Clock Leakage Compensation Method and System for Fiber Optic Communication Systems

Abstract

This invention provides a clock leakage compensation method and system for optical fiber communication systems, comprising: Step S1: establishing a training dataset and constructing a learnable digital signal normalization structure; Step S2: training the learnable digital signal normalization structure using the training dataset; Step S3: using the trained learnable digital signal normalization structure to implement clock leakage compensation and deploying it in the optical fiber communication system. This invention constructs a learnable digital signal normalization structure, a training method, and a deployment process. The method and system can be used to compensate for clock leakage in optical communication systems and improve the transmission system performance of optical fiber communication systems.

Description

Technical Field

[0001] This invention relates to the technical field of optical fiber communication systems, and more specifically, to a clock leakage compensation method and system for optical fiber communication systems. Background Technology

[0002] In the field of fiber optic communication systems, signal quality is affected by various non-ideal effects of devices. Among these, clock leakage in high-speed DACs is one such non-ideal effect affecting signal quality. During transmission, signals may be subject to additional frequency interference within the signal spectrum, leading to a deterioration in the received bit error rate performance.

[0003] Patent document CN106656120A (application number: 201611015057.4) discloses a clock compensation circuit, a clock circuit, and a microcontroller. The clock compensation circuit includes: a detection circuit for detecting capacitance control parameters in the clock circuit that adjust a target capacitor, wherein the target capacitor controls the clock frequency output by the clock circuit; and a control unit connected to the detection circuit for controlling the target capacitor according to the capacitance control parameters detected by the detection circuit, thereby adjusting the clock frequency output by the clock circuit.

[0004] To address the aforementioned shortcomings, this invention proposes a clock leakage compensation method and system for optical fiber communication systems. It proposes a learnable normalized structure and optimizes its parameters through error calculation, backpropagation algorithm, and gradient descent method, thereby achieving clock leakage compensation in optical fiber communication systems, improving signal quality, and reducing the bit error rate after transmission. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the present invention aims to provide a clock leakage compensation method and system for optical fiber communication systems.

[0006] A clock leakage compensation method for an optical fiber communication system provided by the present invention includes:

[0007] Step S1: Establish a training dataset and construct a learnable digital signal normalization structure;

[0008] Step S2: Train a learnable digital signal normalization structure using the dataset;

[0009] Step S3: Utilize the trained, learnable digital signal normalization structure to achieve clock leakage compensation and deploy it in the fiber optic communication system.

[0010] Preferably, the learnable mathematical signal normalization structure adopts:

[0011]

[0012] Where, μ x σ is the mean of the input digital signal. x Let k be the average power of the input digital signal, k be the scaling factor, b be the bias sequence, and μ be the bias value. b is the mean of the bias sequence; k and b are both learnable parameters.

[0013] Preferably, step S2 employs:

[0014] Step S2.1: Input the input digital signal from the training data into the learnable normalization result to obtain the output digital signal;

[0015] Step S2.2: Construct an error function based on the labels of the output digital signal and training data to calculate the error magnitude. Use gradient backpropagation and gradient descent to train the learnable normalized structure. When the training reaches a preset number of times, the training result is obtained.

[0016] Preferably, the error function includes:

[0017]

[0018] Where N is the training batch size, y is the digital signal processing output signal, and x is the training batch size. label These are the label symbols for the training data.

[0019] Preferably, the learnable normalized structure is trained using gradient backpropagation and gradient descent to update the learning parameters of the learnable normalized structure, including:

[0020]

[0021]

[0022] Among them, b i and k i Let b and k represent the parameters after the i-th update. and Let lr represent the gradients of the error functions corresponding to b and k, respectively. i-1 Let be the learning rate for the (i-1)th iteration.

[0023] A clock leakage compensation system for an optical fiber communication system provided by the present invention includes:

[0024] Module M1: Establishes the training dataset and constructs a learnable digital signal normalization structure;

[0025] Module M2: Trains a learnable digital signal normalization structure using a dataset;

[0026] Module M3: Utilizes a trained, learnable digital signal normalization structure to achieve clock leakage compensation, and is deployed in fiber optic communication systems.

[0027] Preferably, the learnable mathematical signal normalization structure adopts:

[0028]

[0029] Where, μ x σ is the mean of the input digital signal. x Let k be the average power of the input digital signal, k be the scaling factor, b be the bias sequence, and μ be the bias value. b is the mean of the bias sequence; k and b are both learnable parameters.

[0030] Preferably, the module M2 adopts:

[0031] Module M2.1: Inputs the input digital signal from the training data into a learnable normalized result to obtain the output digital signal;

[0032] Module M2.2: Constructs an error function based on the labels of the output digital signal and training data to calculate the error magnitude, and trains the learnable normalized structure using gradient backpropagation and gradient descent. When the training reaches a preset number of times, the training result is obtained.

[0033] Preferably, the error function includes:

[0034]

[0035] Where N is the training batch size, y is the digital signal processing output signal, and x is the training batch size. label These are the label symbols for the training data.

[0036] Preferably, the learnable normalized structure is trained using gradient backpropagation and gradient descent to update the learning parameters of the learnable normalized structure, including:

[0037]

[0038]

[0039] Among them, b i and k i Let b and k represent the parameters after the i-th update. and Let lr represent the gradients of the error functions corresponding to b and k, respectively. i-1 Let be the learning rate for the (i-1)th iteration.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] 1. This invention optimizes the parameters of the normalization module using the gradient descent method;

[0042] 2. This invention designs a learnable normalized structure, which compensates for clock leakage effects, improves the signal quality of optical fiber communication, and reduces the bit error rate after transmission.

[0043] 3. This system can be applied to any channel, and in particular, it is suitable for various fiber optic communication scenarios. Attached Figure Description

[0044] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0045] Figure 1 This is a schematic flowchart of the clock leakage compensation method for optical fiber communication systems according to the present invention.

[0046] Figure 2 This is a schematic diagram of the clock leakage compensation system for optical fiber communication systems according to the present invention.

[0047] Figure 3 This is a diagram illustrating the clock leakage compensation result of an optical fiber communication system according to an embodiment of the present invention. Detailed Implementation

[0048] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.

[0049] Example 1

[0050] The clock leakage compensation method for optical fiber communication systems provided by the present invention, such as Figure 1 As shown, it includes:

[0051] Step S1: Establish a training dataset and construct a learnable digital signal normalization structure;

[0052] Step S2: Train a learnable digital signal normalization structure using the training dataset;

[0053] Step S3: Utilize the trained, learnable digital signal normalization structure to achieve clock leakage compensation and deploy it in the fiber optic communication system.

[0054] Specifically, step S1 employs the following:

[0055] Step S1.1: Establish a training dataset for the clock leakage compensation method;

[0056] Step S1.2: Construct a learnable digital signal normalization structure.

[0057] Specifically, the learnable digital signal normalization structure adopts:

[0058]

[0059] Where, μ x σ is the mean of the input digital signal. x Let k be the average power of the input digital signal, k be the scaling factor, b be the bias sequence, and μ be the bias value. b is the mean of the bias sequence; k and b are both learnable parameters.

[0060] The learnable digital signal normalization structure includes:

[0061] Step S1.2.1: Preprocess the received digital signal;

[0062] Step S1.2.2: Perform DC removal and power normalization on the preprocessed signal;

[0063] Step S1.2.3: Set the learnable parameters k and b, where k is the scaling factor, which is a real number; b is the bias sequence, which is a complex sequence with the same length as the input digital signal.

[0064] Step S1.2.4: Scale the output digital signal from step S1.2.2 to obtain the output of a learnable normalized structure.

[0065] Specifically, step S2 employs the following:

[0066] Step S2.1: Input the input digital signal from the training data into a learnable normalized structure, and after back-end digital signal processing, obtain the output digital signal;

[0067] Step S2.2: Based on the labels of the output digital signal and the training data, construct an error function to calculate the error magnitude, and use gradient backpropagation and gradient descent to train the learnable normalized structure. When the number of training iterations reaches the preset number, the training ends.

[0068] Specifically, step S2.2 employs the following:

[0069] Step S2.2.1: Update the learning parameters of the learnable normalized structure according to the gradient;

[0070] Step S2.2.2: Construct the error function and calculate the error between the output digital signal and the training data;

[0071] Step S2.2.3: Calculate the gradient of the learnable parameters using the backpropagation algorithm; iterate the training multiple times, and the training ends when the preset number of training iterations is reached.

[0072] Specifically, step S2.2.2 involves constructing an error function to calculate the error magnitude, using the following formula:

[0073]

[0074] Where N is the training batch size, y is the digital signal processing output signal, and x is the training batch size. label These are the label symbols for the training data.

[0075] Specifically, step S2.2.3 involves updating the learning parameters of the learnable normalized structure based on the gradient, using the following formula:

[0076]

[0077]

[0078] Among them, b i and k i Let b and k represent the parameters after the i-th update. and Let lr represent the gradients of the error functions corresponding to b and k, respectively. i-1 Let be the learning rate for the (i-1)th iteration.

[0079] The clock leakage compensation system for optical fiber communication systems provided by the present invention, such as Figure 2 As shown, it includes:

[0080] Module M1: Establishes the training dataset and constructs a learnable digital signal normalization structure;

[0081] Module M2: Trains a learnable digital signal normalization structure using the training dataset;

[0082] Module M3: Utilizes a trained, learnable digital signal normalization structure to achieve clock leakage compensation and is deployed in fiber optic communication systems.

[0083] Specifically, module M1 adopts:

[0084] Module M1.1: Establishes the training dataset for the clock leakage compensation method;

[0085] Module M1.2: Constructs a learnable digital signal normalization structure.

[0086] Specifically, the learnable digital signal normalization structure adopts:

[0087]

[0088] Where, μ x σ is the mean of the input digital signal. x Let k be the average power of the input digital signal, k be the scaling factor, b be the bias sequence, and μ be the bias value. b is the mean of the bias sequence; k and b are both learnable parameters.

[0089] The learnable digital signal normalization structure includes:

[0090] Module M1.2.1: Preprocesses the received digital signal;

[0091] Module M1.2.2: Performs DC removal and power normalization on the preprocessed signal;

[0092] Module M1.2.3: Set the learnable parameters k and b, where k is the scaling factor, which is a real number; b is the bias sequence, which is a complex sequence with the same length as the input digital signal.

[0093] Module M1.2.4: Scaling up the output digital signal in module M1.2.2 to obtain a learnable normalized structure output.

[0094] Specifically, module M2 adopts:

[0095] Module M2.1: Inputs the input digital signal from the training data into a learnable normalized structure, and after back-end digital signal processing, obtains the output digital signal;

[0096] Module M2.2: Based on the labels of the output digital signal and training data, construct an error function to calculate the error magnitude, and use gradient backpropagation and gradient descent to train the learnable normalized structure. Training ends when the preset number of training iterations is reached.

[0097] Specifically, module M2.2 adopts:

[0098] Module M2.2.1: Updates the learning parameters of the learnable normalized structure based on the gradient;

[0099] Module M2.2.2: Constructs the error function to calculate the error between the output digital signal and the training data;

[0100] Module M2.2.3: Calculates the gradient of the learnable parameters through the backpropagation algorithm; iterates the training multiple times, and ends when the training count reaches the preset number of times.

[0101] Specifically, module M2.2.2 employs the following method to construct an error function and calculate the error magnitude: The specific formula is as follows:

[0102]

[0103] Where N is the training batch size, y is the digital signal processing output signal, and x is the training batch size. label These are the label symbols for the training data.

[0104] Specifically, module M2.2.3 updates the learning parameters of the learnable normalized structure based on the gradient, using the following formula:

[0105]

[0106]

[0107] Among them, b i and k i Let b and k represent the parameters after the i-th update. and Let lr represent the gradients of the error functions corresponding to b and k, respectively. i-1 Let be the learning rate for the (i-1)th iteration.

[0108] Example 2

[0109] Example 2 is a preferred example of Example 1.

[0110] The present invention also provides a clock leakage compensation result illustration of an optical fiber communication system according to an embodiment. The experimental optical fiber communication system has an optical fiber length of 811 kilometers, dual polarization, single channel, 4-bit / symbol modulation, DAC rate of 80 Gsam / s, transmission rate of 400 Gbit / s, and uses coherent modulation and demodulation. Figure 3 The clock leakage compensation results are demonstrated. In the above scenario, the proposed method and system captured the clock leakage present in the DAC and successfully compensated for this effect, solving the non-ideal effect of clock leakage in optical communication and improving transmission performance.

[0111] The technical problem to be solved by this invention is to achieve clock leakage compensation in optical fiber communication systems. This invention optimizes the parameters of the normalization module through the gradient descent method. This invention designs a learnable normalization structure to compensate for clock leakage effects, improve the signal quality of optical fiber communication, and reduce the bit error rate after transmission. The clock leakage compensation method and system for optical fiber communication systems proposed in this invention can be applied to any channel, and in particular, it is suitable for various optical fiber communication scenarios.

[0112] Those skilled in the art will understand that, in addition to implementing the system, apparatus, and their modules provided by this invention in purely computer-readable program code, the same program can be implemented in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers by logically programming the method steps. Therefore, the system, apparatus, and their modules provided by this invention can be considered a hardware component, and the modules included therein for implementing various programs can also be considered structures within the hardware component; alternatively, modules for implementing various functions can be considered both software programs implementing the method and structures within the hardware component.

[0113] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. A clock leakage compensation method for an optical fiber communication system, characterized by, include Step S1: Establish a training dataset and construct a learnable digital signal normalization structure; Step S2: Train a learnable digital signal normalization structure using the dataset; Step S3: Utilize the trained, learnable digital signal normalization structure to achieve clock leakage compensation and deploy it in the fiber optic communication system; The learnable mathematical signal normalization structure adopts: in, The mean of the input digital signal. The average power of the input digital signal, k The scaling factor. b For bias sequences, The mean of the biased sequence; k and b All of these are learnable parameters; Step S2 employs the following: Step S2.1: Input the input digital signal from the training data into the learnable normalization result to obtain the output digital signal; Step S2.2: Construct an error function based on the labels of the output digital signal and training data to calculate the error magnitude. Use gradient backpropagation and gradient descent to train the learnable normalized structure. When the training reaches a preset number of times, the training result is obtained.

2. The clock leakage compensation method for an optical fiber communication system according to claim 1, characterized in that, The error function includes: in, N For training batch size, y For digital signal processing output signal, These are the label symbols for the training data.

3. The clock leakage compensation method for an optical fiber communication system according to claim 1, characterized in that, The learnable normalized structure is trained using gradient backpropagation and gradient descent, and the learning parameters of the learnable normalized structure are updated, including: in, and Indicates the first i Parameters after the last update b and k , and They represent b and k The gradient of the corresponding error function, For the first i-1 The size of the learning rate.

4. A clock leakage compensation system for an optical fiber communication system, characterized in that, include Module M1: Establishes the training dataset and constructs a learnable digital signal normalization structure; Module M2: Trains a learnable digital signal normalization structure using a dataset; Module M3: Utilizes a trained, learnable digital signal normalization structure to achieve clock leakage compensation, and is deployed in fiber optic communication systems; The learnable mathematical signal normalization structure adopts: in, The mean of the input digital signal. The average power of the input digital signal, k The scaling factor. b For bias sequences, The mean of the biased sequence; k and b All of these are learnable parameters; The module M2 adopts: Module M2.1: Inputs the input digital signal from the training data into a learnable normalized result to obtain the output digital signal; Module M2.2: Constructs an error function based on the labels of the output digital signal and training data to calculate the error magnitude, and trains the learnable normalized structure using gradient backpropagation and gradient descent. When the training reaches a preset number of times, the training result is obtained.

5. The clock leakage compensation system for an optical fiber communication system according to claim 4, characterized in that, The error function includes: in, N For training batch size, y For digital signal processing output signal, These are the label symbols for the training data.

6. The clock leakage compensation system for an optical fiber communication system according to claim 4, characterized in that, The learnable normalized structure is trained using gradient backpropagation and gradient descent, and the learning parameters of the learnable normalized structure are updated, including: in, and Indicates the first i Parameters after the last update b and k , and They represent b and k The gradient of the corresponding error function, For the first i-1 The size of the learning rate.

Images