An optical digital-to-analog conversion system and method based on numerical optimization
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHENZHEN ZHONGKE TIANYING TECH CO LTD
- Filing Date
- 2026-05-08
- Publication Date
- 2026-07-21
AI Technical Summary
Existing optical digital-to-analog conversion technology has defects in the joint optimization design of electrode length and binary code driving sequence, resulting in poor conversion linearity, high hardware complexity, high power consumption, poor scalability, and poor compatibility between electrode fabrication and optical chip technology.
By jointly optimizing the segmented electrode length and binary code driving sequence using numerical optimization algorithms, a constrained optimization problem is constructed to minimize the deviation between the actual electrode activation length and the ideal electrode activation length. Non-equal-length segmented phase modulation electrodes are fabricated, and a suitable digital driving circuit is configured to establish a mapping relationship between binary codewords and electrode activation combinations.
This improved the linearity of optical-to-analog conversion, reduced hardware complexity and power consumption, and enhanced the compatibility of electrode fabrication with optical chip technology and the stability of the conversion.
Smart Images
Figure CN122431046A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of optical communication and photonics integration technology, specifically to an optical digital-to-analog conversion system and method based on numerical optimization. Background Technology
[0002] With the rapid development of big data and cloud computing technologies, the data throughput of large-scale distributed computing systems has surged. Optical interconnect systems have placed higher demands on the linearity, power consumption control, and integration scalability of optical digital-to-analog conversion. However, existing optical digital-to-analog conversion technologies that realize multi-level pulse amplitude modulation have obvious defects in electrode and code sequence design, drive circuit architecture, and process implementation adaptation, which cannot meet the needs of practical applications.
[0003] Existing technologies do not jointly optimize the segmented electrode length and binary code driving sequence. They only set the electrode length or use a fixed code sequence, which cannot construct a constrained optimization problem aimed at minimizing the deviation between the actual and ideal electrode activation lengths. This makes it difficult to compensate for the inherent nonlinearity of the electro-optic modulator, resulting in poor conversion linearity. Traditional methods have shortcomings in driving circuit design. They either rely on external electronic digital-to-analog converters and thermometer encoders to achieve signal driving, increasing hardware complexity and power consumption, or use a linearly increasing number of modulation electrodes, causing the driving port to increase significantly with the modulation order, resulting in extremely poor scalability. Existing methods lack standardized electrode fabrication and codeword mapping implementation processes. The electrode length design is not precisely matched with the basic length unit of optical chip technology. There is no unified standard for the fabrication of non-equal length segmented electrodes, and a stable mapping relationship between binary codewords and electrode activation combinations has not been established and pre-stored. This results in poor compatibility between electrode fabrication and optical chip technology, low matching accuracy between codewords and electrode activations, and affects the stability and operability of the conversion. Summary of the Invention
[0004] The purpose of this invention is to provide an optical digital-to-analog conversion system and method based on numerical optimization to solve the problems raised in the prior art.
[0005] To achieve the above objectives, the present invention provides the following technical solution: In a first aspect, the present invention provides an optical digital-to-analog conversion method based on numerical optimization, comprising: Determine the target pulse amplitude modulation order and match the number of segmented phase modulation electrode segments, then calculate the ideal electrode activation length sequence by combining the inherent optical power transfer function of the electro-optic modulator; Construct the binary driving matrix corresponding to the non-zero input codeword, and initialize the normalized electrode length vector and the initial code order arrangement that satisfy the increasing constraint; With the goal of minimizing the maximum deviation between the actual electrode activation length and the ideal electrode activation length sequence, a joint constraint optimization problem is constructed. A numerical optimization algorithm with boundary constraints is used to iteratively solve the joint constraint optimization problem to obtain the optimal normalized electrode length vector and the optimal code order arrangement. The optimal normalized electrode length vector is converted into the actual physical length, and a non-uniform length segmented phase modulation electrode coupled to the optical waveguide is fabricated; a digital driving circuit adapted to the segmented phase modulation electrode is configured. The mapping relationship between binary input codewords and segmented phase modulation electrode segment activation combinations is established based on the optimal code sequence arrangement; the corresponding electrode segment combinations are driven by digital driving circuits to enable the electro-optic modulator to output linearly distributed optical power amplitude levels, thus completing optical digital-to-analog conversion.
[0006] In conjunction with the first aspect, in the first implementation of the first aspect of this application, the step of determining the target pulse amplitude modulation order and matching the number of segmented phase modulation electrode segments, and calculating the ideal electrode activation length sequence in conjunction with the inherent optical power transfer function of the electro-optic modulator, includes: The target pulse amplitude modulation order is set to a value that is greater than or equal to 4 and is an integer power of 2. The number of matching segmented phase modulation electrode segments is the logarithm of the target pulse amplitude modulation order to the base 2. A Mach-Zehnder interferometer is selected as the electro-optic modulator. The inherent optical power transfer function of the Mach-Zehnder interferometer is in the form of a cosine square. Based on the inherent optical power transfer function of the electro-optic modulator, the correlation formula between the optical power amplitude and the electrode activation length is derived. Substituting the equally spaced optical power amplitude parameters into the correlation formula, the ideal electrode activation length sequence, which exhibits a monotonically decreasing change, is calculated. The length of the ideal electrode activation length sequence is consistent with the number of non-zero input codewords, which is one less than the target pulse amplitude modulation order.
[0007] In conjunction with the first aspect, in the second implementation of the first aspect of this application, the step of constructing the binary driving matrix corresponding to the non-zero input codeword and initializing the normalized electrode length vector and initial code order arrangement that satisfy the increasing constraint includes: Construct a binary driving matrix, where the number of rows in the binary driving matrix is the number of non-zero input codewords, the number of columns is the number of segmented phase modulation electrode segments, and the matrix elements are 0 or 1, representing the inactive and active states of the corresponding segmented phase modulation electrodes, respectively. The normalized electrode length vector is initialized as an increasing numerical sequence, in which each subsequent value is greater than the previous value. Arrange the non-zero input codewords in natural number order to obtain the initial code order arrangement; the normalized electrode length vector and the initial code order arrangement are used together as the initial variables for joint optimization.
[0008] In conjunction with the first aspect, in the third implementation of the first aspect of this application, the step of constructing a joint constraint optimization problem with the objective of minimizing the maximum deviation between the actual electrode activation length and the ideal electrode activation length sequence includes: The initialized normalized electrode length vector and the initial code sequence arrangement are used as joint optimization variables; an objective function is constructed, which is to minimize the maximum absolute deviation between the actual electrode activation length and the normalized ideal electrode activation length sequence. The actual electrode activation length is obtained by performing matrix multiplication on the binary driving matrix after rearranging the initial code sequence arrangement and the normalized electrode length vector. Constraints are imposed, including an increasing constraint on the normalized electrode length vector, a constraint on the total physical length of the segmented phase modulation electrodes, and a constraint on the physical feasibility of the electrode length. Specifically, the increasing constraint on the normalized electrode length vector means that the next value in the numerical sequence is greater than the previous value. Specifically, the constraint on the total physical length of the segmented phase modulation electrodes means that the total length is adapted to the layout space and process implementation capability of the optical chip. Specifically, the constraint on the physical feasibility of the electrode length means that the length of each electrode segment is greater than the minimum length that can be fabricated by the optical chip process. The objective function and constraints are combined to form a joint constraint optimization problem.
[0009] In conjunction with the first aspect, in the fourth implementation of the first aspect of this application, the step of using a numerical optimization algorithm with boundary constraints to iteratively solve the joint constraint optimization problem to obtain the optimal normalized electrode length vector and the optimal code order arrangement includes: The L-BFGS-B quasi-Newton optimization algorithm is selected as the numerical optimization algorithm with boundary constraints. During the iteration process, the normalized electrode length vector and the initial code sequence arrangement are updated synchronously. After each iteration, the objective function value is calculated. When the objective function value reaches the preset iteration convergence threshold or the number of iterations reaches the preset maximum number of iterations, the iteration stops and the optimal normalized electrode length vector and the optimal code sequence arrangement that satisfy all constraints are output.
[0010] In conjunction with the first aspect, in the fifth implementation of the first aspect of this application, the step of converting the optimal normalized electrode length vector into the actual physical length and fabricating a non-uniform length segmented phase modulation electrode coupled to the optical waveguide includes: Based on the basic length unit set by the optical chip process, the optimal normalized electrode length vector is converted into an actual physical length sequence. The conversion method is to multiply each value in the optimal normalized electrode length vector by the basic length unit to obtain the corresponding actual physical length. High electro-optic coefficient materials are selected to fabricate optical waveguides. The cross-section of the optical waveguide is rectangular, and the size is determined according to the characteristics of the optical signal mode and the optical chip process specifications. The operating wavelength covers the commonly used window of optical communication. Non-equal length segmented phase modulation electrodes are sequentially fabricated along the optical path in the phase modulation region of the optical waveguide. The electrode lengths are matched with the actual physical length sequence, and the electrode widths are consistent with the width of the optical waveguide.
[0011] In conjunction with the first aspect, in a sixth implementation of the first aspect of this application, the digital driving circuit configured to adapt to the segmented phase modulation electrodes includes: Configure digital drive circuitry, including a parallel high-speed low-power digital driver, a shift register, and control logic consistent with the number of segmented phase modulation electrode segments, with the digital driver corresponding to the segmented phase modulation electrodes; The digital drive circuit receives a binary digital signal that is the same as the number of segmented phase modulation electrode segments, and the number of drive ports is the same as the number of segmented phase modulation electrode segments.
[0012] In conjunction with the first aspect, in the seventh implementation of the first aspect of this application, the step of establishing a mapping relationship between the binary input codeword and the segmented phase modulation electrode segment activation combination based on the optimal code order includes: Based on the obtained optimal code order, the rows of the constructed binary driving matrix are rearranged to obtain the rearranged binary driving matrix. Establish a mapping relationship between each row of binary codewords in the rearranged binary driving matrix and the corresponding segmented phase modulation electrode segment activation combination, specifically, each row of codewords corresponds to the activation state of a set of electrode segments; The mapping relationship is pre-stored in the control logic of the configured digital drive circuit.
[0013] In conjunction with the first aspect, in the eighth implementation of the first aspect of this application, the step of driving the corresponding electrode segment combination through a digital driving circuit to enable the electro-optic modulator to output a linearly distributed optical power amplitude level, thereby completing the optical-to-analog conversion, includes: After receiving the binary input codeword from the external input, the digital drive circuit retrieves the pre-stored mapping relationship from the control logic and controls the corresponding digital driver to output the drive signal according to the mapping relationship, thereby activating the corresponding non-equal length segmented phase modulation electrode combination, so that the actual electrode activation length generated by the electro-optic modulator matches the calculated ideal electrode activation length sequence. The electro-optic modulator outputs optical power amplitude levels based on the matched electrode activation length. The optical power amplitude levels are linearly distributed, realizing the conversion from digital signals to optical analog signals and completing optical digital-to-analog conversion.
[0014] Secondly, the present invention provides an optical-to-analog conversion system based on numerical optimization, comprising: The parameter modeling module includes a parameter matching unit and an ideal length calculation unit. The parameter matching unit determines the target pulse amplitude modulation order and matches the number of segmented phase modulation electrode segments. The ideal length calculation unit calculates the ideal electrode activation length sequence by combining the inherent optical power transfer function of the electro-optic modulator. Matrix and variable initialization module: includes a driving matrix construction unit and an optimization variable initialization unit; wherein, the driving matrix construction unit constructs the binary driving matrix corresponding to the non-zero input codeword; the optimization variable initialization unit initializes the normalized electrode length vector and the initial code order arrangement that satisfy the increasing constraint; The joint optimization solution module includes an optimization problem construction unit and an algorithm iteration solution unit. The optimization problem construction unit constructs a joint constraint optimization problem with the objective of minimizing the maximum deviation between the actual electrode activation length and the ideal electrode activation length sequence. The algorithm iteration solution unit uses the L-BFGS-B quasi-Newton optimization algorithm to iteratively solve the joint constraint optimization problem and outputs the optimal normalized electrode length vector and the optimal code order arrangement. Hardware fabrication and configuration module: including electrode fabrication unit and circuit configuration unit; wherein, the electrode fabrication unit converts the optimal normalized electrode length vector into the actual physical length and fabricates non-uniform length segmented phase modulation electrodes coupled to the optical waveguide; the circuit configuration unit configures digital driving circuits adapted to the segmented phase modulation electrodes. The mapping and driving output module includes a mapping relationship establishment unit and an electrode driving output unit. The mapping relationship establishment unit establishes a mapping relationship between the binary input codeword and the segmented phase modulation electrode segment activation combination according to the optimal code order and stores it in the control logic. The electrode driving output unit drives the corresponding electrode segment combination through a digital driving circuit to make the electro-optic modulator output a linearly distributed optical power amplitude level, thus completing the optical digital-to-analog conversion.
[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention jointly optimizes the segmented electrode length and the driving sequence of the input binary code, and constructs and solves a joint constraint optimization problem with the objective of minimizing the maximum deviation between the actual and ideal electrode activation lengths.
[0016] 2. This invention employs a logarithmic number of segmented phase modulation electrodes and is configured with a drive circuit containing only a corresponding parallel digital driver, shift register, and control logic, without relying on an external DAC and thermometer encoder.
[0017] 3. This invention converts the physical length of the electrode according to the basic length unit of optical chip technology, fabricates non-equal length segmented electrodes along the phase modulation region of the optical waveguide, establishes and pre-stores the mapping relationship between binary codewords and electrode activation combinations. Attached Figure Description
[0018] Figure 1This is a schematic diagram illustrating the steps of an optical digital-to-analog conversion method based on numerical optimization according to the present invention; Figure 2 This is a flowchart of an optical digital-to-analog conversion method based on numerical optimization according to the present invention. Figure 3 This is a system structure diagram of an optical digital-to-analog conversion system based on numerical optimization according to the present invention. Detailed Implementation
[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0020] Example: Figures 1-3 As shown, the present invention provides a technical solution. like Figure 1 A schematic diagram illustrating the steps of a numerically optimized optical digital-to-analog conversion method is shown. This invention provides a numerically optimized optical digital-to-analog conversion method, comprising: Step S100: Determine the target pulse amplitude modulation order and match the number of segmented phase modulation electrode segments; calculate the ideal electrode activation length sequence by combining the inherent optical power transfer function of the electro-optic modulator. Specifically, the target pulse amplitude modulation order is set to N, where N is a value greater than or equal to 4 and a power of 2, and the number of matching segmented phase modulation electrode segments is m, where m = log2N; A Mach-Zehnder interferometer is selected as the electro-optic modulator. The inherent optical power transfer function of the Mach-Zehnder interferometer is in the form of a cosine square, specifically: ; in, To output optical power, For input optical power, For electrode activation length, The half-wavelength of the electro-optic modulator; Based on the inherent optical power transfer function of the electro-optic modulator described above, the correlation formula between the optical power amplitude and the electrode activation length is derived, specifically to make the output optical power linearly related to the digital code value: ; in, = i is a numeric code value, which takes the values 0, 1, ..., N-1; Substituting the inherent optical power transfer function and simplifying, we obtain the correlation formula: ; Substituting the i-values corresponding to the equally spaced optical power amplitude parameters into the correlation formula, the ideal electrode activation length sequence is obtained by solving using inverse trigonometric functions: ; The ideal electrode activation length sequence exhibits a monotonically decreasing trend, with the length consistent with the number of non-zero input codewords, which is N-1.
[0021] Step S200: Construct the binary driving matrix corresponding to the non-zero input codeword, and initialize the normalized electrode length vector and the initial code order arrangement that satisfy the increasing constraint; Specifically, a binary driving matrix A is constructed, where the number of rows in binary driving matrix A is N-1 (the number of non-zero input codewords), the number of columns is m (the number of segmented phase modulation electrode segments), and the matrix elements A... ij ∈{0,1}, A ij =0 indicates that the j-th segmental phase modulation electrode is not activated, A ij =1 indicates that the j-th segmented phase modulation electrode is activated; the construction method is to convert each non-zero input codeword into an m-bit binary number, with each bit corresponding to a column of the matrix, and fill each row of the matrix in the order of the codewords; The normalized electrode length vector is initialized as an increasing numerical sequence, in which each subsequent value is greater than the previous value. Arrange the non-zero input codewords in natural number order to obtain the initial code order arrangement; the normalized electrode length vector and the initial code order arrangement are used together as the initial variables for joint optimization.
[0022] Step S300: With the objective of minimizing the maximum deviation between the actual electrode activation length and the ideal electrode activation length sequence, a joint constraint optimization problem is constructed; a numerical optimization algorithm with boundary constraints is used to iteratively solve the joint constraint optimization problem to obtain the optimal normalized electrode length vector and the optimal code order arrangement; Specifically, the initialized normalized electrode length vector and the initial code order arrangement are used as joint optimization variables; the objective function is constructed as follows: ; in, This is the binary driving matrix after π rearrangement of the code order. This is the normalized result of the ideal electrode activation length sequence. The norm is infinite, and the objective is to minimize the maximum absolute deviation between the actual electrode activation length and the normalized ideal electrode activation length sequence. The actual electrode activation length is obtained by matrix multiplication of the binary driving matrix after rearranging the initial code order with the normalized electrode length vector, as shown in the formula: ; Constraints are imposed, including an increasing constraint on the normalized electrode length vector, a constraint on the total physical length of the segmented phase modulation electrodes, and a constraint on the physical feasibility of the electrode length. Specifically, the increasing constraint on the normalized electrode length vector means that the next value in the numerical sequence is greater than the previous value. Specifically, the constraint on the total physical length of the segmented phase modulation electrodes means that the total length is adapted to the layout space and process implementation capability of the optical chip. Specifically, the constraint on the physical feasibility of the electrode length means that the length of each electrode segment is greater than the minimum length that can be fabricated by the optical chip process. The objective function and constraints are combined to form a joint constraint optimization problem.
[0023] The L-BFGS-B quasi-Newton optimization algorithm is selected as the numerical optimization algorithm with boundary constraints. During the iteration process, the normalized electrode length vector and the initial code sequence arrangement are updated synchronously. After each iteration, the objective function value is calculated. When the objective function value reaches the preset iteration convergence threshold or the number of iterations reaches the preset maximum number of iterations, the iteration stops and the optimal normalized electrode length vector and the optimal code sequence arrangement that satisfy all constraints are output.
[0024] Step S400: Convert the optimal normalized electrode length vector into the actual physical length, fabricate a non-uniform length segmented phase modulation electrode coupled to the optical waveguide; configure a digital driving circuit adapted to the segmented phase modulation electrode. Specifically, based on the basic length unit set by the optical chip process, the optimal normalized electrode length vector is converted into an actual physical length sequence. The conversion method is to multiply each value in the optimal normalized electrode length vector by the basic length unit to obtain the corresponding actual physical length. Optical waveguides are fabricated using materials with high electro-optic coefficients, such as lithium niobate, indium phosphide, or perovskite. The cross-section of the optical waveguide is rectangular, and the dimensions are determined based on the characteristics of the optical signal mode and the specifications of the optical chip process. The operating wavelength covers the commonly used window of optical communication. Non-equal length segmented phase modulation electrodes are sequentially fabricated along the optical path in the phase modulation region of the optical waveguide. The electrode lengths are matched with the actual physical length sequence, and the electrode widths are consistent with the width of the optical waveguide.
[0025] The configuration includes a digital drive circuit, comprising a parallel high-speed low-power digital driver, a shift register, and control logic, which are consistent with the number of segmented phase modulation electrode segments m. The digital driver corresponds one-to-one with the segmented phase modulation electrodes. The digital drive circuit receives m-bit binary digital signals and does not require an external electronic digital-to-analog converter or thermometer encoder. The number of drive ports is m.
[0026] Step S500: Establish a mapping relationship between binary input codewords and segmented phase modulation electrode segment activation combinations according to the optimal code sequence; drive the corresponding electrode segment combinations through digital driving circuits to make the electro-optic modulator output linearly distributed optical power amplitude levels, thus completing optical digital-to-analog conversion.
[0027] Specifically, based on the obtained optimal code order, the rows of the constructed binary driving matrix A are rearranged to obtain the rearranged binary driving matrix; Establish a one-to-one mapping relationship between each row of binary codewords in the rearranged binary driving matrix and the corresponding segmented phase modulation electrode segment activation combination. Specifically, each bit of each row of codewords corresponds to one segmented phase modulation electrode. When the bit value is 0, the corresponding electrode is not activated, and when the bit value is 1, the corresponding electrode is activated. That is, each row of codewords uniquely corresponds to the activation state of a set of electrode segments. The mapping relationship is pre-stored in the control logic of the configured digital drive circuit.
[0028] After receiving the binary input codeword from the external input, the digital drive circuit retrieves the pre-stored mapping relationship from the control logic and controls the corresponding digital driver to output the drive voltage signal according to the mapping relationship, thereby activating the corresponding non-equal length segmented phase modulation electrode combination, so that the deviation between the actual electrode activation length generated by the electro-optic modulator and the calculated ideal electrode activation length sequence is less than the preset deviation threshold. The electro-optic modulator outputs optical power amplitude levels based on matched electrode activation lengths. The linear distribution characteristics are verified by detecting the output optical power. Specifically, an optical power meter is used to measure the output optical power corresponding to each code value to ensure that the optical power amplitude levels satisfy a linear relationship. ; It realizes the conversion from digital signals to optical analog signals, and completes optical digital-to-analog conversion.
[0029] like Figure 2 The flowchart of an optical digital-to-analog conversion method based on numerical optimization is shown. This invention provides an optical digital-to-analog conversion method based on numerical optimization, comprising: After the process begins, the target pulse amplitude modulation order is first determined and set to a value greater than or equal to 4 and a power of 2. The number of segmented phase modulation electrode segments is matched to the base-2 logarithm of the modulation order. A Mach-Zehnder interferometer is selected as the electro-optic modulator. Based on its inherent cosine square form optical power transfer function, the correlation formula between optical power amplitude and electrode activation length is derived. The ideal electrode activation length sequence is calculated by substituting the equally spaced optical power amplitude parameters. Then, the valid sequence is checked in a branch to verify whether the sequence length is equal to the number of non-zero input codewords. Specifically, the modulation order is reduced by one, and the sequence exhibits a monotonically decreasing characteristic. If the conditions are not met, the parameter calculation is re-executed. If the conditions are met, the process proceeds to the next step.
[0030] After sequence verification, a binary driving matrix corresponding to the non-zero input codewords is constructed. A normalized electrode length vector satisfying the increasing constraint is initialized, and the non-zero input codewords are arranged in natural order to obtain an initial code sequence arrangement. Both are used as initial variables for joint optimization. Using the normalized electrode length vector and the initial code sequence arrangement as joint optimization variables, a function is constructed with the objective of minimizing the maximum absolute deviation between the actual electrode activation length and the normalized ideal electrode activation length sequence. Three types of constraints are applied to form a joint constraint optimization problem. The L-BFGS-B quasi-Newton optimization algorithm is used to solve the problem iteratively. The convergence check branch is entered to determine whether the objective function value after each iteration reaches the preset convergence threshold or whether the number of iterations reaches the maximum number. If convergence is not achieved, the optimization is repeated. If convergence is achieved, the optimal normalized electrode length vector and the optimal code sequence arrangement are output.
[0031] After obtaining the optimal parameters, two operations are performed in parallel. The first operation is to convert the optimal normalized electrode length vector into the actual physical length according to the basic length unit of the optical chip process, and to fabricate non-equal length segmented phase modulation electrodes coupled to the optical waveguide. The second operation is to configure a digital driving circuit that matches the number of electrode segments. After the hardware configuration is completed, the rows of the binary driving matrix are rearranged according to the optimal code order, and a mapping relationship is established between each row of binary codewords in the rearranged matrix and the corresponding electrode segment activation combination. This mapping relationship is then pre-stored in the control logic of the digital driving circuit.
[0032] After the mapping relationship is pre-stored, the digital drive circuit receives external binary input codewords, retrieves the pre-stored mapping relationship, controls the corresponding driver to output drive signals, activates non-equal length electrode combinations, makes the actual electrode activation length match the ideal sequence, and finally outputs a linearly distributed optical power amplitude level; enters the accuracy verification branch to verify whether the optical power amplitude level is linearly distributed: if it does not meet the standard, it backtracks to the optimization solution stage to iterate again; if it meets the standard, it completes the digital-to-analog conversion and the process ends.
[0033] like Figure 3 The system structure diagram of an optical digital-to-analog converter based on numerical optimization is shown in the present invention. The present invention provides an optical digital-to-analog converter based on numerical optimization, comprising: The parameter modeling module includes a parameter matching unit and an ideal length calculation unit. The parameter matching unit determines the target pulse amplitude modulation order and matches the number of segmented phase modulation electrode segments. The ideal length calculation unit calculates the ideal electrode activation length sequence by combining the inherent optical power transfer function of the electro-optic modulator. Matrix and variable initialization module: includes a driving matrix construction unit and an optimization variable initialization unit; wherein, the driving matrix construction unit constructs the binary driving matrix corresponding to the non-zero input codeword; the optimization variable initialization unit initializes the normalized electrode length vector and the initial code order arrangement that satisfy the increasing constraint; The joint optimization solution module includes an optimization problem construction unit and an algorithm iteration solution unit. The optimization problem construction unit constructs a joint constraint optimization problem with the objective of minimizing the maximum deviation between the actual electrode activation length and the ideal electrode activation length sequence. The algorithm iteration solution unit uses the L-BFGS-B quasi-Newton optimization algorithm to iteratively solve the joint constraint optimization problem and outputs the optimal normalized electrode length vector and the optimal code order arrangement. Hardware fabrication and configuration module: including electrode fabrication unit and circuit configuration unit; wherein, the electrode fabrication unit converts the optimal normalized electrode length vector into the actual physical length and fabricates non-uniform length segmented phase modulation electrodes coupled to the optical waveguide; the circuit configuration unit configures digital driving circuits adapted to the segmented phase modulation electrodes. The mapping and driving output module includes a mapping relationship establishment unit and an electrode driving output unit. The mapping relationship establishment unit establishes a mapping relationship between the binary input codeword and the segmented phase modulation electrode segment activation combination according to the optimal code order and stores it in the control logic. The electrode driving output unit drives the corresponding electrode segment combination through a digital driving circuit to make the electro-optic modulator output a linearly distributed optical power amplitude level, thus completing the optical digital-to-analog conversion.
[0034] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.
Claims
1. A numerical optimization-based optical digital-to-analog conversion method, characterized in that, include: Determine the target pulse amplitude modulation order and match the number of segmented phase modulation electrode segments, then calculate the ideal electrode activation length sequence by combining the inherent optical power transfer function of the electro-optic modulator; Construct the binary driving matrix corresponding to the non-zero input codeword, and initialize the normalized electrode length vector and the initial code order arrangement that satisfy the increasing constraint; With the goal of minimizing the maximum deviation between the actual electrode activation length and the ideal electrode activation length sequence, a joint constraint optimization problem is constructed. A numerical optimization algorithm with boundary constraints is used to iteratively solve the joint constraint optimization problem to obtain the optimal normalized electrode length vector and the optimal code order arrangement. The optimal normalized electrode length vector is converted into the actual physical length, and a non-uniform length segmented phase modulation electrode coupled to the optical waveguide is fabricated; a digital driving circuit adapted to the segmented phase modulation electrode is configured. Establish a mapping relationship between binary input codewords and segmented phase modulation electrode segment activation combinations based on the optimal code order arrangement; By driving the corresponding electrode segment combination through a digital driving circuit, the electro-optic modulator outputs a linearly distributed optical power amplitude level, thus completing the optical digital-to-analog conversion.
2. The optical digital-to-analog conversion method based on numerical optimization according to claim 1, characterized in that, The process of determining the target pulse amplitude modulation order and matching the number of segmented phase modulation electrode segments, and calculating the ideal electrode activation length sequence in conjunction with the inherent optical power transfer function of the electro-optic modulator, includes: The target pulse amplitude modulation order is set to a value that is greater than or equal to 4 and is an integer power of 2. The number of matching segmented phase modulation electrode segments is the logarithm of the target pulse amplitude modulation order to the base 2. A Mach-Zehnder interferometer is selected as the electro-optic modulator. The inherent optical power transfer function of the Mach-Zehnder interferometer is in the form of a cosine square. Based on the inherent optical power transfer function of the electro-optic modulator, the correlation formula between the optical power amplitude and the electrode activation length is derived. Substituting the equally spaced optical power amplitude parameters into the correlation formula, the ideal electrode activation length sequence, which exhibits a monotonically decreasing change, is calculated. The length of the ideal electrode activation length sequence is consistent with the number of non-zero input codewords, which is one less than the target pulse amplitude modulation order.
3. The optical digital-to-analog conversion method based on numerical optimization according to claim 1, characterized in that, The construction of the binary driving matrix corresponding to the non-zero input codeword, and the initialization of the normalized electrode length vector and the initial code order arrangement satisfying the increasing constraint, include: Construct a binary driving matrix, where the number of rows in the binary driving matrix is the number of non-zero input codewords, the number of columns is the number of segmented phase modulation electrode segments, and the matrix elements are 0 or 1, representing the inactive and active states of the corresponding segmented phase modulation electrodes, respectively. The normalized electrode length vector is initialized as an increasing numerical sequence, in which each subsequent value is greater than the previous value. Arrange the non-zero input codewords in natural number order to obtain the initial code order arrangement; the normalized electrode length vector and the initial code order arrangement are used together as the initial variables for joint optimization.
4. The optical digital-to-analog conversion method based on numerical optimization according to claim 1, characterized in that, The joint constraint optimization problem, which aims to minimize the maximum deviation between the actual electrode activation length and the ideal electrode activation length sequence, includes: The initialized normalized electrode length vector and the initial code sequence arrangement are used as joint optimization variables; an objective function is constructed, which is to minimize the maximum absolute deviation between the actual electrode activation length and the normalized ideal electrode activation length sequence. The actual electrode activation length is obtained by performing matrix multiplication on the binary driving matrix after rearranging the initial code sequence arrangement and the normalized electrode length vector. Constraints are imposed, including an increasing constraint on the normalized electrode length vector, a constraint on the total physical length of the segmented phase modulation electrodes, and a constraint on the physical feasibility of the electrode length. Specifically, the increasing constraint on the normalized electrode length vector means that the next value in the numerical sequence is greater than the previous value. Specifically, the constraint on the total physical length of the segmented phase modulation electrodes means that the total length is adapted to the layout space and process implementation capability of the optical chip. Specifically, the constraint on the physical feasibility of the electrode length means that the length of each electrode segment is greater than the minimum length that can be fabricated by the optical chip process. The objective function and constraints are combined to form a joint constraint optimization problem.
5. The optical digital-to-analog conversion method based on numerical optimization according to claim 1, characterized in that, The numerical optimization algorithm with boundary constraints iteratively solves the joint constraint optimization problem to obtain the optimal normalized electrode length vector and the optimal code order arrangement, including: The L-BFGS-B quasi-Newton optimization algorithm is selected as the numerical optimization algorithm with boundary constraints. During the iteration process, the normalized electrode length vector and the initial code sequence arrangement are updated synchronously. After each iteration, the objective function value is calculated. When the objective function value reaches the preset iteration convergence threshold or the number of iterations reaches the preset maximum number of iterations, the iteration stops and the optimal normalized electrode length vector and the optimal code sequence arrangement that satisfy all constraints are output.
6. The optical digital-to-analog conversion method based on numerical optimization according to claim 1, characterized in that, The process of converting the optimal normalized electrode length vector into the actual physical length and fabricating a non-uniform length segmented phase modulation electrode coupled to the optical waveguide includes: Based on the basic length unit set by the optical chip process, the optimal normalized electrode length vector is converted into an actual physical length sequence. The conversion method is to multiply each value in the optimal normalized electrode length vector by the basic length unit to obtain the corresponding actual physical length. High electro-optic coefficient materials are selected to fabricate optical waveguides. The cross-section of the optical waveguide is rectangular, and the size is determined according to the characteristics of the optical signal mode and the optical chip process specifications. The operating wavelength covers the commonly used window of optical communication. Non-equal length segmented phase modulation electrodes are sequentially fabricated along the optical path in the phase modulation region of the optical waveguide. The electrode lengths are matched with the actual physical length sequence, and the electrode widths are consistent with the width of the optical waveguide.
7. The optical digital-to-analog conversion method based on numerical optimization according to claim 1, characterized in that, The digital driving circuit configured to adapt to the segmented phase modulation electrodes includes: Configure digital drive circuitry, including a parallel high-speed low-power digital driver, a shift register, and control logic consistent with the number of segmented phase modulation electrode segments, with the digital driver corresponding to the segmented phase modulation electrodes; The digital drive circuit receives a binary digital signal that is the same as the number of segmented phase modulation electrode segments, and the number of drive ports is the same as the number of segmented phase modulation electrode segments.
8. The optical digital-to-analog conversion method based on numerical optimization according to claim 1, characterized in that, The process of establishing a mapping relationship between binary input codewords and segmented phase modulation electrode segment activation combinations based on the optimal code order includes: Based on the obtained optimal code order, the rows of the constructed binary driving matrix are rearranged to obtain the rearranged binary driving matrix. Establish a mapping relationship between each row of binary codewords in the rearranged binary driving matrix and the corresponding segmented phase modulation electrode segment activation combination, specifically, each row of codewords corresponds to the activation state of a set of electrode segments; The mapping relationship is pre-stored in the control logic of the configured digital drive circuit.
9. The optical digital-to-analog conversion method based on numerical optimization according to claim 1, characterized in that, The process of driving the corresponding electrode segment combination through a digital driving circuit to enable the electro-optic modulator to output linearly distributed optical power amplitude levels, thereby completing optical-to-analog conversion, includes: After receiving the binary input codeword from the external input, the digital drive circuit retrieves the pre-stored mapping relationship from the control logic and controls the corresponding digital driver to output the drive signal according to the mapping relationship, thereby activating the corresponding non-equal length segmented phase modulation electrode combination, so that the actual electrode activation length generated by the electro-optic modulator matches the calculated ideal electrode activation length sequence. The electro-optic modulator outputs optical power amplitude levels based on the matched electrode activation length. The optical power amplitude levels are linearly distributed, realizing the conversion from digital signals to optical analog signals and completing optical digital-to-analog conversion.
10. A numerically optimized optical digital-to-analog conversion system, using the numerically optimized optical digital-to-analog conversion method according to any one of claims 1-9, characterized in that, include: The parameter modeling module includes a parameter matching unit and an ideal length calculation unit. The parameter matching unit determines the target pulse amplitude modulation order and matches the number of segmented phase modulation electrode segments. The ideal length calculation unit calculates the ideal electrode activation length sequence by combining the inherent optical power transfer function of the electro-optic modulator. Matrix and variable initialization module: includes a driving matrix construction unit and an optimization variable initialization unit; wherein, the driving matrix construction unit constructs the binary driving matrix corresponding to the non-zero input codeword; the optimization variable initialization unit initializes the normalized electrode length vector and the initial code order arrangement that satisfy the increasing constraint; The joint optimization solution module includes an optimization problem construction unit and an algorithm iteration solution unit. The optimization problem construction unit constructs a joint constraint optimization problem with the objective of minimizing the maximum deviation between the actual electrode activation length and the ideal electrode activation length sequence. The algorithm iteration solution unit uses the L-BFGS-B quasi-Newton optimization algorithm to iteratively solve the joint constraint optimization problem and outputs the optimal normalized electrode length vector and the optimal code order arrangement. Hardware fabrication and configuration module: including electrode fabrication unit and circuit configuration unit; wherein, the electrode fabrication unit converts the optimal normalized electrode length vector into the actual physical length and fabricates non-uniform length segmented phase modulation electrodes coupled to the optical waveguide; the circuit configuration unit configures digital driving circuits adapted to the segmented phase modulation electrodes. The mapping and driving output module includes a mapping relationship establishment unit and an electrode driving output unit. The mapping relationship establishment unit establishes a mapping relationship between the binary input codeword and the segmented phase modulation electrode segment activation combination according to the optimal code order and stores it in the control logic. The electrode driving output unit drives the corresponding electrode segment combination through a digital driving circuit to make the electro-optic modulator output a linearly distributed optical power amplitude level, thus completing the optical digital-to-analog conversion.