Method and device for improving polarization tracking speed based on dimension reduction

Abstract

The invention relates to the technical field of communication, and provides a method and device for improving polarization tracking speed based on dimension reduction, and the method comprises the steps: carrying out the sampling of a plurality of periods of a feedback port of a polarization control system, so as to obtain the gradient value of each stage of phase shifter in the polarization control system; in each sampling period, the phase shift amount of each stage of phase shifter is sequentially adjusted according to a preset step length, sampling is carried out once after each pair of stage of phase shifter is adjusted, and the difference value between output power obtained through two adjacent times of sampling is used as the gradient value of the phase shifter; the sampling rate is greater than the change rate of the input polarization state; and processing the gradient value of each stage of phase shifter by using a preset strategy to realize polarization control of the phase shifter. According to the invention, the phase shift amount of each stage of phase shifter is sequentially adjusted, and high-speed sampling is carried out on the feedback port of the polarization control system, so that the number of samples needing to be sampled can be greatly reduced, the sampling rate is improved, the rate of subsequent data processing is improved, and the polarization tracking speed is improved.

Description

Technical Field

[0001] The present invention relates to the field of communication technologies, and particularly to a method and apparatus for improving the polarization tracking speed based on dimensionality reduction. Background Art

[0002] In modern communication systems, polarization control plays an indispensable role. As high-speed optical fiber communication has become an important foundation of modern society and is continuously developing towards miniaturization and integration, the communication system has an increasingly high requirement for the stability of optical signal transmission. Currently, the communication system urgently needs a high-speed polarization controller because of the continuous growth of communication services and the continuous improvement of the requirements for communication quality. Services such as high-definition video transmission and cloud computing data interaction have put forward strict requirements for the stable and high-speed transmission of optical signals. Traditional polarization controllers can no longer meet these needs in terms of speed and accuracy. Only a high-speed polarization controller can effectively overcome the interference caused by the fluctuation of the optical polarization state and ensure the stable and efficient operation of the communication system to meet the growing communication needs of modern society.

[0003] Polarization control is precisely the key link to ensure the stable transmission of optical signals, and it can effectively cope with the fluctuation of the optical polarization state caused by factors such as stress, movement, and vibration on the optical fiber. This kind of fluctuation will cause random changes in the optical intensity and phase of the modulation signal, seriously affecting the communication quality.

[0004] The working principle of the polarization control system is to continuously change the phase shift of each stage of the phase modulation unit according to the randomly changing input polarization state, and then stabilize the polarization state at a certain specific polarization state for output. From the perspective of the Poincaré sphere, it is to stabilize the input polarization state that randomly moves on the sphere, usually at the TE linear polarization state for output.

[0005] Currently, polarization control faces many difficulties. Although high-performance integrated optical devices such as silicon-based phase modulation units and lithium niobate type phase modulation units have gradually replaced the past mechanical extrusion type units, making the tracking speed increase rapidly, the overall polarization control effect is still not satisfactory. For example, although many lithium niobate polarization controllers can achieve a nanosecond-level response, their polarization tracking effect is limited to 10 - 100 Krad / s, which greatly wastes the fast response performance. Therefore, how to improve the utilization rate of performance, achieve a breakthrough in the tracking speed of the polarization controller of Mrad / s, and improve the coordination degree between the polarization controller and the polarization control algorithm has become an urgent problem to be solved currently.

[0006] In view of this, overcoming the defects of the existing technology is an urgent problem to be solved in this technical field. Summary of the Invention

[0007] The technical problem to be solved by the present invention is to provide a method and apparatus for improving the polarization tracking speed based on dimensionality reduction to improve the polarization tracking speed.

[0008] The present invention adopts the following technical solutions:

[0009] In a first aspect, the present invention provides a method for improving the polarization tracking speed based on dimensionality reduction, including:

[0010] Sampling the feedback port of the polarization control system for multiple cycles to obtain the gradient values of the phase shifters at all levels in the polarization control system; wherein, within each sampling cycle, the phase shift amounts of the phase shifters at all levels are sequentially adjusted according to a preset step size. After adjusting each pair of first-level phase shifters, the feedback port is sampled once, and the difference between the output powers obtained from two adjacent samplings is used as the gradient value of the corresponding phase shifter; wherein, the sampling rate of two adjacent samplings is greater than the change rate of the input polarization state.

[0011] Processing the gradient values of the phase shifters at all levels using a preset strategy to achieve polarization control of the phase shifters.

[0012] Preferably, after adjusting each pair of first-level phase shifters, sampling the feedback port once, and using the difference between the output powers obtained from two adjacent samplings as the gradient value of the corresponding phase shifter, specifically includes:

[0013] After adjusting the i-th level phase shifter, performing the (i + 1)-th sampling on the feedback port of the polarization control system.

[0014] Using the output power P obtained from the (i + 1)-th sampling i+1 and the output power P obtained from the i-th sampling i The difference between them is used as the gradient value of the i-th level phase shifter in the current sampling cycle; wherein, before adjusting the first-level phase shifter, performing the first sampling on the feedback port of the polarization control system to obtain the initial output power P1.

[0015] Preferably, processing the gradient values of the phase shifters at all levels using a preset strategy to achieve polarization control of the phase shifters, specifically includes:

[0016] For each cycle of sampling, using the gradient value of the i-th level phase shifter in the current sampling cycle, and calculating the phase shift amount of the i-th level phase shifter in the next sampling cycle in combination with a preset strategy.

[0017] Using the phase shift amount of the next sampling cycle to adjust the i-th level phase shifter for polarization control in the next sampling cycle.

[0018] Using the gradient value of the i-th level phase shifter in the current sampling cycle, and calculating the phase shift amount of the i-th level phase shifter in the next sampling cycle in combination with a preset strategy, specifically includes:

[0019] Using the gradient value of the current sampling cycle to calculate the first moment estimate and the second moment estimate of the current sampling cycle.

[0020] Calibrate the first-order moment estimation to obtain the calibrated first-order moment, and calibrate the second-order moment estimation to obtain the calibrated second-order moment;

[0021] Use the calibrated first-order moment and the calibrated second-order moment to calculate the phase shift amount for the next sampling period.

[0022] Preferably, the use of the gradient value of the current sampling period to calculate the first-order moment estimation and the second-order moment estimation of the current sampling period specifically includes:

[0023] Calculate the first-order moment estimation of the current sampling period as m t =β1m t-1 +(1 - β1)g t , the second-order moment estimation of the current sampling period where β1 and β2 are preset coefficients, g t is the gradient value of the current sampling period, m t-1 is the first-order moment estimation of the previous sampling period, v t-1 is the second-order moment estimation of the previous sampling period.

[0024] Preferably, the calibration of the first-order moment estimation to obtain the calibrated first-order moment and the calibration of the second-order moment estimation to obtain the calibrated second-order moment specifically include:

[0025] Obtain the calibrated first-order moment as The calibrated second-order moment is where m t is the first-order moment estimation of the current sampling period, v t is the second-order moment estimation of the current sampling period, and β1 and β2 are preset coefficients.

[0026] Preferably, the use of the calibrated first-order moment and the calibrated second-order moment to calculate the phase shift amount for the next sampling period specifically includes:

[0027] Calculate the phase shift amount for the next sampling period where η is the learning rate, ε is a preset value, is the calibrated second-order moment of the current sampling period, is the calibrated first-order moment of the current sampling period, θ t is the phase shift amount of the current sampling period.

[0028] Preferably, when the polarization control system includes two-stage silicon-based thermo-optic phase shifters, the Jones matrix of the polarization control system is where, and are the phase shift amounts of each phase shifter, E OUT represents the output light intensity, E FBrepresents the feedback light intensity, E x represents the X - direction polarization of the incident light, E y represents the light intensity of the Y - direction polarization of the incident light.

[0029] Preferably, the structure of each phase shifter is as follows: Among them, is the phase shift amount of the phase shifter;

[0030] In each sampling period, the polarization control system is sampled n + 1 times, where n is the number of phase shifters in the polarization control system.

[0031] In a second aspect, the present invention also provides a device for improving the polarization tracking speed based on dimensionality reduction. The device includes a gradient sampling module and a processing module; and is used to complete the method for improving the polarization tracking speed based on dimensionality reduction according to any one of the first aspects;

[0032] The gradient sampling module is used to sample the feedback port of the polarization control system for multiple periods to obtain the gradient values of each level of phase shifters in the polarization control system; among them, in each sampling period, the phase shift amounts of each level of phase shifters are adjusted sequentially according to a preset step size. After each adjustment of a level of phase shifter, the feedback port is sampled once, and the difference between the output powers obtained from two adjacent samplings is used as the gradient value of the corresponding phase shifter; among them, the sampling rate of two adjacent samplings is greater than the change rate of the input polarization state;

[0033] The processing module is used to process the gradient values of each level of phase shifters using a preset strategy to achieve polarization control of the phase shifters.

[0034] Preferably, the gradient sampling module includes a sampling unit and a gradient calculation unit;

[0035] The sampling unit is used to perform the (i + 1)-th sampling on the feedback port of the polarization control system after adjusting the i - th level of phase shifter;

[0036] The gradient calculation unit is used to use the output power P i+1 obtained from the (i + 1)-th sampling and the output power P i obtained from the i - th sampling, and the difference between them is used as the gradient value of the i - th level of phase shifter in the current sampling period; among them, before adjusting the first - level phase shifter, the feedback port of the polarization control system is sampled for the first time to obtain the initial output power P1.

[0037] Preferably, the processing module includes a phase shift amount calculation unit and a control unit;

[0038] The phase shift amount calculation unit is configured to calculate, for each sampling period, the phase shift amount of the i-th phase shifter in the next sampling period by using the gradient value of the i-th phase shifter in the current sampling period and combining with a preset strategy;

[0039] The control unit is configured to adjust the i-th phase shifter by using the phase shift amount of the next sampling period for polarization control in the next sampling period;

[0040] Wherein, calculating the phase shift amount of the i-th phase shifter in the next sampling period by using the gradient value of the i-th phase shifter in the current sampling period and combining with a preset strategy specifically includes:

[0041] Calculating the first moment estimate and the second moment estimate of the current sampling period by using the gradient value of the current sampling period;

[0042] Correcting the first moment estimate to obtain the corrected first moment, and correcting the second moment estimate to obtain the corrected second moment;

[0043] Calculating the phase shift amount of the next sampling period by using the corrected first moment and the corrected second moment.

[0044] Preferably, calculating the first moment estimate and the second moment estimate of the current sampling period by using the gradient value of the current sampling period specifically includes:

[0045] Calculating the first moment estimate of the current sampling period as m t =β1m t-1 +(1 - β1)g t , the second moment estimate v t =β2v t-1 +(1 - β2)g t 2 ; wherein, β1 and β2 are preset coefficients, g t is the gradient value of the current sampling period, m t-1 is the first moment estimate of the previous sampling period, and v t-1 is the second moment estimate of the previous sampling period.

[0046] Preferably, correcting the first moment estimate to obtain the corrected first moment, and correcting the second moment estimate to obtain the corrected second moment specifically includes:

[0047] Obtaining the corrected first moment as The corrected second moment is Wherein, m t is the first moment estimate of the current sampling period, v t is the second moment estimate of the current sampling period, and β1 and β2 are preset coefficients.

[0048] Preferably, calculating the phase shift amount for the next sampling period by using the corrected first moment and the corrected second moment specifically includes:

[0049] Calculating the phase shift amount for the next sampling period where η is the learning rate and ε is a preset value. is the corrected second moment for the current sampling period. is the corrected first moment for the current sampling period, and θ t is the phase shift amount for the current sampling period.

[0050] Preferably, when two-stage silicon-based thermo-optic phase shifters are included in the polarization control system, the Jones matrix of the polarization control system is where and are the phase shift amounts of the phase shifters at all levels, E OUT represents the output light intensity, E FB represents the feedback light intensity, E x represents the light polarized in the X direction of the incident light, and E y represents the light intensity of the incident light polarized in the Y direction.

[0051] Preferably, the structure of each phase shifter is: where is the phase shift amount of the phase shifter;

[0052] In each sampling period, the polarization control system is sampled n + 1 times, where n is the number of phase shifters in the polarization control system.

[0053] In a third aspect, the present invention further provides a device for improving the polarization tracking speed based on dimensionality reduction, which is used to implement the method for improving the polarization tracking speed based on dimensionality reduction described in the first aspect. The device includes:

[0054] At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions, when executed by the processor, are used to execute the method for improving the polarization tracking speed based on dimensionality reduction described in the first aspect.

[0055] In a fourth aspect, the present invention further provides a non-volatile computer storage medium, which stores computer-executable instructions, and the computer-executable instructions, when executed by one or more processors, are used to complete the method described in the first aspect.

[0056] In a fifth aspect, a chip is provided, including: a processor and an interface, which are used to call and run a computer program stored in a memory and execute the method as described in the first aspect.

[0057] In a sixth aspect, a computer program product including instructions is provided, which, when the instructions are run on a computer or a processor, cause the computer or the processor to execute the method as in the first aspect.

[0058] By sequentially adjusting the phase shift amounts of phase shifters at all levels and performing high-speed sampling on the feedback ports of the polarization control system, the present invention can greatly reduce the number of required sampling samples, thereby improving the sampling rate on the one hand and the subsequent data processing rate on the other hand, and thus improving the polarization tracking speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required to be used in the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only 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.

[0060] Figure 1 is a flowchart showing a method for improving polarization tracking speed based on dimensionality reduction provided by an embodiment of the present invention;

[0061] Figure 2 is a flowchart showing a method for improving polarization tracking speed based on dimensionality reduction provided by an embodiment of the present invention;

[0062] Figure 3 is a flowchart showing a method for improving polarization tracking speed based on dimensionality reduction provided by an embodiment of the present invention;

[0063] Figure 4 is a flowchart showing a method for improving polarization tracking speed based on dimensionality reduction provided by an embodiment of the present invention;

[0064] Figure 5 is a schematic diagram of the architecture of a polarization control system provided by an embodiment of the present invention;

[0065] Figure 6 is a schematic diagram of a method for improving polarization tracking speed based on dimensionality reduction provided by an embodiment of the present invention;

[0066] Figure 7 is a comparative schematic diagram of the gradient descent process of the prior art and the gradient descent process of the method for improving polarization tracking speed based on dimensionality reduction provided by an embodiment of the present invention;

[0067] Figure 8 is a comparative effect schematic diagram before and after polarization control in the prior art;

[0068] Figure 9It is a schematic diagram of the comparison effect before and after polarization control by a method for improving polarization tracking speed based on dimensionality reduction provided by an embodiment of the present invention;

[0069] Figure 10 It is a schematic diagram of the architecture of a device for improving polarization tracking speed based on dimensionality reduction provided by an embodiment of the present invention. Detailed implementation manners

[0070] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0071] Unless otherwise required by the context, throughout the specification and claims, the term "including" is interpreted in an open, inclusive sense, i.e., "including, but not limited to". In the description of the specification, the terms "one embodiment", "some embodiments", "exemplary embodiments", "examples", "specific examples" or "some examples" etc. are intended to indicate that specific features, structures, materials or characteristics related to the embodiment or example are included in at least one embodiment or example of the present disclosure. The schematic representations of the above terms are not necessarily referring to the same embodiment or example. In addition, the specific features, structures, materials or characteristics may be included in any one or more embodiments or examples in any appropriate manner, that is, although they may be carried in the above-mentioned embodiments or examples due to reasons such as the order and position of appearance, they are not limited to being carried by one embodiment or example in a combined manner.

[0072] In the description of the present invention, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly indicating the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the embodiments of the present disclosure, unless otherwise stated, the meaning of "a plurality" is two or more. In addition, for example, in the description, for the same type of nouns, the method of adding "A" and "B" at the end is used to describe them as two independent individuals. In this case, the features defined with "A" and "B" are only used for the purpose of distinguishing similar individuals and cannot be construed as indicating or implying relative importance or implicitly indicating the quantity of the indicated technical features.

[0073] In the description of the present invention, there will be a description of the form "A and / or B" (where A and B are used to formally represent specific feature contents), and the corresponding description methods include the following three combinations: only A, only B, and the combination of A and B.

[0074] As used in the present invention, "about", "substantially" or "approximately" includes the stated value and the average value within an acceptable deviation range of the specific value, where the acceptable deviation range is determined by a person of ordinary skill in the art considering the measurement being discussed and the error associated with the measurement of a specific quantity (i.e., the limitations of the measurement system).

[0075] In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0076] Example 1:

[0077] Due to the carrier fading phenomenon caused by the random fluctuation of the polarization state of the local oscillator light in the relevant optical communication system, a polarization controller is required to track the polarization state. However, in harsh and extreme environments, such as lightning strikes, etc., the polarization state in the optical fiber will rotate at a speed of Mrad / s. Therefore, a high-speed automatic polarization controller is of great significance. In recent years, thin-film lithium niobate polarization controllers have attracted extensive attention from researchers due to their ultra-fast electro-optic response speed. However, although the response time of the lithium niobate polarization controller has reached the nanosecond level, the polarization tracking rate of the lithium niobate polarization controller is still not high. To solve this problem, Example 1 of the present invention provides a method for improving the polarization tracking speed based on dimensionality reduction, as Figure 1 shown, including:

[0078] In step 201, multiple cycles of sampling are performed on the feedback port of the polarization control system to obtain the gradient values of the phase shifters at all levels in the polarization control system; wherein, within each sampling cycle, the phase shift amounts of the phase shifters at all levels are adjusted in sequence according to a preset step size. After adjusting each pair of first-level phase shifters, a sampling is performed on the feedback port, and the difference between the output powers obtained from two adjacent samplings is used as the gradient value of the corresponding phase shifter; wherein, the sampling rate between two adjacent samplings is greater than the change rate of the input polarization state; in actual use, multiple samplings within a sampling cycle can be implemented according to a preset sampling rate, and the preset sampling rate is obtained by a person skilled in the art through requirement analysis. The preset step size is obtained by a person skilled in the art through empirical analysis. Adjusting the phase shift amounts of the phase shifters at all levels according to the preset step size can be: increasing or decreasing the phase shift amount of the phase shifter by the preset step size.

[0079] In step 202, a preset strategy is used to process the gradient values of the phase shifters at all levels to achieve polarization control of the phase shifters. Among them, there are often multiple phase shifters in a polarization control system, and the feedback port is the output port after passing through all the phase shifters. In each sampling cycle, the polarization control system is sampled n + 1 times, where n is the number of phase shifters in the polarization control system, and the first sampling is performed before the phase shifters are adjusted.

[0080] For example, assume that T is the time length of a sampling period. If the method described in this embodiment starts to be executed at the beginning of the t0 sampling period, then the first sampling period is from t0 to t0 + T, the second sampling period is from t0 + T to t0 + 2T, the third sampling period is from t0 + 2T to t0 + 3T, and so on. Within the first sampling period from t0 to t0 + T, n + 1 samplings are performed. For example: within the first sampling period from t0 to t0 + T, the first sampling is carried out first. After the first sampling is completed, the phase shift amount of the first-stage phase shifter is increased by a preset step size, and the second sampling is performed. And so on, then the phase shift amount of the second-stage phase shifter is increased by the preset step size, and the third sampling is performed until after the phase shift amount of the last-stage phase shifter is increased by the preset step size and the sampling is carried out, the first sampling period ends. Among them, the time interval between every two samplings is less than the preset interval, that is, the sampling rate of adjacent two samplings is greater than the change rate of the input polarization state, so that the input polarization state of adjacent two samplings does not change.

[0081] In this embodiment, by sequentially adjusting the phase shift amounts of each stage of phase shifters and performing high-speed sampling on the feedback port of the polarization control system (the sampling speed is higher than the change speed of the input polarization state), it can be approximated that the input polarization states within the two sampling intervals are the same. Thus, the difference between the output powers obtained from two adjacent samplings can be used as the change value of the output power of the phase shifter, that is, the gradient value. In this way, dimensionality reduction is achieved, and only n + 1 samplings are required in one sampling period, where n is the number of phase shifters in the polarization control system. That is, in one sampling period, an initial value and the output powers after the step size adjustments of each phase shifter are sampled. Compared with the prior art where 2 n samplings are required in one sampling period, this embodiment can greatly reduce the number of sampling samples required. Thus, on the one hand, the sampling rate is improved, and on the other hand, the rate of subsequent data processing is improved, thereby improving the polarization tracking speed.

[0082] In a specific application scenario, after each pair of first-stage phase shifters is adjusted, the feedback port is sampled once, and the difference between the output powers obtained from two adjacent samplings is used as the gradient value of the corresponding phase shifter. As Figure 2 shown, it specifically includes:

[0083] In step 301, after the i-th stage phase shifter is adjusted and before the (i + 1)-th stage phase shifter is adjusted, the (i + 1)-th sampling of the feedback port of the polarization control system is performed;

[0084] In step 302, the difference between the output power P i+1 obtained from the (i + 1)-th sampling and the output power P i obtained from the i-th sampling is used as the gradient value of the i-th stage phase shifter within the current sampling period, that is, the gradient value D i= P i+1 -P i ; where, before adjusting the first - stage phase shifter, the feedback port of the polarization control system is sampled for the first time to obtain the initial output power P1, and the gradient value D1 of the first - stage phase shifter is D1 = P2 - P1.

[0085] In an optional implementation manner, the gradient values of each - stage phase shifters are processed using a preset strategy to achieve polarization control of the phase shifter, as Figure 3 shown, specifically including:

[0086] In step 401, for each sampling period, the gradient value of the i - th stage phase shifter in the current sampling period is used, and in combination with the preset strategy, the phase - shift amount of the i - th stage phase shifter in the next sampling period is calculated.

[0087] In step 402, the phase - shift amount of the next sampling period is used to adjust the i - th stage phase shifter for polarization control in the next sampling period.

[0088] The above steps 401 - step 402 can be understood as: setting the phase - shift amount of the i - th stage phase shifter calculated for each sampling period to the i - th stage phase shifter, so as to use this phase - shift amount to perform phase - shift on the subsequent signal. Then, in the next sampling period, this step is repeated. Thus, through continuous iterative optimization, after a period of time, the phase - shift amount of the i - th stage phase shifter gradually converges, so as to stabilize the output polarization state at the TE linear polarization state for output. It should be noted here that the above steps 401 - step 402 are described for one phase shifter (i.e., the i - th stage phase shifter). In actual use, the operations described in steps 401 - step 402 need to be performed for each phase shifter in the offset control system, and the calculated phase - shift amount is set to the phase shifter after the end of one sampling period. Within one sampling period, the change of the phase - shift amount of each phase shifter only depends on the preset step size.

[0089] In an actual application scenario, using the gradient value of the i - th stage phase shifter in the current sampling period, and in combination with the preset strategy to calculate the phase - shift amount of the i - th stage phase shifter in the next sampling period, as Figure 4 shown, specifically including:

[0090] In step 501, using the gradient value of the current sampling period, the first - order moment estimate and the second - order moment estimate of the current sampling period are calculated.

[0091] In step 502, the first - order moment estimate is corrected to obtain the corrected first - order moment, and the second - order moment estimate is corrected to obtain the corrected second - order moment.

[0092] In step 503, the phase shift amount for the next sampling period is calculated using the corrected first moment and the corrected second moment. The above steps 501 to 503 are all described for the i-th stage phase shifter. For the sake of simplicity of expression, in this embodiment, the "gradient value of the current sampling period" is directly used as an alternative expression for the "gradient value of the i-th stage phase shifter in the current sampling period", and the "first moment estimate of the current sampling period" is used as an alternative expression for the "first moment estimate of the i-th stage phase shifter in the current sampling period", and so on.

[0093] Among them, the calculation of the first moment estimate and the second moment estimate of the current sampling period using the gradient value of the current sampling period specifically includes: calculating that the first moment estimate of the current sampling period is m t = β1m t-1 +(1 - β1)g t , and the second moment estimate of the current sampling period Among them, β1 and β2 are preset coefficients, g t is the gradient value of the current sampling period, m t-1 is the first moment estimate of the previous sampling period, and v t-1 is the second moment estimate of the previous sampling period. Both β1 and β2 are obtained by those skilled in the art through empirical analysis.

[0094] The first moment estimate of the initial sampling period is a vector of all zeros, and the second moment estimate of the initial sampling period is a vector of all zeros. That is, in the first sampling period, that is, when t = 1, v0 is a vector of all zeros, and m0 is a vector of all zeros.

[0095] The correction of the first moment estimate to obtain the corrected first moment and the correction of the second moment estimate to obtain the corrected second moment specifically include: obtaining the corrected first moment as The corrected second moment is Among them, m t is the first moment estimate of the current sampling period, v t is the second moment estimate of the current sampling period, and β1 and β2 are preset coefficients.

[0096] The calculation of the phase shift amount for the next sampling period using the corrected first moment and the corrected second moment specifically includes: calculating the phase shift amount for the next sampling period Among them, η is the learning rate, ε is a preset value, is the corrected second moment of the current sampling period, is the corrected first moment of the current sampling period, and θ t is the phase shift amount of the current sampling period. In the first sampling period, that is, when t = 1, θ1 is obtained by those skilled in the art through empirical analysis, that is, the phase shift amount initially set for the corresponding phase shifter.

[0097] In a specific application scenario, the structures of phase shifters at all levels are as follows: Among them, is the phase shift amount of the phase shifter. This 2×2 matrix represents the Jones matrix of the phase shifter. For example, represents the Jones matrix of the 3dB coupler in the polarization controller, that is, the phase shifter and the 3dB coupler are alternately arranged to form a multi-stage thermally tuned polarization controller.

[0098] When there are two levels of silicon-based thermally tuned phase shifters in the polarization control system, the Jones matrix of the polarization control system is Among them, and are the phase shift amounts of the phase shifters at all levels. E OUT represents the output light intensity, E FB represents the feedback light intensity, E x represents the polarization of the incident light in the X direction, and E y represents the light intensity of the polarization of the incident light in the Y direction.

[0099] In this embodiment, by using the historical cumulative first moment, second moment, and the current gradient, the gradient direction close to the current true value is calculated, and combined with the modulo function, high-speed and high-precision polarization control without reset can be achieved, realizing a polarization tracking effect with higher speed and better tracking effect.

[0100] Embodiment 2:

[0101] Based on the method described in Embodiment 1, the present invention combines a specific application scenario and uses the technical expressions in the relevant scenario to elaborate the implementation process in the characteristic scenario of the present invention.

[0102] This embodiment takes the four-level polarization controller structure as shown in Figure 5 as an example. In this embodiment, a field-programmable gate array is used to execute a method for improving the polarization tracking speed based on dimensionality reduction, specifically including:

[0103] 1. Perform high-speed sampling on the feedback port of the polarization control system (the sampling speed is higher than the change speed of the input polarization state), then it can be approximated that the input polarization state is the same within two sampling intervals, thereby reducing the n+1-dimensional function to an n-dimensional function.

[0104] 2. Use the difference between the output powers of the i-th phase shifter obtained from two adjacent samplings as the gradient Di of the i-th phase shifter.

[0105] 3. Process the gradient information using a preset strategy to find the direction that best fits the actual gradient of the function in n dimensions.

[0106] The polarization control process includes:

[0107] Among them, the control principle of the polarization controller is specifically as follows: Any polarized light is input into the adaptive polarization controller. The PSR at the input port of the polarization controller converts the any polarized light into the X-direction polarization state and the Y-direction polarization state of the TE mode. After passing through the multi-stage phase shifter optical waveguide structure, phase modulation is first performed, and then interference occurs, and finally output light and feedback light are obtained. According to the intensity of the feedback light, the additional phase φ of each stage of the phase shifter is changed by using a feedback algorithm, so that the output light is in a determined polarization state of the TE mode and has the maximum light intensity.

[0108] The so-called high-speed sampling specifically includes:

[0109] According to the above polarization control process, it can be abstracted into an optimization process of a multi-dimensional loss function. The independent variables of the multi-dimensional loss function are the input polarization state and the phase shift amounts φ1-φn of the n-stage phase shifters, and the dependent variable is the output polarization state. By performing high-speed sampling on the feedback port, and the sampling speed is higher than the change speed of the external input polarization state, it can be approximately considered that the input polarization states of two samplings are the same, and then the above n+1-dimensional loss function can be reduced to an n-dimensional one, excluding the interference of the input polarization state on the polarization control system.

[0110] Taking the difference between the output powers obtained by the i-th stage phase shifter in two adjacent samplings as the gradient Di of the i-th stage phase shifter specifically includes:

[0111] 1) Record the optical power of the current feedback port.

[0112] 2) Increase or decrease the i-th stage phase shifter by a small step (i.e., the preset step).

[0113] 3) Sample the change in the optical power of the feedback port after step 2), calculate the difference between the two samplings, and thus obtain Di.

[0114] 4) Repeat the above steps, and obtain the gradients D1-Dn of the output polarization state with respect to the 1st - nth stage phase shifters through multiple high-speed samplings.

[0115] In an embodiment, the structure of the polarization controller is specifically:

[0116]

[0117] Among them, is the phase shift amount of the phase shifter, and the above 2×2 matrix represents the Jones matrix of the phase shifter.

[0118] For example:

[0119] The above 2×2 matrix represents the Jones matrix of the 3dB coupler in the polarization control device, and the phase shifters and 3dB couplers are alternately arranged and combined into a multi-stage thermally tunable phase polarization controller.

[0120] For example, the Jones matrix of a two-stage silicon-based thermal phase shifter can be written as:

[0121]

[0122] where E OUT represents the output light intensity, E FB represents the feedback light intensity, and E x E y represent the light intensities of the incident light polarized in the X direction and the Y direction respectively, represent the phase shifts of the two-stage phase shifters respectively.

[0123] Processing the gradient information using a preset strategy to find the direction that best fits the actual gradient of the function in n dimensions, specifically including:

[0124] 1) First, set initial values for some key variables of the algorithm. Set the variable used to record the first-order moment estimate of the gradient (similar to the gradient mean) to a vector of all 0s, and set the variable used to record the second-order moment estimate of the gradient (similar to the uncentered variance) to a vector of all 0s. At the same time, set the time step of the iteration to 0.

[0125] 2) Calculate the gradient of the current i-th level according to the current model and data.

[0126] 3) Use the gradient of the current i-th level calculated in the previous step to update the first-order moment estimate. It synthesizes the previous and current gradient information and can reflect the general trend of the gradient.

[0127] 4) Similarly, use the gradient of the current i-th level to update the second-order moment estimate to obtain the second-order moment estimate of the current sampling period, which reflects the change amplitude of the gradient. The obtained first-order moment and second-order moment are as follows:

[0128] mt = β1mt-1+(1-β1)gt

[0129]

[0130] where mt is the historical cumulative first-order moment estimate, vt is the historical cumulative second-order moment estimate, β1 and β2 represent two coefficients respectively, and gt is the current gradient.

[0131] Through specific calculation methods, correct the first-order moment estimate and the second-order moment estimate respectively to obtain the corrected first-order moment estimate and second-order moment estimate, as follows:

[0132]

[0133] Update the parameters of the model according to the learning rate, the corrected first-order moment estimate, and the corrected second-order moment estimate.

[0134]

[0135] Where η is the learning rate and ε is a very small value, whose function is to prevent the denominator from being zero.

[0136] 5) Repeat the above steps separately in different dimensions. As the number of iterations increases, the parameters of the model will continuously update in the direction of minimizing the multi-dimensional loss function.

[0137] Different from the existing multi-dimensional gradient descent algorithm, that is, a total of 2^n + 1 samplings are performed on the n-level phase shifter. As Figure 6 shown, the number of samplings in a single cycle of this implementation is n + 1 times, which are respectively the initial value recording and the gradient calculations in n directions, greatly reducing the number of samplings, thereby increasing the speed limit of the algorithm.

[0138] Moreover, this embodiment uses a multi-dimensional algorithm. As Figure 7 shown, this is because there is a gradient direction closest to the actual change of the external input polarization state in the gradient descent algorithm. Controlling only in a single dimension will lead to improper convergence. The multi-dimensional algorithm can make full use of the gradients measured by the n-level phase shifter for calculation, making the tracking gradient direction of the polarization control system closer to the actual change gradient, thereby significantly reducing the number of iterations and improving the polarization tracking effect. This embodiment adopts a preset strategy to process the gradient information, so as to find the direction that best fits the actual gradient of the function in n dimensions. Multi-dimensional gradient descent can update the parameters by using the gradient information of multiple variables at the same time. It can move more accurately in the direction of the minimum value of the objective function. Compared with one-dimensional gradient descent that searches along only one dimension each time, multi-dimensional gradient descent can find the path to the optimal solution more directly, so it usually has a faster convergence speed.

[0139] As Figure 8 and Figure 9 shown, the left side is the high-speed polarization perturbation input, and the trajectory on the right is the position of the polarization state output after algorithm control on the Poincaré sphere. Figure 8 is the comparison effect diagram before and after the treatment of the prior art. Figure 9 is the comparison effect diagram before and after the treatment of the method described in this embodiment. Through comparison, it can be seen that the convergence ability of the polarization controller is greatly improved by dimensionality reduction, compared with the long-term unlocking of the fixed-step gradient descent algorithm, and dimensionality reduction has achieved a leap in control accuracy and tracking speed.

[0140] The present invention introduces a multi-dimensional processing strategy, which improves the convergence speed and convergence accuracy of the polarization controller. In contrast, it reduces the sampling times of the previous multi-dimensional gradient descent algorithm, thereby enhancing the tracking speed. Furthermore, by calculating the historical cumulative first moment, second moment, and the current gradient, a gradient direction closer to the current true value is obtained. In combination with the modulo function, high-speed and high-precision polarization control without reset can be achieved.

[0141] In this embodiment, multi-dimensional function dimensionality reduction is achieved through high-speed sampling, excluding the influence of the input polarization state on the polarization control process. After obtaining richer gradient information compared to one-dimensional gradient descent through multi-dimensional gradient measurement, and by combining historical gradient information with current gradient information, the convergence speed and convergence accuracy of the polarization controller are improved. In contrast, it reduces the sampling times of the previous multi-dimensional gradient descent algorithm, thereby enhancing the speed upper limit of the algorithm. In combination with the modulo function, high-speed and high-precision polarization control without reset can be achieved. It can be foreseen that the method proposed in this embodiment has great application prospects in actual industrial applications.

[0142] This embodiment also provides an experimental device and evaluation criteria for measuring the polarization tracking control effect, including:

[0143] Laser: As the light source input.

[0144] Polarization scrambler: Randomly scramble the polarization state of the light output from the light source.

[0145] Erbium-doped optical amplifier: Amplify the light with a random polarization state.

[0146] Oscilloscope: Observe and record the optical signals converted by the photodetector at the output and feedback ports.

[0147] Polarization controller: Silicon-based polarization controller or lithium niobate polarization controller.

[0148] Control circuit: Comprising a digital-to-analog / analog-to-digital converter, a photodetector, a transimpedance amplifier, and a field-programmable gate array for signal processing.

[0149] The evaluation criterion for polarization control adopts the relative intensity error RIE, and its calculation formula is:

[0150]

[0151] where I FB is the feedback light intensity, and I OUT is the light intensity at the output port. And the complementary cumulative distribution function 1 - F(RIE) (the probability that RIE becomes worse than the value given on the horizontal axis, and F(RIE) is calculated by the proportion of the number of all points where RIE is less than a certain value in this group of data to the sample size).

[0152] Embodiment 3:

[0153] As shown in Figure 10 , it is a schematic structural diagram of the device for improving the polarization tracking speed based on dimensionality reduction according to an embodiment of the present invention. The device for improving the polarization tracking speed based on dimensionality reduction in this embodiment includes one or more processors 21 and a memory 22. Among them, Figure 10 One processor 21 is taken as an example herein.

[0154] The processor 21 and the memory 22 can be connected through a bus or other means, Figure 10 Taking connection through a bus as an example herein.

[0155] The memory 22, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs and non-volatile computer-executable programs, such as the method for improving the polarization tracking speed based on dimensionality reduction in Embodiment 1. The processor 21 executes the method for improving the polarization tracking speed based on dimensionality reduction by running the non-volatile software programs and instructions stored in the memory 22.

[0156] The memory 22 may include a high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage devices. In some embodiments, the memory 22 may optionally include a memory remotely provided with respect to the processor 21, and these remote memories can be connected to the processor 21 through a network. Examples of the above-mentioned network include but are not limited to the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0157] The program instructions / modules are stored in the memory 22, and when executed by the one or more processors 21, execute the method for improving the polarization tracking speed based on dimensionality reduction in Embodiment 1 above.

[0158] It should be noted that for the content such as information interaction and execution process among the modules and units within the above-mentioned device and system, since it is based on the same concept as the method embodiment of the present invention, the specific content can be referred to the description in the method embodiment of the present invention and will not be elaborated herein.

[0159] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the embodiments can be completed by instructing relevant hardware through a program, and this program can be stored in a computer-readable storage medium. The storage medium may include: a read-only memory (ROM, Read Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk, an optical disk, or the like.

[0160] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for improving the polarization tracking speed based on dimensionality reduction, characterized in that, Including: Sampling the feedback port of the polarization control system for multiple cycles to obtain the gradient values of the phase shifters at all levels in the polarization control system; wherein, within each sampling cycle, the phase shift amounts of the phase shifters at all levels are adjusted sequentially according to a preset step size, and after adjusting each pair of first-level phase shifters, the feedback port is sampled once, and the difference between the output powers obtained from two adjacent samplings is used as the gradient value of the corresponding phase shifter; wherein, the sampling rate of two adjacent samplings is greater than the change rate of the input polarization state; Processing the gradient values of the phase shifters at all levels using a preset strategy to achieve polarization control of the phase shifters.

2. The method for improving the polarization tracking speed based on dimensionality reduction according to claim 1, wherein The step of, after adjusting each pair of first-level phase shifters, sampling the feedback port once and using the difference between the output powers obtained from two adjacent samplings as the gradient value of the corresponding phase shifter specifically includes: After adjusting the i-th level phase shifter, performing the (i + 1)-th sampling on the feedback port of the polarization control system; The output power P obtained from the (i + 1)-th sampling i+1 and the output power P obtained from the i-th sampling i The difference between them is used as the gradient value of the i-th phase shifter within the current sampling period; among them, before adjusting the first-stage phase shifter, the feedback port of the polarization control system is sampled for the first time to obtain the initial output power P1.

3. The method for improving the polarization tracking speed based on dimensionality reduction according to claim 1, wherein The step of processing the gradient values of the phase shifters at all levels using a preset strategy to achieve polarization control of the phase shifters specifically includes: For each cycle of sampling, using the gradient value of the i-th level phase shifter in the current sampling cycle and combining with the preset strategy to calculate the phase shift amount of the i-th level phase shifter in the next sampling cycle; Using the phase shift amount of the next sampling cycle to adjust the i-th level phase shifter for polarization control in the next sampling cycle; The step of using the gradient value of the i-th level phase shifter in the current sampling cycle and combining with the preset strategy to calculate the phase shift amount of the i-th level phase shifter in the next sampling cycle specifically includes: Using the gradient value of the current sampling cycle to calculate the first moment estimate and the second moment estimate of the current sampling cycle; Correcting the first moment estimate to obtain the corrected first moment, and correcting the second moment estimate to obtain the corrected second moment; Using the corrected first moment and the corrected second moment to calculate the phase shift amount of the next sampling cycle.

4. The method for improving the polarization tracking speed based on dimensionality reduction according to claim 3, wherein The step of using the gradient value of the current sampling cycle to calculate the first moment estimate and the second moment estimate of the current sampling cycle specifically includes: The first moment estimate for the current sampling period is calculated as m t = β1m t-1 +(1 - β1)g t , and the second moment estimate for the current sampling period where β1 and β2 are preset coefficients, and g t is the gradient value for the current sampling period, m t-1 is the first moment estimate for the previous sampling period, and v t-1 is the second moment estimate for the previous sampling period.

5. The method for improving the polarization tracking speed based on dimensionality reduction according to claim 3, wherein The step of correcting the first moment estimate to obtain the corrected first moment and correcting the second moment estimate to obtain the corrected second moment specifically includes: The corrected first moment is The corrected second moment is where m t is the first moment estimate of the current sampling period, v t is the second moment estimate of the current sampling period, and β1 and β2 are preset coefficients.

6. The method for improving the polarization tracking speed based on dimensionality reduction according to claim 3, wherein The step of using the corrected first moment and the corrected second moment to calculate the phase shift amount of the next sampling cycle specifically includes: Calculate the phase shift amount for the next sampling period where η is the learning rate and ε is a preset value, is the corrected second moment for the current sampling period, is the corrected first moment for the current sampling period, and θ t is the phase shift amount for the current sampling period.

7. The method for improving the polarization tracking speed based on dimensionality reduction according to claim 1, characterized in that When two - stage silicon - based thermo - phase modulators are included in the polarization control system, the Jones matrix of the polarization control system is where and are the phase - shift amounts of each stage of the phase shifter, E OUT represents the output light intensity, E FB represents the feedback light intensity, E x represents the light polarized in the X - direction of the incident light, and E y represents the light intensity of the incident light polarized in the Y - direction.

8. The method for improving the polarization tracking speed based on dimensionality reduction according to any one of claims 1 to 7, characterized in that The structures of the phase shifters at all levels are as follows: wherein, is the phase shift amount of the phase shifter; In each sampling cycle, performing n + 1 samplings on the polarization control system, where n is the number of phase shifters in the polarization control system.

9. A non-volatile computer storage medium, characterized in that The computer storage medium stores computer-executable instructions, which are executed by one or more processors to complete the method for improving the polarization tracking speed based on dimensionality reduction according to any one of claims 1 to 8.

10. A device for improving the polarization tracking speed based on dimensionality reduction, characterized in that, The device includes a gradient sampling module and a processing module; The gradient sampling module is used to sample the feedback port of the polarization control system for multiple cycles to obtain the gradient values of the phase shifters at all levels in the polarization control system. Among them, within each sampling cycle, the phase shift amounts of the phase shifters at all levels are adjusted in sequence according to a preset step size. After each adjustment of a pair of first-level phase shifters, the feedback port is sampled once, and the difference between the output powers obtained from two adjacent samplings is used as the gradient value of the corresponding phase shifter. Among them, the sampling rate of two adjacent samplings is greater than the change rate of the input polarization state. The processing module is used to process the gradient values of the phase shifters at all levels using a preset strategy to achieve polarization control of the phase shifters.

11. The device for improving the polarization tracking speed based on dimensionality reduction according to claim 10, characterized in that, The gradient sampling module includes a sampling unit and a gradient calculation unit. The sampling unit is used to perform the (i + 1)-th sampling on the feedback port of the polarization control system after adjusting the i-th phase shifter. The gradient calculation unit is used to use the output power P obtained from the (i + 1)-th sampling i+1 and the output power P obtained from the i-th sampling i The difference between them is used as the gradient value of the i-th phase shifter in the current sampling period; among them, before adjusting the first-stage phase shifter, the feedback port of the polarization control system is sampled for the first time to obtain the initial output power P1.

12. The apparatus for enhancing polarization tracking speed based on dimensionality reduction according to claim 10, wherein The processing module includes a phase shift amount calculation unit and a control unit. The phase shift amount calculation unit is used to, for each cycle of sampling, use the gradient value of the i-th phase shifter in the current sampling cycle and combine it with a preset strategy to calculate the phase shift amount of the i-th phase shifter in the next sampling cycle. The control unit is used to use the phase shift amount of the next sampling cycle to adjust the i-th phase shifter for polarization control in the next sampling cycle. Among them, the using the gradient value of the i-th phase shifter in the current sampling cycle and combining it with a preset strategy to calculate the phase shift amount of the i-th phase shifter in the next sampling cycle specifically includes: Using the gradient value of the current sampling cycle to calculate the first moment estimate and the second moment estimate of the current sampling cycle. Correcting the first moment estimate to obtain the corrected first moment, and correcting the second moment estimate to obtain the corrected second moment. Using the corrected first moment and the corrected second moment to calculate the phase shift amount of the next sampling cycle.

13. The device for improving the polarization tracking speed based on dimensionality reduction according to claim 12, characterized in that, The using the gradient value of the current sampling cycle to calculate the first moment estimate and the second moment estimate of the current sampling cycle specifically includes: The first moment estimate for the current sampling period is calculated as m t = β1m t-1 + (1 - β1)g t , and the second moment estimate for the current sampling period where β1 and β2 are preset coefficients, and g t is the gradient value for the current sampling period, m t-1 is the first moment estimate for the previous sampling period, and v t-1 is the second moment estimate for the previous sampling period.

14. The device for improving the polarization tracking speed based on dimensionality reduction according to claim 12, wherein The correcting the first moment estimate to obtain the corrected first moment, and correcting the second moment estimate to obtain the corrected second moment specifically includes: The corrected first moment is The corrected second moment is where m t is the first moment estimate of the current sampling period, v t is the second moment estimate of the current sampling period, and β1 and β2 are preset coefficients.

15. The device for improving the polarization tracking speed based on dimensionality reduction according to claim 12, wherein The using the corrected first moment and the corrected second moment to calculate the phase shift amount of the next sampling cycle specifically includes: Calculate the phase shift amount for the next sampling period where η is the learning rate and ε is a preset value is the corrected second moment for the current sampling period is the corrected first moment for the current sampling period, and θ t is the phase shift amount for the current sampling period 16. The device for improving the polarization tracking speed based on dimensionality reduction according to claim 10, wherein When two - stage silicon - based thermo - phase modulators are included in the polarization control system, the Jones matrix of the polarization control system is where and are the phase - shift amounts of the phase - shifters at all levels, E OUT represents the output optical intensity, E FB represents the feedback optical intensity, E x represents the polarization of the incident light in the X - direction, and E y represents the optical intensity of the polarization of the incident light in the Y - direction.

17. The device for improving the polarization tracking speed based on dimensionality reduction according to any one of claims 10 to 16, characterized in that, The structure of each phase shifter is as follows: Among them, is the phase shift amount of the phase shifter; In each sampling cycle, the polarization control system is sampled n + 1 times, where n is the number of phase shifters in the polarization control system.

Images