A nonlinear calculation method and system based on mach-zehnder modulator
By adjusting the bias point on a Mach-Zehnder modulator and using optical signal processing to achieve nonlinear calculations, the traditional computer memory wall and nonlinear operation problems are solved, providing a fast, low-power solution and reducing dependence on electrical components.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- WUHAN POST & TELECOMM RES INST CO LTD
- Filing Date
- 2023-12-04
- Publication Date
- 2026-07-24
AI Technical Summary
Current integrated circuits face development bottlenecks. Traditional computers suffer from memory wall problems due to data transfer between the CPU and memory, and it is difficult to achieve efficient nonlinear operations. Electronic devices are overly dependent on electrical components.
Nonlinear calculations are achieved using a Mach-Zehnder modulator (MZM) at different bias points. By applying a combination of optical carrier, DC bias voltage, and AC voltage, the voltage difference is adjusted to achieve different bias points, and optical signal processing is performed to obtain nonlinear calculation results.
It achieves fast, low-energy nonlinear computation, reduces dependence on electrical components, and features programmability and high integration, solving the memory wall problem and nonlinear computation challenges of traditional computers.
Smart Images
Figure CN117675023B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of optical fiber communication technology, specifically to a nonlinear calculation method and system based on a Mach-Zehnder modulator. Background Technology
[0002] With the explosive growth of application demands, AI has emerged. Simultaneously, the demand for information processing has increased dramatically, leading to an exponential increase in the requirements for chip computing power and memory. Furthermore, neural networks contain multiple levels of nonlinear operations. If only linear transformations are used, the expressive power of an arbitrary-layer fully connected neural network is no different from that of a single-layer neural network.
[0003] However, the development of integrated circuits has now entered the post-Moore era, and traditional computers are facing development bottlenecks. Furthermore, in traditional von Neumann architecture computing systems, data needs to be moved back and forth between the CPU and memory. Since the CPU has a faster processing speed and the memory access speed is slower, this leads to the so-called "memory wall" problem.
[0004] Given the aforementioned problems faced by electronic devices, how to reduce reliance on electrical components and achieve nonlinear computation has become a pressing issue that needs to be addressed. Summary of the Invention
[0005] This application provides a nonlinear calculation method and system based on Mach-Zehnder modulators, which utilizes the different manifestations of MZM at different bias points to achieve fast response time and low power consumption nonlinear calculation, reducing the dependence on electrical components.
[0006] In a first aspect, embodiments of this application provide a nonlinear calculation method based on a Mach-Zehnder modulator, the nonlinear calculation method based on a Mach-Zehnder modulator comprising:
[0007] The optical carrier generated by the laser is loaded into the inlet of the Mach-Zehnder modulator (MZM).
[0008] A DC bias voltage is applied to the upper and lower arms of the MZM, and the data that needs to be calculated nonlinearly is converted into an electrical analog signal and then applied as an AC voltage to the upper and lower arms of the MZM.
[0009] Adjust the voltage difference between the DC bias voltage applied to the upper and lower arms to place the MZM at different bias points.
[0010] The nonlinear calculation results are obtained based on the output optical signal of the bias point under different states of MZM.
[0011] In conjunction with the first aspect, in one embodiment, adjusting the voltage difference of the DC bias voltage applied to the upper and lower arms to place the MZM at bias points in different states includes:
[0012] Adjust the voltage difference between the two DC bias voltages applied to the upper and lower arms of the MZM, so that:
[0013] The voltage difference is 0, so that the MZM is at the peak bias point where the output light intensity is the maximum;
[0014] Or, the voltage difference is V π This is to ensure that MZM is at the null bias point where the output light intensity is minimized;
[0015] Or, the voltage difference is So that the MZM is at the -Quadrature bias point where the output light intensity is half of the maximum light intensity;
[0016] Or, the voltage difference is So that the MZM is at the Quadrature bias point where the output light intensity is half of the maximum light intensity;
[0017] Among them, V π This is the half-wave voltage of the MZM.
[0018] In conjunction with the first aspect, in one implementation, it further includes:
[0019] The output optical signal is photoelectrically converted and the result after nonlinear calculation by MZM is displayed on an oscilloscope.
[0020] In conjunction with the first aspect, in one embodiment, a photodetector (PD) is used to perform photoelectric conversion on the output optical signal.
[0021] In conjunction with the first aspect, in one embodiment, before performing photoelectric conversion on the output optical signal, a step of amplifying the output optical signal is further included.
[0022] In conjunction with the first aspect, in one embodiment, the output optical signal is amplified using an erbium-doped fiber amplifier (EDFA).
[0023] In conjunction with the first aspect, in one embodiment, the half-wave voltage V of the MZM π The voltage is 4 volts, the insertion loss is 5 dB, and the splitting ratio of the upper and lower arms is: Where, ε r =10 ExtRatio / 10 The extinction ratio (ExtRatio) is 20 dB.
[0024] The initial DC bias voltage applied to both the upper and lower arms of the MZM was set to 0 volts.
[0025] Secondly, embodiments of this application provide a nonlinear computing system based on a Mach-Zehnder modulator, the nonlinear computing system based on a Mach-Zehnder modulator comprising:
[0026] A laser, used to output an optical carrier wave;
[0027] A Mach-Zehnder modulator (MZM) includes an upper arm and a lower arm, and receives the optical carrier based on the splitting ratio of the upper and lower arms.
[0028] A DC power supply is used to apply a DC bias voltage to the upper and lower arms, and to adjust the voltage difference between the DC bias voltages applied to the upper and lower arms to make the MZM biased at different states.
[0029] An AC signal source is used to convert data that needs to be calculated nonlinearly into an electrical analog signal, which is then applied as an AC voltage to the upper and lower arms.
[0030] In conjunction with the second aspect, in one implementation, it further includes:
[0031] A photoelectric conversion unit is used to perform photoelectric conversion on the output optical signal to obtain an electrical signal;
[0032] An oscilloscope is used to receive the electrical signal to display the result after nonlinear calculations by MZM.
[0033] In conjunction with the second aspect, in one implementation, it further includes:
[0034] An optical signal amplification unit is used to amplify the output optical signal and input it to the photoelectric conversion unit.
[0035] The beneficial effects of the technical solutions provided in this application include at least the following:
[0036] The nonlinear calculation method based on a Mach-Zehnder modulator in this application involves loading an optical carrier generated by a laser onto the input port of a Mach-Zehnder modulator (MZM); applying a DC bias voltage to the upper and lower arms of the MZM; converting the data to be calculated into an analog electrical signal and then applying it as an AC voltage to the upper and lower arms of the MZM; adjusting the voltage difference between the DC bias voltages applied to the upper and lower arms to place the MZM at different bias points; and obtaining the nonlinear calculation results based on the output optical signals at the different bias points of the MZM.
[0037] Compared with existing technologies, this application utilizes the different manifestations of MZM at different bias points to achieve nonlinear calculations with fast response time and low power consumption. It has the advantages of programmability, high speed and integrability, and reduces the dependence on electrical components. Attached Figure Description
[0038] Figure 1 This is a flowchart illustrating an embodiment of the nonlinear calculation method based on a Mach-Zehnder modulator according to this application;
[0039] Figure 2 This is a schematic diagram of a typical bias point of the Mach-Zehnder modulator in this application;
[0040] Figure 3 This is a graph showing the output optical signal when the MZM of this application is at the peak bias point;
[0041] Figure 4 This is a graph showing the output optical signal when the MZM of this application is at the null bias point;
[0042] Figure 5 This is a graph showing the output optical signal when the MZM of this application is at the -Quadrature bias point;
[0043] Figure 6 This is a graph showing the output optical signal when the MZM of this application is at the Quadrature bias point;
[0044] Figure 7 This is a structural block diagram of an embodiment of the nonlinear computing system based on a Mach-Zehnder modulator according to this application. Detailed Implementation
[0045] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.
[0046] The terms "comprising" and "having," and any variations thereof, in the specification, claims, and accompanying drawings of this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus. The terms "first," "second," and "third," etc., are used to distinguish different objects, etc., and do not indicate a sequence, nor do they limit "first," "second," and "third" to different types.
[0047] In the description of the embodiments of this application, terms such as "exemplary," "for example," or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary," "for example," or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary," "for example," or "for instance" is intended to present the relevant concepts in a concrete manner.
[0048] In the description of the embodiments of this application, unless otherwise stated, " / " means "or". For example, A / B can mean A or B. The "and / or" in the text is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist simultaneously, and B exists alone. In addition, in the description of the embodiments of this application, "multiple" means two or more.
[0049] In some processes described in the embodiments of this application, multiple operations or steps are included in a specific order. However, it should be understood that these operations or steps may not be executed in the order they appear in the embodiments of this application, or they may be executed in parallel. The sequence number of the operation is only used to distinguish different operations, and the sequence number itself does not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed sequentially or in parallel, and these operations or steps may be combined.
[0050] It is worth noting that optical neural networks have emerged to address the problems faced by existing electronic devices. The high bandwidth and low latency processing capabilities of photonics, combined with the distributed processing of artificial neural networks, provide better computing power. Given the current bandwidth and speed limitations in photoelectric conversion, utilizing existing optical devices such as MZMs (Mach-Zehnder Modulators) for nonlinear calculations offers advantages such as high speed and high integrability.
[0051] MZMs are interference structures made of materials with strong electro-optic effects (such as LiNbO3, GaAs, and InP). Applying an electric field to the arms changes the optical path length, thus achieving phase modulation. Combining two arms with different phase modulations converts phase modulation into intensity modulation.
[0052] MZMs are simple in structure, easy to manufacture, and low in cost. They can also couple well with lasers and optical fibers with low insertion loss, so MZMs are widely used in optical communication.
[0053] This application utilizes MZM to achieve nonlinear computation, reducing reliance on electrical components and providing significant support for seizing the main battlefield of post-Moore's Law semiconductor technology in the field of new intelligent computing.
[0054] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.
[0055] In a first aspect, embodiments of this application provide a nonlinear calculation method based on a Mach-Zehnder modulator.
[0056] In one embodiment, reference is made to Figure 1 , Figure 1 This is a schematic flowchart illustrating an embodiment of the nonlinear calculation method based on a Mach-Zehnder modulator according to this application. Figure 1 As shown, the nonlinear calculation method based on the Mach-Zehnder modulator includes:
[0057] S1. Load the optical carrier generated by the laser into the inlet of the Mach-Zehnder modulator (MZM).
[0058] This embodiment first utilizes a laser to provide a light source, because the MZM is an optical device and requires light to function. In this embodiment, the laser generates a raw optical carrier at 1550nm and 16dBm, which is then split according to the MZM's splitting ratio and loaded onto the MZM's inlet.
[0059] S2. Apply a DC bias voltage to the upper and lower arms of the MZM, and convert the data that needs to be calculated nonlinearly into an electrical analog signal and apply it as an AC voltage to the upper and lower arms of the MZM.
[0060] It is worth noting that the dual-drive MZM has two drive electrodes, which can be controlled by two voltages. The working principle of the dual-drive MZM is as follows:
[0061] Let the incident light be: Where E0 is the amplitude of the incident light and ω0 is the angular frequency of the incident light.
[0062] Let α be the insertion loss of the MZM, γ be the splitting ratio of the upper and lower arms, and V π It is the half-wave voltage of the modulator, V b1 V b2 Let Vt be the DC bias voltage applied to the upper arm and V2 be the DC bias voltage applied to the upper arm and V3 be the AC bias voltage applied to the upper arm and V4 be the AC bias voltage applied to the lower arm, respectively. Then we have:
[0063]
[0064] V1(t) = -V2(t)
[0065] S3. Adjust the voltage difference between the DC bias voltage applied to the upper arm and the lower arm to make the MZM biased at different states.
[0066] It's worth noting that the AC voltage value represents the value to be calculated, and this is used as input information loaded into the upper and lower arms of the MZM. This is achieved by controlling the DC bias voltage V. b1 V b2 The voltage difference causes the MZM to be at different bias points.
[0067] See Figure 2 As shown, in this embodiment, let φ = V b1 -V b2 When φ = 0, the output light intensity is at its maximum, which is called the peak offset point; when φ = V... π When the output light intensity is at its minimum, it is called the null bias point; when When the output light intensity is half of the maximum light intensity, it is called the -Quadrature bias point; when When the output light intensity is half of the maximum light intensity, it is called the +Quadrature bias point.
[0068] S4. Obtain nonlinear calculation results based on the output optical signal of the bias point under different states of MZM.
[0069] Furthermore, to more intuitively display the nonlinear calculation results, in some embodiments, the output optical signal is also photoelectrically converted, and the result after nonlinear calculation by the MZM is displayed on an oscilloscope. Specifically, a photodetector (PD) can be used to perform photoelectric conversion on the output optical signal.
[0070] Furthermore, before performing photoelectric conversion on the output optical signal, the process includes an amplification step. Specifically, an erbium-doped fiber amplifier (EDFA) can be used to amplify the output optical signal. The amplified output optical signal facilitates further processing.
[0071] The following is a specific example to further illustrate this:
[0072] A laser generates 1550nm, 16dBm light, which is applied to the upper and lower arms of a Mach-Zehnder modulator (MZM). Simultaneously, the half-wave voltage V of the MZM is used. π The voltage is 4 volts, the insertion loss is 5 dB, and the splitting ratio of the upper and lower arms is: Where, ε r =10 ExtRatio / 10 The extinction ratio ExtRatio is 20dB; and the initial DC bias voltage applied to the upper and lower arms of the MZM is set to 0V.
[0073] Adjust the voltage difference between the two DC bias voltages applied to the upper and lower arms of the MZM, so that:
[0074] The voltage difference is 0, so that the MZM is at the peak bias point where the output light intensity is the maximum;
[0075] Or, the voltage difference is V π This is to ensure that MZM is at the null bias point where the output light intensity is minimized;
[0076] Or, the voltage difference is So that the MZM is at the -Quadrature bias point where the output light intensity is half of the maximum light intensity;
[0077] Or, the voltage difference is This is to place the MZM at the Quadrature bias point where the output light intensity is half of the maximum light intensity.
[0078] It is worth noting that the aforementioned voltage difference V π , These are all ideal values. In reality, there may be manufacturing errors. In actual implementation, the error between the actual voltage difference and the above ideal value is within the set threshold range.
[0079] When MZM is at the peak bias point, E out The expression for (t) is:
[0080]
[0081] At this time, E out See the image of (t) Figure 3 As shown, Figure 3 The horizontal axis is V1(t), and the vertical axis is E. out The absolute value of (t), the same below.
[0082] When MZM is at the null bias point, E out The expression for (t) is:
[0083]
[0084] At this time, E out See the image of (t) Figure 4 As shown.
[0085] When MZM is at the -Quadrature bias point, E out The expression for (t) is:
[0086]
[0087] At this time, E out See the image of (t) Figure 5 As shown.
[0088] When MZM is at the Quadrature bias point, E out The expression for (t) is:
[0089]
[0090] At this time, E out See the image of (t) Figure 6 As shown.
[0091] As can be seen from the above, MZM exhibits different forms at different bias points, corresponding to different nonlinear equations, thus possessing the ability to perform nonlinear calculations.
[0092] In summary, the nonlinear calculation method based on a Mach-Zehnder modulator in this application involves loading an optical carrier generated by a laser onto the input port of a Mach-Zehnder modulator (MZM); applying a DC bias voltage to the upper and lower arms of the MZM; converting the data to be calculated into an analog electrical signal and then applying it as an AC voltage to the upper and lower arms of the MZM; adjusting the voltage difference between the DC bias voltages applied to the upper and lower arms to place the MZM at different bias points; and obtaining the nonlinear calculation results based on the output optical signals at the different bias points of the MZM.
[0093] Compared with existing technologies, this application utilizes the different manifestations of MZM at different bias points to achieve nonlinear calculations with fast response time and low power consumption. It has the advantages of programmability, high speed and integrability, and reduces the dependence on electrical components.
[0094] Secondly, embodiments of this application also provide a nonlinear computing system based on a Mach-Zehnder modulator.
[0095] In one embodiment, reference is made to Figure 7 , Figure 7 This is a structural block diagram of an embodiment of the nonlinear computing system based on a Mach-Zehnder modulator according to this application. Figure 7 As shown, the nonlinear computing system based on the Mach-Zehnder modulator includes:
[0096] A laser, used to output an optical carrier wave;
[0097] A Mach-Zehnder modulator (MZM) includes an upper arm and a lower arm, and receives the optical carrier based on the splitting ratio of the upper and lower arms.
[0098] A DC power supply is used to apply a DC bias voltage to the upper and lower arms, and to adjust the voltage difference between the DC bias voltages applied to the upper and lower arms to make the MZM biased at different states.
[0099] An AC signal source is used to convert data that needs to be calculated nonlinearly into an electrical analog signal, which is then applied as an AC voltage to the upper and lower arms.
[0100] Furthermore, in one embodiment, it also includes:
[0101] A photoelectric conversion unit is used to perform photoelectric conversion on the output optical signal to obtain an electrical signal;
[0102] An oscilloscope is used to receive the electrical signal to display the result after nonlinear calculations by MZM.
[0103] Preferably, a photodetector (PD) can be used to perform photoelectric conversion on the output optical signal. Of course, other devices can also be used for photoelectric conversion, such as phototransistors, photoresistors, etc., and this embodiment does not impose any limitations.
[0104] Furthermore, in one embodiment, it also includes:
[0105] An optical signal amplification unit is used to amplify the output optical signal and input it to the photoelectric conversion unit.
[0106] Preferably, the output optical signal can be amplified using an erbium-doped fiber amplifier (EDFA).
[0107] In the above-mentioned nonlinear calculation system based on Mach-Zehnder modulator, the functions of each device correspond to the steps in the above-mentioned nonlinear calculation method based on Mach-Zehnder modulator. Their functions and implementation processes will not be described in detail here.
[0108] In summary, the nonlinear calculation system based on a Mach-Zehnder modulator in this application includes a laser, a Mach-Zehnder modulator (MZM), a DC power supply, and an AC signal source. It works by loading an optical carrier generated by the laser onto the input port of the MZM; applying a DC bias voltage to the upper and lower arms of the MZM; converting the data to be calculated into an analog electrical signal and then applying it as an AC voltage to the upper and lower arms of the MZM; adjusting the voltage difference between the DC bias voltages applied to the upper and lower arms to place the MZM at different bias points; and obtaining the nonlinear calculation results based on the output optical signals at the different bias points of the MZM.
[0109] Compared with existing technologies, this application utilizes the different manifestations of MZM at different bias points to achieve nonlinear calculations with fast response time and low power consumption. It has the advantages of programmability, high speed and integrability, and reduces the dependence on electrical components.
[0110] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.
Claims
1. A nonlinear calculation method based on a Mach-Zehnder modulator, characterized in that: The nonlinear calculation method based on the Mach-Zehnder modulator includes: The optical carrier generated by the laser is loaded into the inlet of the Mach-Zehnder modulator (MZM). A DC bias voltage is applied to the upper and lower arms of the MZM, and the data that needs to be calculated nonlinearly is converted into an electrical analog signal and then applied as an AC voltage to the upper and lower arms of the MZM. Adjust the voltage difference between the DC bias voltage applied to the upper and lower arms to place the MZM at different bias points. The nonlinear calculation results are obtained based on the output optical signal of the bias point under different states of MZM; The adjustment of the voltage difference between the DC bias voltage applied to the upper and lower arms, so that the MZM is at different bias points, includes: Adjust the voltage difference between the two DC bias voltages applied to the upper and lower arms of the MZM, so that: The voltage difference is 0, so that the MZM is at the peak bias point where the output light intensity is the maximum; Or, the voltage difference is This is to ensure that MZM is at the null bias point where the output light intensity is minimized; Or, the voltage difference is This is to place the MZM at the -Quadrature bias point where the output light intensity is half of the maximum light intensity; Or, the voltage difference is This is to place the MZM at the Quadrature bias point where the output light intensity is half of the maximum light intensity; in, This is the half-wave voltage of the MZM.
2. The nonlinear calculation method based on Mach-Zehnder modulator as described in claim 1, characterized in that, Also includes: The output optical signal is photoelectrically converted and the result after nonlinear calculation by MZM is displayed on an oscilloscope.
3. The nonlinear calculation method based on Mach-Zehnder modulator as described in claim 2, characterized in that: The output optical signal is converted into photoelectric signal using a photodetector (PD).
4. The nonlinear calculation method based on Mach-Zehnder modulator as described in claim 2, characterized in that: Before performing photoelectric conversion on the output optical signal, the method further includes a step of amplifying the output optical signal.
5. The nonlinear calculation method based on Mach-Zehnder modulator as described in claim 4, characterized in that: The output optical signal is amplified using an erbium-doped fiber amplifier (EDFA).
6. The nonlinear calculation method based on Mach-Zehnder modulator as described in claim 1, characterized in that: The half-wave voltage of the MZM The voltage is 4 volts, the insertion loss is 5 dB, and the splitting ratio of the upper and lower arms is: ,in, Extinction ratio 20dB; The initial DC bias voltage applied to both the upper and lower arms of the MZM was set to 0 volts.
7. A nonlinear computing system based on a Mach-Zehnder modulator, characterized in that, The nonlinear computation system based on the Mach-Zehnder modulator includes: A laser, used to output an optical carrier wave; A Mach-Zehnder modulator (MZM) includes an upper arm and a lower arm, and receives the optical carrier based on the splitting ratio of the upper and lower arms. A DC power supply is used to apply a DC bias voltage to the upper and lower arms, and to adjust the voltage difference between the DC bias voltages applied to the upper and lower arms to make the MZM biased at different states. An AC signal source is used to convert data that needs to be calculated nonlinearly into an electrical analog signal, which is then applied as an AC voltage to the upper and lower arms. The DC power supply adjusts the voltage difference between the DC bias voltage applied to the upper and lower arms to place the MZM at different bias points, including: Adjust the voltage difference between the two DC bias voltages applied to the upper and lower arms of the MZM, so that: The voltage difference is 0, so that the MZM is at the peak bias point where the output light intensity is the maximum; Or, the voltage difference is This is to ensure that MZM is at the null bias point where the output light intensity is minimized; Or, the voltage difference is This is to place the MZM at the -Quadrature bias point where the output light intensity is half of the maximum light intensity; Or, the voltage difference is This is to place the MZM at the Quadrature bias point where the output light intensity is half of the maximum light intensity; in, This is the half-wave voltage of the MZM.
8. The nonlinear computing system based on a Mach-Zehnder modulator as described in claim 7, characterized in that, Also includes: A photoelectric conversion unit is used to perform photoelectric conversion on the output optical signal of the MZM at a bias point to obtain an electrical signal; An oscilloscope is used to receive the electrical signal to display the result after nonlinear calculations by MZM.
9. The nonlinear computing system based on a Mach-Zehnder modulator as described in claim 8, characterized in that, Also includes: An optical signal amplification unit is used to amplify the output optical signal and input it to the photoelectric conversion unit.