Dynamic channel adjustment method and system based on SerDes high-speed serial communication link

Abstract

The invention relates to a dynamic channel adjustment method and system based on a SerDes high-speed serial communication link, and belongs to the technical field of communication links. The method comprises the following steps: acquiring a transmission signal, and obtaining a compensation transmission signal through a double-SRU neural network model according to the transmission signal; judging a feedback equalization model through two taps of a half-rate structure according to the compensation transmission signal to obtain a compensation full-rate signal; obtaining a self-adaptive equalization signal through a self-adaptive model according to the compensation full-rate signal; and obtaining a channel transmission signal through a filtering model according to the adaptive equalization signal. According to the invention, dynamic channel adjustment based on the SerDes high-speed serial communication link is realized, and the stability of signal transmission is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of communication links, and in particular relates to a channel dynamic adjustment method and system based on a SerDes high-speed serial communication link. Background Art

[0002] With the development of information technology, artificial intelligence, the Internet of Things, and 5G communications are deeply integrated into daily life. The realization of these technologies is inseparable from the support of high-bandwidth, high-speed, and low-latency communication links. Traditional parallel data transmission methods require a large number of IO interfaces, and application scenarios are severely limited. The high-speed serial communication link based on SerDes has gradually replaced the traditional parallel data transmission method and become the mainstream with its advantages of small number of pins, strong anti-interference ability, and low cost. However, with the increasing data transmission efficiency, the inter-symbol crosstalk caused by the non-ideal characteristics of the channel is a key factor affecting the transmission performance of the high-speed serial communication link based on SerDes. How to reduce inter-symbol interference and improve the stability of data transmission through dynamic adjustment of the channel is the current research difficulty and hot spot.

[0003] Equalization technology reduces the inter-symbol interference caused by the channel and ensures the integrity of high-speed serial signals by designing circuits with opposite characteristics to the channel characteristics at the receiver end. It is a key technology to improve the quality of channel transmission signals. Equalization methods based on equalizers include equalization methods based on forward feedback equalizers, equalization methods based on continuous time linear equalizers, and equalization methods based on decision feedback equalizers. They are not able to dynamically adjust the channel according to changes in signal loss. Equalization methods based on analog adaptive equalization structures cannot adapt well to increasingly high-speed data transmission and are prone to errors. Therefore, the above methods are not able to dynamically adjust the channel in high-speed serial communication links to reduce inter-symbol interference and improve the stability of data transmission. Summary of the invention

[0004] In order to solve the above problems existing in the prior art, the present invention provides a channel dynamic adjustment method and system based on SerDes high-speed serial communication link. The purpose of the present invention can be achieved by the following technical solutions: A channel dynamic adjustment method based on a SerDes high-speed serial communication link, comprising: Acquire a transmission signal, and obtain a compensated transmission signal through a dual SRU neural network model according to the transmission signal; According to the compensated transmission signal, a compensated full-rate signal is obtained by judging a feedback equalization model with two taps of a half-rate structure; Obtaining an adaptive equalization signal through an adaptive model according to the compensated full-rate signal; A channel transmission signal is obtained through a filtering model according to the adaptive equalization signal.

[0005] Preferably, obtaining a compensated transmission signal through a dual SRU neural network model according to the transmission signal comprises: The dual SRU neural network model includes a real part training SRU neural network and an imaginary part training SRU neural network; Acquire the real part and the imaginary part of the transmission signal according to the transmission signal, and train the SRU neural network according to the real part of the transmission signal to calculate the real part of the compensated transmission signal; The imaginary part of the transmission signal is calculated by training the SRU neural network with the imaginary part according to the imaginary part of the transmission signal to obtain the compensated transmission signal imaginary part; The compensated transmission signal is obtained by combining the real part of the compensated transmission signal and the imaginary part of the compensated transmission signal.

[0006] Preferably, the real part training SRU neural network training process includes: Preset an initial real part training SRU neural network, wherein the initial real part training SRU neural network includes an input layer, an SRU hidden layer, and an output layer, and the SRU hidden layer includes a forget gate and a reset gate; Acquire the real part of the training transmission signal, and obtain the real part of the training linear signal through linear change calculation according to the real part of the training transmission signal; Obtain a forget gate output by calculating the real part of the training transmission signal through the forget gate; Obtaining a reset gate output by calculating the real part of the training transmission signal through the reset gate; Obtain a hidden state through hidden state calculation according to the forget gate output and the real part of the training linear signal; The real part prediction value of the transmission signal is calculated by the SRU hidden layer according to the hidden state, the reset gate output, and the real part of the training transmission signal; Obtaining a true value of the real part of the transmission signal, and obtaining a loss function by calculating the loss function according to the predicted value of the real part of the transmission signal and the true value of the real part of the transmission signal; The initial real part training SRU neural network is updated by updating the gradient of the loss function to obtain the real part training SRU neural network.

[0007] Preferably, the half-rate structure two-tap judgment feedback equalization model includes an odd path and an even path.

[0008] Preferably, obtaining the compensated full-rate signal by judging the feedback equalization model with a half-rate structure two-tap according to the compensated transmission signal comprises: The compensation transmission signal is sampled at a rising edge and a falling edge of a half-rate clock to obtain an odd-branch half-rate signal and an even-branch half-rate signal; Obtain an odd-branch half-rate feedback signal through a one-stage decision delay according to the odd-branch half-rate signal; eliminate the inter-symbol interference of the even-path according to the odd-branch half-rate feedback signal to obtain an interference-free even-path; Obtain an even-branch half-rate feedback signal through a one-stage decision delay according to the even-branch half-rate signal; eliminate the inter-symbol interference of the odd-path according to the even-branch half-rate feedback signal to obtain an interference-free odd-path; Obtain an odd-branch half-rate delayed feedback signal through a delay unit according to the odd-branch half-rate signal; eliminate the inter-symbol interference of the odd-branch half-rate signal according to the odd-branch half-rate delayed feedback signal to obtain an interference-free odd-branch half-rate signal; Obtain an even-branch half-rate delayed feedback signal through a delay unit according to the even-branch half-rate signal; eliminate the inter-symbol interference of the even-branch half-rate signal according to the even-branch half-rate delayed feedback signal to obtain an interference-free even-branch half-rate signal; Obtaining a compensated odd-branch half-rate signal through the non-interference odd-branch path according to the non-interference odd-branch half-rate signal; Obtaining a compensated even-shunt half-rate signal through the non-interfering even-path path according to the non-interfering even-shunt half-rate signal; The compensated odd-branch half-rate signal and the compensated even-branch half-rate signal are restored by a selector to obtain the compensated full-rate signal.

[0009] Preferably, obtaining an adaptive equalization signal through an adaptive model according to the compensated full-rate signal comprises: Obtaining a signal attenuation amount by error calculation according to the compensated full-rate signal and the transmission signal; Obtaining an updated tap coefficient by updating calculation according to the signal attenuation and the compensated full-rate signal; Presetting an initial adaptive model, and updating the initial adaptive model according to the updated tap coefficients to obtain the adaptive model; The adaptive equalization signal is obtained through the adaptive model according to the compensated full-rate signal.

[0010] Preferably, obtaining a channel transmission signal according to a filtering model based on the adaptive equalization signal includes: Obtain a time domain signal through an ideal low-pass filter according to the adaptive equalization signal; A preset roll-off factor is used to obtain a raised cosine roll-off signal by multiplying the roll-off factor and the time domain signal; The channel transmission signal is obtained by Fourier transforming the raised cosine roll-off signal.

[0011] A channel dynamic adjustment system based on a SerDes high-speed serial communication link, comprising a signal compensation module, a signal equalization module, a signal adaptation module, and a signal filtering module; The signal compensation module is used to obtain a transmission signal, and obtain a compensated transmission signal according to the transmission signal through a dual SRU neural network model; The signal equalization module is used to obtain a compensated full-rate signal according to the compensated transmission signal through a half-rate structure two-tap judgment feedback equalization model; The signal adaptation module is used to obtain an adaptive equalization signal through an adaptive model according to the compensated full-rate signal; The signal filtering module is used to obtain a channel transmission signal according to the adaptive equalization signal according to a filtering model.

[0012] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the program, the channel dynamic adjustment method based on the SerDes high-speed serial communication link is implemented.

[0013] A storage medium containing computer executable instructions, wherein the computer executable instructions are used to execute the above-mentioned channel dynamic adjustment method based on SerDes high-speed serial communication link when executed by a computer processor.

[0014] The beneficial effects of the present invention are: (1) Through the dual SRU neural network model, the real and imaginary parts of the transmission signal are processed independently, which can more accurately correct the phase and amplitude deviations in the signal, dynamically compensate for the distortion and attenuation of the signal during transmission, and improve the signal integrity and anti-interference ability.

[0015] (2) Through the half-rate structure two-tap judgment feedback equalization model, the half-rate signal sampling is utilized by odd-even splitting, and the inter-symbol interference is eliminated through the one-level decision delay and delay unit, which significantly reduces the signal bit error rate and improves the stability of signal transmission.

[0016] (3) Through the adaptive model, the signal attenuation is calculated in real time and the tap coefficient is adjusted dynamically to ensure that the half-rate two-tap judgment feedback equalization model can adapt to different channel conditions, further improving the stability and adaptability of signal transmission.

[0017] (4) By performing filtering optimization at the receiving end, the roll-off factor and the time domain signal are multiplied to obtain a raised cosine roll-off signal. The roll-off factor can adapt to various bandwidth requirements and maximize bandwidth utilization; the raised cosine roll-off signal can reduce signal distortion; the channel transmission signal is obtained through Fourier transform based on the raised cosine roll-off signal, further improving the stability of channel transmission. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0019] Figure 1 The present invention is a flow chart of a method for dynamically adjusting a channel based on a SerDes high-speed serial communication link. DETAILED DESCRIPTION

[0020] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0021] See also Figure 1 , a channel dynamic adjustment method based on a SerDes high-speed serial communication link, comprising: S1: Acquire a transmission signal, and obtain a compensated transmission signal through a dual SRU neural network model according to the transmission signal; S2: obtaining a compensated full-rate signal through a half-rate structure two-tap judgment feedback equalization model according to the compensated transmission signal; S3: Obtaining an adaptive equalization signal through an adaptive model according to the compensated full-rate signal; S4: Obtain a channel transmission signal through a filtering model according to the adaptive equalization signal.

[0022] Specifically, the SerDes high-speed serial communication link is a serializer / deserializer high-speed serial communication link, which means that a low-speed parallel signal is converted into a high-speed serial signal through a serializer at the transmitting end and transmitted through a transmission medium, and the high-speed serial signal is restored to a low-speed parallel transmission signal through a deserializer at the receiving end.

[0023] In step S1, the dual SRU neural network model includes: S101: The dual SRU neural network model includes a real part training SRU neural network and an imaginary part training SRU neural network; S102: acquiring a real part of a transmission signal and an imaginary part of a transmission signal according to the transmission signal, and training an SRU neural network according to the real part of the transmission signal to calculate a real part of a compensated transmission signal; S103: calculating the imaginary part of the compensated transmission signal through the imaginary part training SRU neural network according to the imaginary part of the transmission signal, wherein the imaginary part training SRU neural network is the same as the real part training SRU neural network; S104: Obtain the compensated transmission signal by combining the real part of the compensated transmission signal and the imaginary part of the compensated transmission signal.

[0024] Specifically, the real part training SRU neural network training process includes: S102-1: Preset an initial real part training SRU neural network, wherein the initial real part training SRU neural network includes an input layer, an SRU hidden layer, and an output layer, wherein the SRU hidden layer includes a forget gate and a reset gate; S102-2: Obtain the real part of the training transmission signal, and obtain the real part of the training linear signal through linear change calculation according to the real part of the training transmission signal; The linear change calculation expression is: , in, represents the real part of the training linear signal, is the weight parameter, X n represents the real part of the training transmission signal; S102-3: Obtain a forget gate output by calculating the real part of the training transmission signal through the forget gate; The forget gate calculation expression is: , in, f n is the output of the forget gate, s represents the Sigmod function, is the weight parameter, is the bias parameter, X n represents the real part of the training transmission signal; S102-4: Obtain a reset gate output by calculating the real part of the training transmission signal through the reset gate; The reset gate calculation expression is: , in, r n is the reset gate output, s represents the Sigmod function, is the weight parameter, is the bias parameter, X n represents the real part of the training transmission signal; S102-5: Obtaining a hidden state through hidden state calculation according to the forget gate output and the real part of the training linear signal; The hidden state calculation expression is: , in, C n is the hidden state, f n is the output of the forget gate, C n-1 is the hidden state of the previous layer, represents the Hadamarda product, represents the real part of the training linear signal; S102-6: Calculate the real part prediction value of the transmission signal through the SRU hidden layer according to the hidden state, the reset gate output, and the real part of the training transmission signal; The SRU hidden layer calculation expression is: , in, h n is the real part prediction value of the transmission signal, r n is the reset gate output, g is the activation function, C n is the hidden state, X n represents the real part of the training transmission signal, It means Hadamarda product; S102-7: Obtain a true value of the real part of the transmission signal, and calculate a loss function according to the predicted value of the real part of the transmission signal and the true value of the real part of the transmission signal through a loss function; The loss function calculation expression is: , in, F rL represents the loss function, M represents the real part of the transmitted signal, j represents the real part of the jth transmitted signal, is the real value of the real part of the transmission signal, is a predicted value of the real part of the transmission signal; S102-8: Update the initial real part training SRU neural network by updating the gradient of the loss function to obtain the real part training SRU neural network.

[0025] In step S2, the half-rate structure two-tap judgment feedback equalization model includes: S201: The half-rate structure two-tap judgment feedback equalization model includes an odd path and an even path; S202: sampling the compensation transmission signal at a rising edge and a falling edge of a half-rate clock to obtain an odd-branch half-rate signal and an even-branch half-rate signal; S203: Obtain an odd-branch half-rate feedback signal through a one-stage decision delay according to the odd-branch half-rate signal; eliminate the inter-symbol interference of the even path according to the odd-branch half-rate feedback signal to obtain an interference-free even path; S204: Obtain an even-branch half-rate feedback signal through a one-stage decision delay according to the even-branch half-rate signal; eliminate the inter-symbol interference of the odd path according to the even-branch half-rate feedback signal to obtain an interference-free odd path; S205: Obtain an odd-branch half-rate delayed feedback signal through a delay unit according to the odd-branch half-rate signal; eliminate the inter-symbol interference of the odd-branch half-rate signal according to the odd-branch half-rate delayed feedback signal to obtain an interference-free odd-branch half-rate signal; S206: Obtain an even-branch half-rate delayed feedback signal through a delay unit according to the even-branch half-rate signal; eliminate the inter-symbol interference of the even-branch half-rate signal according to the even-branch half-rate delayed feedback signal to obtain an interference-free even-branch half-rate signal; S207: Obtain a compensated odd-branch half-rate signal through the interference-free odd-branch path according to the interference-free odd-branch half-rate signal; S208: Obtain a compensated even-split half-rate signal through the non-interference even-path path according to the non-interference even-split half-rate signal; S209: Restoring the compensated odd-branch half-rate signal and the compensated even-branch half-rate signal through a selector to obtain the compensated full-rate signal.

[0026] In this embodiment, the half-rate structure two-tap judgment feedback equalization model is a half-rate structure two-tap high-speed decision feedback equalizer, and the half-rate structure two-tap high-speed decision feedback equalizer includes an odd path and an even path. The Slicer is controlled by a half-rate clock to sample the compensation transmission signal at the rising edge and the falling edge of the clock respectively to obtain an odd-branch half-rate signal and an even-branch half-rate signal. The compensation transmission signal is a full-rate signal. Since odd and even sampling is performed alternately, the odd-branch half-rate feedback signal obtained after a one-level decision delay of the odd path is used to eliminate the inter-code interference of the even path to obtain an interference-free even path; similarly, the even-branch half-rate feedback signal obtained after a one-level decision delay of the even path is used to eliminate the inter-code interference of the odd path to obtain an interference-free odd path; and then the inter-code interference of the signal itself is eliminated by a delay unit, and the compensation full-rate signal is restored by a 2:1 multiplexer.

[0027] In step S3, the adaptive model is based on a symbol-to-symbol least mean square algorithm, specifically including: S301: Obtaining a signal attenuation amount through error calculation according to the compensated full-rate signal and the transmission signal; The error calculation expression is: , in, e(n) is the signal attenuation, x(n) is the compensated full rate signal, for said transmission signal; S302: obtaining an updated tap coefficient through updating calculation according to the signal attenuation and the compensated full-rate signal; The update calculation expression is: , in, ω(n+1) is the updated tap coefficient, ω(n) is the tap coefficient before updating, m is the update coefficient, sign [e(n)] Indicates that the signal attenuation is symbolized, sign[x(n)] Indicates that the compensated full-rate signal is symbolized, e(n) is the signal attenuation, x(n) is the compensated full-rate signal; S303: presetting an initial adaptive model, and updating the initial adaptive model according to the updated tap coefficients to obtain the adaptive model; S304: Obtain the adaptive equalization signal through the adaptive model according to the compensated full-rate signal.

[0028] Specifically, the symbolizing of the signal attenuation includes converting the signal attenuation into 1 if the signal attenuation is a positive number, and converting the signal attenuation into 0 if the signal attenuation is a negative number. The symbolizing of the compensated full-rate signal can be obtained similarly.

[0029] In step S4, the filtering model includes: S401: Obtain a time domain signal through an ideal low-pass filter according to the adaptive equalization signal; S402: Preset a roll-off factor, and obtain a raised cosine roll-off signal by multiplying the roll-off factor and the time domain signal; The multiplication calculation expression is: , in, is the time domain signal, is the roll-off factor, h(t) is the raised cosine roll-off signal, T B is the rate, is the cut-off angular frequency; S403: Obtain the channel transmission signal through Fourier transform according to the raised cosine roll-off signal; The Fourier transform expression is: , Wherein, y(t) is the channel transmission signal, x0(t) is the filtered adaptive equalization signal, n0(t) is Gaussian white noise, X(f) is the Fourier transform of the adaptive equalization signal, H(f) is the Fourier transform of the raised cosine roll-off signal, j represents an imaginary unit, and t represents time.

[0030] In this embodiment, the Gaussian white noise has a mean of 0 and a variance of σ 2 , Gaussian white noise with bilateral power spectral density n0 / 2.

[0031] A channel dynamic adjustment system based on a SerDes high-speed serial communication link, comprising a signal compensation module, a signal equalization module, a signal adaptation module, and a signal filtering module; The signal compensation module is used to obtain a transmission signal, and obtain a compensated transmission signal according to the transmission signal through a dual SRU neural network model; The signal equalization module is used to obtain a compensated full-rate signal according to the compensated transmission signal through a half-rate structure two-tap judgment feedback equalization model; The signal adaptation module is used to obtain an adaptive equalization signal through an adaptive model according to the compensated full-rate signal; The signal filtering module is used to obtain a channel transmission signal according to the adaptive equalization signal according to a filtering model.

[0032] The computer storage medium of the embodiment of the present invention may adopt any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device.

[0033] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0034] The program code included on the computer readable medium can be transmitted with any appropriate medium, including but not limited to wireless, electric wire, optical cable, RF, etc., or any suitable combination of the above. The computer program code for performing the operation of the present invention can be written in one or more programming languages ​​or their combinations, and the programming language includes object-oriented programming languages-such as Java, Smalltalk, C++, and also includes conventional procedural programming languages-such as "C" language or similar programming languages. The program code can be executed completely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on the remote computer, or completely on the remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (for example, using an Internet service provider to connect through the Internet).

[0035] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention without departing from the content of the technical solution of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A channel dynamic adjustment method based on a SerDes high-speed serial communication link, characterized in that: include: Acquire a transmission signal, and obtain a compensated transmission signal through a dual SRU neural network model according to the transmission signal; According to the compensated transmission signal, a compensated full-rate signal is obtained by judging a feedback equalization model with two taps of a half-rate structure; Obtaining an adaptive equalization signal through an adaptive model according to the compensated full-rate signal; A channel transmission signal is obtained through a filtering model according to the adaptive equalization signal.

2. The method for dynamically adjusting a channel based on a SerDes high-speed serial communication link according to claim 1, characterized in that: The obtaining of a compensated transmission signal through a dual SRU neural network model according to the transmission signal comprises: The dual SRU neural network model includes a real part training SRU neural network and an imaginary part training SRU neural network; Acquire the real part and the imaginary part of the transmission signal according to the transmission signal, and train the SRU neural network according to the real part of the transmission signal to calculate the real part of the compensated transmission signal; The imaginary part of the transmission signal is calculated by training the SRU neural network with the imaginary part according to the imaginary part of the transmission signal to obtain the compensated transmission signal imaginary part; The compensated transmission signal is obtained by combining the real part of the compensated transmission signal and the imaginary part of the compensated transmission signal.

3. The channel dynamic adjustment method based on SerDes high-speed serial communication link according to claim 2, characterized in that: The real part training SRU neural network training process includes: Preset an initial real part training SRU neural network, wherein the initial real part training SRU neural network includes an input layer, an SRU hidden layer, and an output layer, and the SRU hidden layer includes a forget gate and a reset gate; Acquire the real part of the training transmission signal, and obtain the real part of the training linear signal through linear change calculation according to the real part of the training transmission signal; Obtain a forget gate output by calculating the real part of the training transmission signal through the forget gate; Obtaining a reset gate output by calculating the real part of the training transmission signal through the reset gate; Obtain a hidden state through hidden state calculation according to the forget gate output and the real part of the training linear signal; The real part prediction value of the transmission signal is calculated by the SRU hidden layer according to the hidden state, the reset gate output, and the real part of the training transmission signal; Obtaining a true value of the real part of the transmission signal, and obtaining a loss function by calculating the loss function according to the predicted value of the real part of the transmission signal and the true value of the real part of the transmission signal; The initial real part training SRU neural network is updated by updating the gradient of the loss function to obtain the real part training SRU neural network.

4. The method for dynamically adjusting a channel based on a SerDes high-speed serial communication link according to claim 1, characterized in that: The half-rate structure two-tap judgment feedback equalization model includes an odd path and an even path.

5. The channel dynamic adjustment method based on SerDes high-speed serial communication link according to claim 4, characterized in that: The step of obtaining a compensated full-rate signal by using a half-rate structure two-tap judgment feedback equalization model according to the compensated transmission signal comprises: The compensation transmission signal is sampled at a rising edge and a falling edge of a half-rate clock to obtain an odd-branch half-rate signal and an even-branch half-rate signal; Obtain an odd-branch half-rate feedback signal through a one-stage decision delay according to the odd-branch half-rate signal; eliminate the inter-symbol interference of the even-path according to the odd-branch half-rate feedback signal to obtain an interference-free even-path; Obtain an even-branch half-rate feedback signal through a one-stage decision delay according to the even-branch half-rate signal; eliminate the inter-symbol interference of the odd-path according to the even-branch half-rate feedback signal to obtain an interference-free odd-path; Obtain an odd-branch half-rate delayed feedback signal through a delay unit according to the odd-branch half-rate signal; eliminate the inter-symbol interference of the odd-branch half-rate signal according to the odd-branch half-rate delayed feedback signal to obtain an interference-free odd-branch half-rate signal; Obtain an even-branch half-rate delayed feedback signal through a delay unit according to the even-branch half-rate signal; eliminate the inter-symbol interference of the even-branch half-rate signal according to the even-branch half-rate delayed feedback signal to obtain an interference-free even-branch half-rate signal; Obtaining a compensated odd-branch half-rate signal through the non-interference odd-branch path according to the non-interference odd-branch half-rate signal; Obtaining a compensated even-shunt half-rate signal through the non-interference even-path path according to the non-interference even-shunt half-rate signal; The compensated odd-branch half-rate signal and the compensated even-branch half-rate signal are restored by a selector to obtain the compensated full-rate signal.

6. The method for dynamically adjusting a channel based on a SerDes high-speed serial communication link according to claim 1, characterized in that: The step of obtaining an adaptive equalization signal through an adaptive model according to the compensated full-rate signal comprises: Obtaining a signal attenuation amount by error calculation according to the compensated full-rate signal and the transmission signal; Obtaining an updated tap coefficient by updating calculation according to the signal attenuation and the compensated full-rate signal; Presetting an initial adaptive model, and updating the initial adaptive model according to the updated tap coefficients to obtain the adaptive model; The adaptive equalization signal is obtained through the adaptive model according to the compensated full-rate signal.

7. The channel dynamic adjustment method based on SerDes high-speed serial communication link according to claim 1, characterized in that: The obtaining of a channel transmission signal according to a filtering model based on the adaptive equalization signal comprises: Obtain a time domain signal through an ideal low-pass filter according to the adaptive equalization signal; A preset roll-off factor is used to obtain a raised cosine roll-off signal by multiplying the roll-off factor and the time domain signal; The channel transmission signal is obtained by Fourier transforming the raised cosine roll-off signal.

8. A channel dynamic adjustment system based on a SerDes high-speed serial communication link, comprising a signal compensation module, a signal equalization module, a signal adaptation module, and a signal filtering module, characterized in that: include: The signal compensation module is used to obtain a transmission signal, and obtain a compensated transmission signal according to the transmission signal through a dual SRU neural network model; The signal equalization module is used to obtain a compensated full-rate signal according to the compensated transmission signal through a half-rate structure two-tap judgment feedback equalization model; The signal adaptation module is used to obtain an adaptive equalization signal through an adaptive model according to the compensated full-rate signal; The signal filtering module is used to obtain a channel transmission signal according to the adaptive equalization signal according to a filtering model.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the channel dynamic adjustment method based on the SerDes high-speed serial communication link is implemented as described in any one of claims 1-7.

10. A storage medium containing computer executable instructions, characterized in that: The computer executable instructions are used to execute the channel dynamic adjustment method based on the SerDes high-speed serial communication link as described in any one of claims 1-7 when executed by a computer processor.

Images