A method for constructing a parameterized optoelectronic co-simulation platform for optoelectronic chip design

Abstract

The application relates to the field of photonic device modeling, and discloses a parameterized optoelectronic joint simulation platform construction method for optoelectronic chip design, which is characterized in that, in an electronic design automation platform, a unified model framework of a photonic device is constructed based on a hardware description language, and conversion between frequency domain response and time domain response is realized in the model, including the following steps: a general VerilogA model framework is constructed based on the frequency domain response of the photonic device; a frequency-time domain numerical conversion mechanism is realized in the model; and for the photonic device with dispersion characteristics, a parameterized time domain model is constructed based on a finite impulse response. The application can directly embed the photonic device model into an electronic circuit simulation environment, co-simulate the electronic device, support time-frequency domain analysis of the cascade structure and the loop structure, thereby avoiding the complicated steps of separate simulation by relying on external simulation tools in the process of optoelectronic chip design, and further improving the design efficiency and iteration speed of the chip.

Description

Technical Field

[0001] This invention relates to the field of photonic device modeling, and in particular to a method for constructing a parametric optoelectronic co-simulation platform for optoelectronic chip design. Background Technology

[0002] With the development of high-speed computing, data center interconnects, and heterogeneous chip systems, photonic devices are increasingly being introduced into optoelectronic fusion design to achieve high-speed, low-power data transmission. In the design process of optoelectronic fusion chips, it is typically necessary to co-model and simulate photonic devices and electronic circuits to evaluate signal integrity, timing characteristics, and link performance at the system level. Therefore, introducing photonic device models into electronic design automation (EDA) platforms and using hardware description languages ​​to model photonic devices, thereby achieving co-design and co-simulation analysis of optoelectronic systems at the schematic layer level, has become an important development direction.

[0003] Among existing photonic device modeling techniques, a common approach is to construct a compact model of the photonic device using analytical formulas or equivalent circuits within a hardware description language. This method is typically used for photonic devices that are wavelength-independent within their operating bandwidth, describing only the amplitude attenuation or fixed time delay. This modeling approach is applicable to devices that are wavelength-independent or whose dispersion effects are negligible. However, when the device exhibits significant group delay differences or higher-order dispersion effects, the frequency domain response no longer has a simple correspondence with the time domain response. The aforementioned methods struggle to accurately reflect the dynamic behavior of the signal, such as broadening and distortion, in the time domain, leading to discrepancies between eye diagram simulation results and actual device characteristics.

[0004] To overcome the aforementioned problems, another existing technology employs time-domain modeling methods based on scattering parameters. For example, algorithms such as rational function fitting or complex vector fitting are used to convert frequency-domain scattering parameters into a time-domain equivalent model. Theoretically, this approach can guarantee the causality and passivity of the model and describe the dispersive characteristics of photonic devices well. However, this method typically requires complex numerical optimization and matrix operations, resulting in a lengthy computational process. Furthermore, when the device has a long propagation delay or its spectrum contains multiple resonance characteristics, the resulting model has a high order and large scale, which is unsuitable for use in large-scale circuit or system-level simulations, thus limiting its application efficiency in practical engineering design.

[0005] Furthermore, in existing EDA simulation workflows, time-domain simulations based on scattering parameters (S-parameters) often rely on externally generated fixed parameter files. Once the parameter file is generated, its frequency domain characteristics are difficult to adjust flexibly during simulation based on device structure or physical parameters. This leads to frequent calls to external simulation tools during design iterations or parameter optimization, increasing the complexity and time cost of the design process. Simultaneously, the lack of uniformity in data formats and interfaces between different simulation platforms further reduces the overall efficiency of optoelectronic co-simulation.

[0006] Therefore, the following technical bottlenecks exist in existing optoelectronic chip designs:

[0007] First, existing electronic design automation platforms lack a unified and standardized modeling architecture for photonic devices. Especially in hardware description language environments such as Verilog-A, different photonic device models are usually built independently by their respective modelers. The modeling methods, port definitions, model functions, and applicable scenarios are inconsistent, making it difficult to unify the frequency domain model and the time domain model. Consequently, it is difficult to form a reusable and scalable unified time-frequency domain model system in the optoelectronic chip design process, which restricts the standardization and efficient development of optoelectronic chip design.

[0008] Secondly, while existing technologies can achieve schematic-level co-simulation of photonic devices and electronic circuits within electronic design automation (EDA) platforms, current photonic device models primarily employ simplified time-domain descriptions, failing to fully consider wavelength-dependent optical characteristics such as dispersion and group delay. The fundamental reason lies in the limitations of hardware description languages ​​in numerical expression and domain transformation, making it difficult to directly describe frequency-dependent propagation characteristics. External numerical calculation tools are typically required for model calculations, resulting in a lack of accurate time-domain models for wavelength-dependent photonic devices such as dispersive waveguides and linear optical fibers. Consequently, it is difficult to complete system-level time-domain design and performance analysis relying on these devices within EDA platforms, hindering the design and verification of complex optoelectronic fusion systems.

[0009] Third, in existing technologies, schematic-level packaged photonic device models are typically built based on the results of physical-level full-wave simulations. The model parameters are fixed after generation, similar to Touchstone parameter files in electronic design. However, in actual optoelectronic chip design, when designers need to adjust device parameters for different application scenarios or system performance indicators, existing models cannot be directly optimized within electronic design automation platforms. Often, a new device-level full-wave simulation is required to generate new model data. This process is computationally expensive and time-consuming, resulting in low device model update efficiency and severely restricting the design iteration efficiency and engineering application flexibility of optoelectronic chips. Summary of the Invention

[0010] To address the aforementioned technical problems, this invention provides a method for constructing a parametric optoelectronic co-simulation platform for optoelectronic chip design. This method enables precise time-frequency domain modeling of photonic devices within an electronic design automation platform, accurately characterizing their optical properties. It is also compatible with electrical simulation solvers, ultimately achieving unified modeling and time-frequency domain co-simulation with electronic circuits within the same schematic environment. This improves the overall efficiency and consistency of optoelectronic chip design.

[0011] To achieve the above objectives, the technical solution of the present invention is as follows:

[0012] A method for constructing a parametric optoelectronic co-simulation platform for optoelectronic chip design, comprising the following steps: In an electronic design automation platform, a unified model framework for photonic devices is constructed based on a hardware description language, and the conversion between frequency domain response and time domain response is implemented within the model.

[0013] Step 1: Construct a general Verilog A model framework based on the frequency domain response of photonic devices, where the photonic devices are modeled as multi-port linear systems, each port corresponds to the input or output of an optical signal, and the transmission characteristics of the optical signal in the device are described by a scattering matrix;

[0014] Step 2: Implement a frequency-domain and time-domain numerical conversion mechanism within the Verilog A model, including writing a discrete Fourier transform algorithm and a discrete convolution operation algorithm within the model to convert the frequency domain response of the photonic device into a time-domain impulse response;

[0015] Step 3: For photonic devices with dispersive characteristics, construct a parameterized time-domain model based on finite impulse response. The key optical parameters of the model are set as user-adjustable input variables, enabling dynamic adjustment of device parameters and time-domain simulation within the electronic design automation platform.

[0016] In the above scheme, in step one, the general Verilog A model framework represents the optical signal by separating the real part and the imaginary part. Each optical signal is split into a real part channel and an imaginary part channel, and the correspondence is realized through port layout. At the same time, it supports the distinction between forward propagation signals and backward propagation signals.

[0017] In the above scheme, in step one, the general Verilog A model framework supports the extension of multiple modes or multiple polarization states, and the evolution relationship mapping between different modes is realized through the extension port.

[0018] In the above scheme, in step two, the Fourier transform adopts a discrete frequency-time domain conversion method within the framework of the general Verilog A model, representing the frequency domain response as a discrete sampling point sequence, converting the rotation factor in the Fourier transform into a trigonometric function expression; separating the real and imaginary complex numbers, and obtaining the time-domain impulse response through iterative calculation.

[0019] In the above scheme, in step two, the convolution operation is implemented by introducing a time advance function with a fixed time step to achieve discrete convolution calculation, which is used to describe the time delay and memory effect of photonic devices.

[0020] In the above scheme, step three, constructing the parameterized time-domain model, includes the following sub-steps:

[0021] (1) Discretization of spectral response: Based on the physical parameters or analytical model of the dispersive device, the amplitude response and phase response at each discrete frequency sampling point are calculated within a preset spectral range to form a discrete frequency domain response;

[0022] (2) Determination of time domain response length: Based on the group delay characteristics in the discrete frequency domain response and the system simulation time step, calculate the number of taps required for the time domain finite impulse response model;

[0023] (3) Frequency domain response resampling: Based on the number of taps determined in step (2), the discrete frequency domain response obtained in step (1) is resampled so that the number of frequency domain sampling points matches the length of the time domain model.

[0024] (4) Subsampling delay compensation: When the group delay is not an integer multiple of the simulation time step, the frequency domain response after resampling is compensated by a fractional delay filter to correct the phase deviation at the subsampling level;

[0025] (5) Causality violation suppression: Apply a window function to the compensated frequency domain response to suppress non-causal components that may be caused by frequency domain truncation and band-limited sampling;

[0026] (6) Time-domain model generation: The frequency domain response processed in step (5) is converted into a time-domain finite impulse response through inverse Fourier transform to obtain the parameterized time-domain model.

[0027] In the above scheme, step three involves key optical parameters including at least one of device length, effective refractive index, and dispersion coefficient.

[0028] In a further technical solution, the specific method for discretizing the spectral response in step (1) is as follows:

[0029] First, the frequency range required for frequency domain modeling is determined based on the operating bandwidth of the dispersive photonic device, the system sampling rate, and the simulation requirements.

[0030] Subsequently, based on the Nyquist sampling theorem, the number of discrete frequency sampling points and the corresponding frequency resolution are determined within this frequency range, thereby constructing a discrete frequency range for modeling.

[0031] Secondly, at the determined frequency sampling points, the amplitude response and phase response of the dispersive photonic device at each frequency point are calculated using the physical parameters or analytical model of the device, forming a discrete frequency domain response.

[0032] In a further technical solution, the formula for the number of taps required to calculate the finite-impulse response model in step (2) is as follows:

[0033] ;

[0034] in, This indicates the number of taps required for the time-domain model. Indicates group delay. This indicates the simulation time step.

[0035] Through the above technical solution, the parametric optoelectronic co-simulation platform construction method for optoelectronic chip design provided by the present invention has the following beneficial effects:

[0036] (1) This invention achieves unified construction and consistent representation of frequency domain and time domain models of photonic devices. In the prior art, frequency domain analysis and time domain simulation often rely on different modeling methods or different simulation tools, and there is a lack of a unified model representation between the frequency domain response and the time domain response, which can easily lead to inconsistent system-level analysis results. This invention uses frequency domain scattering parameters as a unified description basis, and by constructing a corresponding time domain model within the model, it makes the frequency domain response and the time domain response consistent under the same modeling framework;

[0037] (2) This invention can accurately characterize the group delay and dispersion effect of dispersive devices, significantly improving the accuracy of time-domain modeling. In view of the problems that traditional Verilog-A photonic device models generally use fixed delay or instantaneous dot product approximation, which are difficult to reflect the pulse broadening and waveform distortion caused by dispersion, this invention uses a finite impulse response filter modeling method to simultaneously reflect the overall propagation delay and frequency-dependent delay characteristics of the device in the time-domain model;

[0038] (3) This invention ensures both modeling accuracy and numerical stability and engineering feasibility. By determining the number of taps based on group delay, resampling the frequency domain response, and introducing a fractional delay filter, this invention effectively avoids the phase deviation problem caused by discretization error. At the same time, by introducing window function processing when constructing the time domain model, non-causal components that may be caused by frequency domain truncation and band-limited sampling are suppressed, thereby improving the physical rationality of the time domain impulse response and the stability of transient simulation.

[0039] Based on the above technical solution, this invention constructs a standardized Verilog-A photonic device model framework and port layout, realizing numerical conversion and propagation calculations between the frequency and time domains within the model. Furthermore, it employs a parametric modeling approach based on finite impulse response, enabling accurate characterization of the group delay and dispersion characteristics of dispersive photonic devices within an electronic design automation platform. This achieves joint frequency domain analysis and time-domain transient simulation of photonic devices and electronic circuits at the same schematic layer level, avoiding the problems of non-reusable models, difficult parameter adjustment, and reliance on external full-wave simulations in existing technologies. This improves the feasibility and design iteration efficiency of optoelectronic chip system-level design and simulation. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0041] Figure 1 This is a schematic diagram of a method for constructing a parametric optoelectronic co-simulation platform for optoelectronic chip design, as disclosed in an embodiment of the present invention.

[0042] Figure 2 This is a schematic diagram of the Verilog-A model framework;

[0043] Figure 3 This is a schematic diagram of the port layout for a Verilog-A model.

[0044] Figure 4 This is a schematic diagram of an optoelectronic integrated circuit.

[0045] Figure 5 The simulation comparison results of the optoelectronic integrated link are as follows: (a) frequency domain comparison results, (b) time domain results of commercial software, and (c) time domain results of Verilog-A model.

[0046] Figure 6 The following are test diagrams for linear optical fibers: (a) is a schematic diagram of the Verilog-A test principle, and (b) is a simulation comparison result of pulse broadening characteristics.

[0047] Figure 7 (a) shows the on-chip optical waveguide test diagram, (b) shows the Verilog-A test schematic diagram, and (c) shows the simulation comparison results of the time delay characteristics.

[0048] Figure 8 To verify the causal violation results, (a) shows the comparison results of the Verilog-A window function, and (b) shows the effect of causal violation suppression. Detailed Implementation

[0049] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0050] This invention provides a method for constructing a parametric optoelectronic co-simulation platform for optoelectronic chip design. This method enables precise time-frequency domain modeling of photonic devices within an electronic design automation (EDA) platform, accurately characterizing their optical properties, and is compatible with electrical simulation solvers. Ultimately, it achieves unified modeling and time-frequency domain co-simulation with electronic circuits within the same schematic environment, thereby improving the overall efficiency and consistency of optoelectronic chip design. Figure 1 As shown.

[0051] This invention constructs a unified model framework for photonic devices based on a hardware description language within an electronic design automation platform, and implements the conversion between frequency domain response and time domain response within the model. The method of this invention includes the following steps:

[0052] Step 1: Construct a general Verilog A model framework based on the frequency domain response of photonic devices, in which photonic devices are modeled as multi-port linear systems, each port corresponding to the input or output of optical signals, and the transmission characteristics of optical signals in the device are described by the scattering matrix.

[0053] Specifically, the frequency domain response of photonic devices is determined based on their analytical representations, and the transmission characteristics of optical signals within the devices are described in the form of a scattering matrix. The scattering matrix simultaneously considers the bidirectional transmission of optical signals and their polarization state or multi-mode characteristics, used to characterize the evolution relationships between different modes and the transmission paths of signals between ports. This standardized model architecture is applicable to most photonic devices. By unifying the port format of the Verilog-A model and its correspondence with the scattering matrix, different types of photonic devices can be embedded into electronic design automation platforms in a consistent manner, thus providing a standardized model foundation for the subsequent joint design and simulation of photonic devices and electronic components at the schematic level.

[0054] This model framework is based on the scattering matrix of photonic devices. By constructing a unified matrix structure and port mapping method within the Verilog-A model, it enables callable, cascadeable, and scalable modeling of photonic devices in electronic circuit simulation environments. Figure 2 As shown.

[0055] In this model framework, photonic devices are abstracted as multi-port linear systems, with each port corresponding to the input or output relationship of an optical signal. Considering that optical signals inherently contain both amplitude and phase information, and that EDA platforms typically do not directly support complex signal operations, this invention represents optical signals by separating the real and imaginary parts in the model. Specifically, each optical signal is split into a real part channel (Re) and an imaginary part channel (Im), and the correspondence is achieved through port layout, such as... Figure 3 As shown.

[0056] To further support the bidirectional propagation characteristics in photonic devices, this invention distinguishes between forward and backward propagation signals at the port level, enabling the model to describe the reflection, propagation, and bidirectional scattering relationships between ports within the device. Simultaneously, this port layout offers good scalability; when describing polarization state evolution or multimode transmission, one side of the port can be expanded into multiple sets of real / imaginary channels, thereby achieving unified modeling of different polarization modes or spatial modes, such as... Figure 3 The diagram shows modes from 1 to n. When the photonic device only has polarization states, modes 1 and 2 can be considered as TE and TM modes, respectively. The mapping relationship between their mode evolution and the scattering matrix is ​​as follows:

[0057] ;

[0058] in, Indicates a port. Indicates a two-port port. and Even numbers represent the real part, and odd numbers represent the imaginary part. The S-parameters represent the evolution from TE mode to TE mode, where i and j represent ports, and TE and TM represent polarization states, and the same applies to other modes.

[0059] Through the aforementioned model framework and port layout design, this invention enables photonic devices to be embedded into electronic design automation platforms in the form of standard Verilog-A models. This allows the optical and transmission characteristics of photonic devices to be described using electronic design standards, providing a unified foundation for subsequent frequency domain analysis and time domain transient simulation.

[0060] Step 2: Implement the frequency-domain and time-domain numerical conversion mechanism within the Verilog A model, including writing discrete Fourier transform and discrete convolution algorithms within the model to convert the frequency domain response of the photonic device into the time-domain impulse response.

[0061] By constructing Fourier transform and convolution algorithms within the general model, the challenges of frequency-domain to time-domain transfer and loop computation in the hardware environment are solved. This enables photonic device models to undergo time-frequency domain transformation within the Verilog-A general model and supports loop structure analysis of photonic links. Based on this numerical computation mechanism, the necessary computational foundation is provided for the subsequent construction of parameterized time-domain models of photonic devices that can accurately characterize group delay and dispersion properties.

[0062] First, regarding the Fourier transform operation, this invention employs a discrete-form frequency-to-time domain transformation method within the Verilog-A model. Considering the limited support for complex number operations in the Verilog-A environment, this invention represents frequency and time domain quantities as two unified parts: real and imaginary. Second, the frequency domain response of the photonic device is represented as a numerical sequence at discrete sampling points (…). ), the rotation factor in the Fourier transform Convert to trigonometric function expression. Separate the real and imaginary aspects of the complex number, and iterate through all... The time-domain sequence is obtained through iterative calculation. This establishes the mapping relationship between frequency domain components and time domain sampling points. In this way, the model can internally generate the corresponding time domain impulse response based on the discrete frequency domain response, thereby implementing the Fourier forward and inverse transforms of the following formulas in the EDA platform;

[0063] ;

[0064] ;

[0065] in, Let N be the numerical sequence of discrete sampling points, where N is the number of sampling points, and n and k are indices. It is a time-domain sequence. This represents the rotation factor.

[0066] Secondly, regarding convolution operations, this invention implements a discrete convolution calculation mechanism suitable for transient simulation within the Verilog-A model, as shown in the following formula, which describes the time-domain propagation characteristics of photonic devices with time delay and memory effects.

[0067] ;

[0068] in, For output signal, For input signal, For unit impulse response, For time.

[0069] In this mechanism, the model introduces a time-stepping function `@timer(dt)` with a fixed time step to force a uniform step size, thus discretizing the continuous convolution. Using a `for` loop, at each time step, the loop executes once, weighting and accumulating the historical input samples, thereby achieving discrete convolution. The entire simulation process is repeated to obtain the time-domain output of the complete link.

[0070] By implementing Fourier transform and convolution operations within the Verilog-A model, this invention endows photonic device models with fundamental numerical computation capabilities, enabling them to achieve a unified description in the frequency and time domains within an EDA platform.

[0071] Step 3: For photonic devices with dispersive characteristics, construct a parameterized time-domain model based on finite impulse response. The key optical parameters of the model are set as user-adjustable input variables, enabling dynamic adjustment of device parameters and time-domain simulation within the electronic design automation platform.

[0072] By constructing a finite-impulse response (FIR) model of a photonic device and setting key optical parameters of the device as user-adjustable input variables, the model can be flexibly adjusted within an electronic design automation platform according to design requirements. The adjustable parameters of the model include parameters related to the optical and transmission characteristics of the device, such as device length, effective refractive index, and dispersion coefficient, thereby enabling dynamic optimization and efficient iteration of the dispersive photonic device model during the design process.

[0073] For photonic devices such as on-chip waveguides and linear optical fibers that have significant group delay and dispersion characteristics, this invention proposes a time-domain modeling method based on finite impulse response filters, which can be used to construct a time-domain model that reflects dispersion characteristics in the Verilog-A environment.

[0074] Constructing a parameterized time-domain model includes the following sub-steps:

[0075] (1) Discretization of spectral response: Based on the physical parameters or analytical model of the dispersive device, the amplitude response and phase response at each discrete frequency sampling point are calculated within a preset spectral range to form a discrete frequency domain response.

[0076] The specific method is as follows:

[0077] First, the frequency range required for frequency domain modeling is determined based on the operating bandwidth of the dispersive photonic device, the system sampling rate, and the simulation requirements.

[0078] Subsequently, based on the Nyquist sampling theorem, the number of discrete frequency sampling points and the corresponding frequency resolution are determined within this frequency range, thereby constructing a discrete frequency range for modeling.

[0079] Secondly, at the determined frequency sampling points, the amplitude response and phase response of the dispersive photonic device at each frequency point are calculated using the physical parameters or analytical model of the device, forming a discrete frequency domain response.

[0080] The propagation constant of the on-chip waveguide is shown in the following formula, which can fully characterize the propagation characteristics of the dispersive device as a function of frequency, including the phase change characteristics caused by group delay and higher-order dispersion.

[0081] ;

[0082] in, Represents the propagation constant. Indicates the reference frequency. Represents the center phase constant. Represents the first-order dispersion coefficient. Represents the second-order dispersion coefficient. This represents the third-order dispersion coefficient.

[0083] (2) Determination of time domain response length: Based on the group delay characteristics in the discrete frequency domain response and the system simulation time step, calculate the number of taps required for the time domain finite impulse response model.

[0084] In the process of constructing the time-domain model, this invention uses the obtained frequency-domain response as input and constructs a time-domain model in the form of a finite impulse response by combining time delay and dispersion characteristics. First, based on the group delay characteristics reflected by the frequency-domain phase response, the maximum delay range that the device needs to cover in the time domain is determined, and then based on the group delay... and simulation time step Number of taps required to build a time-domain model As shown in the following formula, the obtained time-domain model can fully encompass the main propagation delay characteristics of the device;

[0085] ;

[0086] in, This indicates the number of taps required for the time-domain model. Indicates group delay. This indicates the simulation time step.

[0087] (3) Frequency domain response resampling: In order to ensure the consistency between the time domain model and the frequency domain model, the discrete frequency domain response obtained in step (1) is resampled based on the number of taps determined in step (2) so that the number of frequency domain sampling points matches the length of the time domain model.

[0088] (4) Subsampling delay compensation: When the group delay is not an integer multiple of the simulation time step, the frequency domain response after resampling is compensated by a fractional delay filter to correct the phase deviation at the subsampling level, thereby ensuring the accuracy of the time domain model in terms of phase and delay.

[0089] (5) Causality violation suppression: Apply a window function to the compensated frequency domain response to suppress non-causal components that may be caused by frequency domain truncation and band-limited sampling.

[0090] Considering that discrete frequency domain sampling and finite bandwidth may introduce non-causal components in the time domain, this invention applies a window function to the frequency domain response when constructing the time domain model to suppress spectral leakage and mitigate the resulting violation of causality.

[0091] (6) Time-domain model generation: The frequency domain response processed in step (5) is converted into a time-domain finite impulse response through inverse Fourier transform to obtain a parameterized time-domain model.

[0092] After processing with a window function, the frequency domain response is converted into the corresponding time domain impulse response, ultimately yielding a time domain model with a finite length.

[0093] Through the above steps, the time-domain model constructed by this invention can not only accurately reflect the overall propagation delay of the dispersive device, but also reflect the influence of the dispersion effect on the time-domain waveform, avoiding the introduction of obvious phase errors or non-physical responses due to discretization processing, thereby obtaining time-domain response results consistent with the frequency-domain model.

[0094] Within an EDA platform, the principle of an optoelectronic integrated Mach-Zehnder modulated transceiver composed of photonic devices and electronic circuits is as follows: Figure 4 As shown, this link uses a laser as its light source. The Y branch and Y combiner are cascaded with upper and lower arm waveguides. Four-level amplitude pulse (PAM4) signal modulation and conversion are achieved through dual-arm pattern modulation. The pattern generator, low-pass filter, and transimpedance amplifier are electronic circuit components, while the rest are Verilog-A photonic devices. Using this schematic, the optoelectronic co-simulation platform of this invention is compared with the mainstream commercial simulation software Lumerical. The frequency domain response results are as follows: Figure 5 As shown in (a) above, the time-domain response results are as follows: Figure 5 As shown in (b) and (c), it can be seen that the frequency domain response and time domain transient waveform obtained by the present invention maintain good consistency in characteristic trends and key indicators, thereby improving the reliability of optoelectronic fusion system analysis.

[0095] like Figure 6 (a) and Figure 7 As shown in Figure (a), the linear fiber model and dispersive waveguide model constructed by testing on the EDA platform are compared with the simulation results of the commercial software Lumerical. Figure 6 (b) and Figure 7 As shown in Figure (b), it can be seen that after adopting the model of the present invention, the time-domain output waveform can correctly reflect the phenomena such as pulse broadening, edge distortion and time delay caused by dispersion. Its trend is consistent with the results of commercial photonic simulation tools and is significantly better than the traditional simplified model.

[0096] Simulation results related to causal violation, such as Figure 8 As shown in (a) and (b), after the above processing, the non-physical forward components in the time domain response are significantly suppressed, and the model output is smoother and more stable.

[0097] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for constructing a parametric optoelectronic co-simulation platform for optoelectronic chip design, characterized in that, In an electronic design automation platform, a unified model framework for photonic devices is constructed based on a hardware description language, and the conversion between frequency domain response and time domain response is implemented within the model. The method includes the following steps: Step 1: Construct a general Verilog A model framework based on the frequency domain response of photonic devices, where the photonic devices are modeled as multi-port linear systems, each port corresponds to the input or output of an optical signal, and the transmission characteristics of the optical signal in the device are described by a scattering matrix; Step 2: Implement a frequency-domain and time-domain numerical conversion mechanism within the Verilog A model, including writing a discrete Fourier transform algorithm and a discrete convolution operation algorithm within the model to convert the frequency domain response of the photonic device into a time-domain impulse response; Step 3: For photonic devices with dispersive characteristics, construct a parameterized time-domain model based on finite impulse response. The key optical parameters of the model are set as user-adjustable input variables to realize dynamic adjustment of device parameters and time-domain simulation within the electronic design automation platform. Step three, constructing the parameterized time-domain model, includes the following sub-steps: (1) Discretization of spectral response: Based on the physical parameters or analytical model of the dispersive device, the amplitude response and phase response at each discrete frequency sampling point are calculated within a preset spectral range to form a discrete frequency domain response; (2) Determination of time domain response length: Based on the group delay characteristics in the discrete frequency domain response and the system simulation time step, calculate the number of taps required for the time domain finite impulse response model; (3) Frequency domain response resampling: Based on the number of taps determined in step (2), the discrete frequency domain response obtained in step (1) is resampled so that the number of frequency domain sampling points matches the length of the time domain model. (4) Subsampling delay compensation: When the group delay is not an integer multiple of the simulation time step, the frequency domain response after resampling is compensated by a fractional delay filter to correct the phase deviation at the subsampling level; (5) Causality violation suppression: Apply a window function to the compensated frequency domain response to suppress non-causal components that may be caused by frequency domain truncation and band-limited sampling; (6) Time-domain model generation: The frequency domain response processed in step (5) is converted into a time-domain finite impulse response through inverse Fourier transform to obtain the parameterized time-domain model.

2. The method for constructing a parametric optoelectronic co-simulation platform for optoelectronic chip design according to claim 1, characterized in that, In step one, the general Verilog A model framework represents optical signals by separating the real and imaginary parts. Each optical signal is split into a real part channel and an imaginary part channel, and the correspondence is realized through port layout. At the same time, it supports the distinction between forward propagation signals and backward propagation signals.

3. The method for constructing a parametric optoelectronic co-simulation platform for optoelectronic chip design according to claim 1, characterized in that, In step one, the general Verilog A model framework supports multi-mode or multi-polarization state extensions, and the evolutionary relationship mapping between different modes is realized through the extension port.

4. The method for constructing a parametric optoelectronic co-simulation platform for optoelectronic chip design according to claim 1, characterized in that, In step two, the Fourier transform employs a discrete frequency-time domain conversion method within the framework of the general Verilog A model. This method represents the frequency domain response as a discrete sampling point sequence and converts the rotation factor in the Fourier transform into a trigonometric function expression. The complex number is separated into real and imaginary components, and the time-domain impulse response is obtained through iterative calculation.

5. The method for constructing a parametric optoelectronic co-simulation platform for optoelectronic chip design according to claim 1, characterized in that, In step two, the convolution operation is implemented by introducing a time-progression function with a fixed time step to achieve discrete convolution calculation, which is used to describe the time delay and memory effect of photonic devices.

6. The method for constructing a parametric optoelectronic co-simulation platform for optoelectronic chip design according to claim 1, characterized in that, In step three, the key optical parameters include at least one of the following: device length, effective refractive index, and dispersion coefficient.

7. The method for constructing a parametric optoelectronic co-simulation platform for optoelectronic chip design according to claim 1, characterized in that, The specific method for discretizing the spectral response in step (1) is as follows: First, the frequency range required for frequency domain modeling is determined based on the operating bandwidth of the dispersive photonic device, the system sampling rate, and the simulation requirements. Subsequently, based on the Nyquist sampling theorem, the number of discrete frequency sampling points and the corresponding frequency resolution are determined within this frequency range, thereby constructing a discrete frequency range for modeling. Secondly, at the determined frequency sampling points, the amplitude response and phase response of the dispersive photonic device at each frequency point are calculated using the physical parameters or analytical model of the device, forming a discrete frequency domain response.

8. The method for constructing a parametric optoelectronic co-simulation platform for optoelectronic chip design according to claim 1, characterized in that, In step (2), the formula for the number of taps required to calculate the finite-impulse response model in the time domain is as follows: ； in, This indicates the number of taps required for the time-domain model. Indicates group delay. This indicates the simulation time step.

Images