure 4 are the g adaptive result and the RC adaptive result respectively, where the RC adaptive result is reflected by the corner frequency. Figure 3 and Figure 4 show that g and RC simultaneously adaptively find the target optimal value from the initial 0 state. Among them, g quickly converges to around 1, and RC finally converges to around 5.1, completing the convergence and reaching system stability.

[0133] The embodiment of the present invention also provides a baseline drift correction method, which is applied to the above baseline drift correction circuit. Refer to Figure 5 the flowchart of a baseline drift correction method shown, and this baseline drift correction method includes the following steps S510 to step S550:

[0134] Step S510, calculate the compensated signal according to the acquired current received signal and the previous compensation signal;

[0135] Step S520, determine the current decision signal corresponding to the compensated signal;

[0136] Step S530, perform adaptive update of the gain and time constant through the gradient descent algorithm according to the compensated signal and the current decision signal to obtain the current gain and the current time constant;

[0137] Step S540, determine the current low-pass filtered signal according to the current time constant and the current decision signal;

[0138] Step S550: Determine a current compensation signal based on a current gain and a current low-pass filtered signal, where the current compensation signal is used for compensating a next received signal.

[0139] The baseline drift correction method provided by the embodiments of the present invention adaptively updates the gain and the time constant by using a gradient descent algorithm, realizes two-dimensional optimization of the gain and the cut-off frequency, improves the accuracy of offset compensation, and thus improves the stability of the performance of the communication system.

[0140] Further, after the above step S530, the method further includes: calculating a current cut-off frequency according to the current time constant. The current cut-off frequency can be calculated according to the formula to calculate the current cut-off frequency; where Fc represents the current cut-off frequency and RC represents the current time constant.

[0141] The embodiments of the present invention also provide a communication system, including a receiving end, and the receiving end includes the above-mentioned baseline drift correction circuit.

[0142] For the baseline drift correction method and the communication system provided by the present embodiment, the implementation principle and the technical effects generated are the same as those of the foregoing embodiment of the baseline drift correction circuit. For the sake of brief description, for the parts not mentioned in the embodiments of the baseline drift correction method and the communication system, reference may be made to the corresponding content in the foregoing embodiment of the baseline drift correction circuit.

[0143] The term "and / or" in this article is merely a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the term "at least one" in this article means any one of a plurality or any combination of at least two of a plurality. For example, including at least one of A, B, and C can represent including any one or more elements selected from the set composed of A, B, and C.

[0144] In all the examples shown and described here, any specific value should be construed as merely exemplary, rather than as a limitation. Therefore, other examples of the exemplary embodiments may have different values.

[0145] In addition, in the description of the embodiments of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

[0146] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A baseline drift correction circuit, characterized in that: Applied to a receiving end of a communication system, the baseline drift correction circuit comprises a compensator, a judger, an adaptive updater, a low-pass filter and a multiplier; the input end of the compensator is connected to the output end of the multiplier, the output end of the compensator is respectively connected to the input end of the judger and the input end of the adaptive updater, the output end of the judger is respectively connected to the input end of the adaptive updater and the input end of the low-pass filter, the output end of the adaptive updater is respectively connected to the input end of the multiplier and the input end of the low-pass filter, and the output end of the low-pass filter is connected to the input end of the multiplier; The compensator is used to output a compensated signal according to the acquired current received signal and the previous compensation signal output by the multiplier; The decision device is used to output a current decision signal corresponding to the compensated signal; The adaptive updater is used to adaptively update the gain and the time constant according to the compensated signal and the current decision signal through a gradient descent algorithm to obtain the current gain and the current time constant; The low-pass filter is used to output a current low-pass filter signal according to the current time constant and the current decision signal; The multiplier is used to output a current compensation signal according to the current gain and the current low-pass filter signal, and the current compensation signal is used for compensating a next received signal.

2. The baseline drift correction circuit according to claim 1, characterized in that: The adaptive updater is specifically used to: calculate the current error according to the compensated signal and the current decision signal; calculate the gain partial derivative and the time constant partial derivative according to the current decision signal, the previous gain and the previous time constant; wherein the gain partial derivative is the partial derivative of the output of the low-pass filter with respect to the gain, and the time constant partial derivative is the partial derivative of the output of the low-pass filter with respect to the time constant; and perform a gradient descent adaptive update on the previous gain and the previous time constant according to the gain partial derivative, the time constant partial derivative and the current error to obtain the current gain and the current time constant.

3. The baseline drift correction circuit according to claim 2, characterized in that: The gain partial derivative is calculated by the following formula: The time constant partial derivative is calculated by the following formula: in, represents the gain partial derivative, represents the time constant partial derivative, T s represents the sampling period, RC(n) represents the previous time constant, g(n) represents the previous gain, and X(n+1) represents the current decision signal.

4. The baseline drift correction circuit according to claim 2, characterized in that: The current gain is calculated by the following formula: The current time constant is calculated by the following formula: Wherein, g(n+1) represents the current gain, g(n) represents the previous gain, μg represents the preset gain learning rate, and e(n+1) represents the current error. represents the gain partial derivative, RC(n+1) represents the current time constant, RC(n) represents the previous time constant, μ RC represents the preset time constant learning rate, represents the time constant partial derivative.

5. The baseline drift correction circuit according to claim 1, characterized in that: The current low-pass filter signal is calculated by the following formula: Among them, B pre (n+1) represents the current low-pass filtered signal, B pre (n) represents the previous low-pass filter signal, g(n) represents the previous gain, T s represents the sampling period, RC(n+1) represents the current time constant, and X(n+1) represents the current decision signal.

6. The baseline drift correction circuit according to claim 1, characterized in that: The current compensation signal is calculated by the following formula: BW(n+1)=g(n+1)*B pre (n+1); Wherein, BW(n+1) represents the current compensation signal, g(n+1) represents the current gain, B pre (n+1) represents the current low-pass filtered signal.

7. The baseline drift correction circuit according to claim 1, characterized in that: The compensated signal is calculated by the following formula: C(n+1)=HP(n+1)+BW(n); Among them, C(n+1) represents the compensated signal, HP(n+1) represents the current received signal, and BW(n) represents the previous compensated signal.

8. A baseline drift correction method, characterized in that: A baseline drift correction circuit applied to any one of claims 1 to 7; the baseline drift correction method comprising: Calculate and obtain a compensated signal based on the currently received signal and the previous compensated signal; Determining a current decision signal corresponding to the compensated signal; According to the compensated signal and the current decision signal, adaptively updating the gain and the time constant by a gradient descent algorithm to obtain a current gain and a current time constant; Determining a current low-pass filter signal according to the current time constant and the current decision signal; A current compensation signal is determined according to the current gain and the current low-pass filter signal, and the current compensation signal is used for compensating a next received signal.

9. The baseline drift correction method according to claim 8, characterized in that: After adaptively updating the gain and the time constant by a gradient descent algorithm according to the compensated signal and the current decision signal to obtain the current gain and the current time constant, the baseline drift correction method further comprises: The current cutoff frequency is calculated according to the current time constant.

10. A communication system, characterized in that: It comprises a receiving end, wherein the receiving end comprises the baseline drift correction circuit according to any one of claims 1 to 7.

Images