Broadband signal measuring instrument

Abstract

By using narrow-bandwidth elements commonly found in highly versatile systems, it is possible to measure broadband signals with high reproducibility. [Solution] This broadband signal measuring instrument comprises a reservoir, a plurality of signal conversion elements connected to the reservoir, and a signal processing unit that processes the signals output from the plurality of signal conversion elements. The reservoir converts an input signal input to the reservoir into a plurality of signals. Each of the plurality of signal conversion elements receives one of the plurality of signals. Each of the plurality of signal conversion elements satisfies at least one of the following two conditions: a first condition that its bandwidth is narrower than the bandwidth of the input signal, and a second condition that its sampling rate is less than twice the bandwidth of the input signal.

Description

Technical Field

[0001] The present invention relates to a broadband signal measuring instrument.

Background Art

[0002] Broadband signals are used to transmit large-capacity information at high speed. In order to receive a broadband signal at a transmission destination, the receiving element also needs to be an element capable of detecting broadband signals.

[0003] Elements capable of detecting broadband signals are expensive, and it is difficult to incorporate them into highly versatile systems. For example, Patent Document 1 discloses a method for extracting a plurality of frequency components contained in an optical signal, which is a broadband signal.

Prior Art Literature

Patent Literature

[0004]

Patent Literature 1

Summary of the Invention

Problem to be Solved by the Invention

[0005] There is a need for a method that can detect broadband signals even with highly versatile systems.

[0006] The present invention has been made in view of the above problems, and an object of the present invention is to provide a broadband signal measuring instrument that can measure broadband signals with good reproducibility using elements used in highly versatile systems.

Means for Solving the Problem

[0007] The present invention provides the following means to solve the above problems.

[0008] A broadband signal measuring instrument according to the first embodiment comprises a reservoir, a plurality of signal conversion elements connected to the reservoir, and a signal processing unit that processes signals output from the plurality of signal conversion elements. The reservoir converts an input signal input to the reservoir into a plurality of signals. Each of the plurality of signal conversion elements receives one of the plurality of signals. Each of the plurality of signal conversion elements satisfies at least one of the following two conditions: a first condition that its bandwidth is narrower than the bandwidth of the input signal, and a second condition that its sampling rate is less than twice the bandwidth of the input signal. [Effects of the Invention]

[0009] The broadband signal measuring instrument according to this embodiment can measure broadband signals with high reproducibility using highly versatile elements. [Brief explanation of the drawing]

[0010] [Figure 1] This is a schematic diagram of the broadband signal measuring instrument according to this embodiment. [Figure 2] This is a schematic block diagram showing a specific example of a signal processing unit. [Figure 3] This is a schematic diagram illustrating the function of the broadband signal measuring instrument according to this embodiment. [Figure 4] This is a schematic diagram illustrating the function of another example of a broadband signal measuring instrument according to this embodiment. [Figure 5] This graph shows the results of Example 2. [Figure 6] This shows the accuracy of the output signal from the broadband signal measuring instrument of Example 2 in relation to the number of photodiodes. [Figure 7] This is a schematic diagram illustrating the function of another example of a broadband signal measuring instrument according to this embodiment. [Figure 8] This graph shows the results of Example 3. [Figure 9] This graph shows the results of Example 4. [Modes for carrying out the invention]

[0011] Hereinafter, the present embodiment will be described in detail. The following description is an example of the present invention, and the present invention is not limited thereto, and can be appropriately modified and implemented without changing the gist of the present invention.

[0012] FIG. 1 is a schematic diagram of a broadband signal measuring instrument 100 according to the present embodiment. The broadband signal measuring instrument 100 includes a reservoir 10, a signal conversion unit 20, and a signal processing unit 30.

[0013] The reservoir 10 is an element, device, or system used as part of reservoir computing. Reservoir computing is a type of recurrent neural network. An input signal S is supplied to the reservoir 10 in is input, and the reservoir 10 stores the input signal S in The reservoir 10 does not need to directly store the input signal S in as it is, and may store the input signal S in as another signal that partially retains the information of the input signal S.

[0014] The input signal S in is, for example, a broadband signal. The input signal S in is, for example, an optical signal. The input signal S in is not limited to optical signals, and may be an electrical signal, a magnetic signal, or the like. The band of the input signal S in is not particularly limited as long as it is a broadband. The band of the input signal S in is, for example, 1 GHz or more, may be 10 GHz or more, or may be 1 THz or more.

[0015] The reservoir 10 has, for example, a plurality of nodes 11. The plurality of nodes 11 are, for example, randomly connected. As the input signal S in propagates between the plurality of randomly connected nodes 11, the input signal S in is replaced with another signal. The replaced other signal retains part of the information of the input signal S in .

[0016] Reservoir 10 is, for example, an optical element in which the concept of reservoir computing is physically implemented. The optical element is, for example, an optical integrated circuit chip fabricated using silicon photonics technology. Reservoir 10 is not limited to an optical element as long as it is in which the concept of reservoir computing is physically implemented, and the input signal S in Anything that matches the requirements is acceptable. For example, reservoir 10 can be an electrical component.

[0017] Reservoir 10 outputs multiple signals. Each of the multiple signals is an input signal S in This is converted from the input signal S in Includes information.

[0018] The signal conversion unit 20 has a plurality of signal conversion elements. The signal conversion unit 20 is connected to the reservoir 10. The plurality of signal conversion elements include, for example, a plurality of first signal conversion elements 22 and a plurality of second signal conversion elements 24.

[0019] Each of the first signal conversion elements 22 is connected to, for example, the reservoir 10. Multiple signals output from the reservoir 10 are input to, for example, one of the multiple first signal conversion elements 22. There is a one-to-one correspondence between the multiple signals output from the reservoir 10 and the multiple first signal conversion elements 22. The signal input to each of the first signal conversion elements 22 is the input signal S in These are different signals that contain information about each other.

[0020] The first signal conversion element 22 converts the first signal to the second signal. For example, it replaces the first signal, which is an optical signal, with the second signal, which is an electrical signal. For example, the first signal conversion element 22 is a photoelectric converter. For example, the first signal conversion element 22 is a photodiode. Here, we illustrate the case where the first signal conversion element 22 replaces an optical signal with an electrical signal, but it is not limited to this case.

[0021] Each of the second signal conversion elements 24 is connected, for example, to one of the first signal conversion elements 22. The second signal converted by the first signal conversion elements 22 is input to, for example, one of the multiple second signal conversion elements 24. The first signal conversion elements 22 and the second signal conversion elements 24 correspond to each other, for example, one-to-one.

[0022] The second signal conversion element 24 converts the second signal into a third signal. For example, the second signal conversion element 24 converts an analog signal into a digital signal. The analog signal is, for example, an electrical signal converted by the first signal conversion element 22. For example, the second signal conversion element 24 is an analog-to-digital converter. Here, we illustrate the case where the second signal conversion element 24 replaces an analog signal with a digital signal, but it is not limited to this case.

[0023] Multiple signal conversion elements constituting the signal conversion unit 20 have a bandwidth that matches the input signal S in The first condition is that the bandwidth is narrower than the input signal S, and the sampling rate is narrower than the input signal S. in The second condition is that it is less than twice the bandwidth of and that at least one of the following conditions is met.

[0024] First, let's explain the first condition. Generally, a narrowband signal conversion element is used to convert a wideband input signal S in Attempting to convert this results in the loss of much of the high-speed information from the input signal S. in It is not possible to accurately restore the input signal S. in Even when using multiple signal conversion elements with a bandwidth narrower than the bandwidth, machine learning can be used to convert the input signal S in It can be restored with high accuracy.

[0025] The bandwidth of the first signal conversion element 22 is, for example, the input signal S in It is narrower than the bandwidth of the first signal conversion element 22. The bandwidth of the first signal conversion element 22 is, for example, the input signal S inThe bandwidth may be 1 / 10 or less, 1 / 100 or less, or 1 / 1000 or less. For example, the bandwidth of the first signal conversion element 22 may be 1 GHz or less, 10 MHz or less, or 1 MHz or less.

[0026] The bandwidth of the second signal conversion element 24 is, for example, the input signal S in It is narrower than the bandwidth of the second signal conversion element 24. The bandwidth of the second signal conversion element 24 is, for example, the input signal S in The bandwidth may be 1 / 10 or less, 1 / 100 or less, or 1 / 1000 or less. For example, the bandwidth of the second signal conversion element 24 may be 1 GHz or less, 10 MHz or less, or 1 MHz or less.

[0027] The bandwidth of the second signal conversion element 24 may be approximately the same as that of the first signal conversion element 22. Here, "approximately the same" means that the ratio obtained by dividing the bandwidth of the second signal conversion element 24 by the bandwidth of the first signal conversion element 22 is between 0.1 and 10. For example, the bandwidth of the second signal conversion element 24 is preferably greater than or equal to the bandwidth of the first signal conversion element 22, and may be the same as that of the first signal conversion element 22.

[0028] Note that if the first signal conversion element 22 satisfies the first condition, the second signal conversion element 24 does not need to satisfy the first condition. Similarly, if the second signal conversion element 24 satisfies the first condition, the second signal conversion element 24 does not need to satisfy the first condition. Both the first signal conversion element 22 and the second signal conversion element 24 may satisfy the first condition. When the signal conversion unit 20 satisfies both the first and second conditions, the input signal S in The difficulty of restoring it increases. On the other hand, when the signal conversion unit 20 satisfies both the first and second conditions, the input signal S in If the signal can be sufficiently restored, less expensive components can be applied to the signal conversion unit 20, increasing the versatility of the broadband signal measuring instrument.

[0029] Next, let's explain the second condition. The sampling rate is a value that represents how many times per second data is taken when an analog signal is digitized. In other words, the second condition is a requirement imposed on the analog-to-digital converter, and for example, the second signal conversion element 24 described above is required to satisfy it. For example, if the first signal conversion element 22 is an element that converts light, which is an analog signal, into electricity, which is an analog signal, then the concept of sampling rate does not apply to the first signal conversion element 22.

[0030] Generally, when converting an analog signal to a digital signal using the sampling theorem, the signal conversion element receives the input signal S in A sampling rate of at least twice the maximum frequency (upper limit of bandwidth) is required. For example, the second signal conversion element 24 receives the input signal S in A sampling rate of more than twice the bandwidth is required. In contrast, the wideband signal measuring instrument 100 according to this embodiment uses the input signal S in Even when using multiple signal conversion elements with sampling rates smaller than twice the bandwidth, machine learning can be used to analyze the input signal S in It can be restored with high accuracy.

[0031] The sampling rate of the second signal conversion element 24 is, for example, the input signal S in The sampling rate of the second signal conversion element 24 may be less than twice the bandwidth. in The bandwidth may be smaller than the input signal S in The sampling rate of the second signal conversion element 24 may be less than 1 / 10, less than 1 / 100, or less than 1 / 1000 of the bandwidth. For example, the sampling rate of the second signal conversion element 24 may be less than 1 GHz, less than 10 MHz, or less than 1 MHz.

[0032] The signal conversion unit only needs to satisfy either the first or second condition. For example, the bandwidth of the first signal conversion element 22 and the second signal conversion element 24 is equal to the input signal S in If the bandwidth is narrower than that, the sampling rate of the second signal conversion element 24 is not particularly important. Also, for example, if the sampling rate of the second signal conversion element 24 is narrower than that of the input signal Sin If the bandwidth is less than twice the bandwidth of the input signal S, the bandwidth of the first signal conversion element 22 and the second signal conversion element 24 will be less than twice the bandwidth of the input signal S. in The bandwidth may be equivalent to that of the input signal S. in The bandwidth is narrower than that, and the sampling rate of the second signal conversion element 24 is the input signal S in It may be less than twice the bandwidth.

[0033] The signal processing unit 30 is connected to the signal conversion unit 20. Figure 2 is a schematic block diagram showing a specific example of the signal processing unit 30. The signal processing unit 30 is composed of, for example, a personal computer or a server device. The signal processing unit 30 includes, for example, an input unit 31, an output unit 32, a storage unit 33, and a control unit 34.

[0034] The input section 31 is the part where signals are input from the signal conversion section 20. For example, signals output from each of the second signal conversion elements 24 are input to the input section 31. The input section 31 is, for example, an interface that handles the connection with the signal conversion section 20.

[0035] The output unit 32 outputs information in a format that is recognizable to the user. The output unit 32 outputs the estimated solution. The output unit 32 may be, for example, an image display device, or an interface for connecting an image display device to the broadband signal measuring instrument 100. In this case, the output unit 32 generates a video signal for displaying image data and outputs the video signal to the image display device connected to it.

[0036] The storage unit 33 is configured using a storage medium such as a magnetic hard disk drive or a semiconductor memory device. The storage unit 33 stores data used for controlling the broadband signal measuring instrument 100. The storage unit 33 also stores data necessary for the control unit 34 to perform processing.

[0037] The memory unit 33 includes, for example, a training data storage unit 33A and a coupling coefficient storage unit 33B.

[0038] The training data storage unit 33A stores pre-input training data. Training data is not always necessary, and in this case, the training data storage unit 33A may be omitted. The coupling coefficient storage unit 33B stores, for example, the weights (coupling coefficients) to be applied to the signals output from each of the second signal conversion elements 24. The coupling coefficient storage unit 33B may update the stored weights after each learning process.

[0039] The control unit 34 is composed of a processor 34A, such as a CPU (Central Processing Unit), and memory 34B. The control unit 34 functions when the processor 34A executes a program. Note that all or part of the functions of the control unit 34 may be implemented using hardware such as an ASIC (Application Specific Integrated Circuit), PLD (Programmable Logic Device), or FPGA (Field Programmable Gate Array).

[0040] The program is stored in memory 34B. Memory 34B may be a computer-readable recording medium. Computer-readable recording media include, for example, portable media such as flexible disks, magneto-optical disks, ROMs, CD-ROMs, and semiconductor memory devices (e.g., SSDs: Solid State Drives), as well as storage devices such as hard disks and semiconductor memory devices built into computer systems. The above program may be transmitted to the broadband signal measuring instrument 100 via a telecommunications line.

[0041] Memory 34B includes a learning program that performs learning processing for the wideband signal measuring instrument 100, and an inference program that performs inference processing.

[0042] Next, the functions of the broadband signal measuring instrument 100 according to this embodiment will be described. The broadband signal measuring instrument 100 performs learning processing and inference processing. Figure 3 is a schematic diagram illustrating the functions of the broadband signal measuring instrument 100 according to this embodiment.

[0043] First, let's explain the learning process. First, the wideband signal measuring instrument 100 receives the input signal S of waveform u(t). in Input signal S. in The bandwidth is determined by the input signal S that is input to the broadband signal measuring instrument 100 during actual inference. in It is preferable that the bandwidth is close to that. Input signal S in This is input to reservoir 10. Input signal S in For example, this is an optical signal.

[0044] Input signal S input to reservoir 10 in It is distributed to multiple signals. For example, input signal S in If the input signal is an optical signal, the optical signal can be distributed into multiple optical signals by using an optical integrated circuit chip fabricated using silicon photonics technology. Each of the distributed signals is the input signal S in This includes information about the distribution of signals. Each of the distributed signals does not need to be the same signal; for example, they may be different signals.

[0045] The signal output from the reservoir 10 is input to one of the elements of the signal conversion unit 20. For example, the signal output from the reservoir 10 is input to the first signal conversion element 22. When the broadband signal measuring instrument 100 satisfies the first condition, the bandwidth of the first signal conversion element 22 is narrower than the bandwidth of the signal input to the first signal conversion element 22. In this case, the narrowband first signal conversion element 22 loses high-speed information when converting the input signal. The narrowband first signal conversion element 22 cannot accurately reconstruct the input signal. For example, a narrowband photoelectric conversion element cannot accurately reconstruct a broadband optical signal as an electrical signal.

[0046] In the signal conversion unit 20, the signal converted by the first signal conversion element 22 may be further converted by the second signal conversion element 24. For example, the second signal conversion element 24 may convert the analog signal converted into an electrical signal by the first signal conversion element 22 into a digital signal. When the broadband signal measuring instrument 100 satisfies the second condition, the sampling rate of the second signal conversion element 24 is the input signal Sin It may be less than twice the bandwidth.

[0047] The signal processing unit 30 processes the output signal from the signal conversion unit 20. The signal processing in the signal processing unit 30 may be linear regression, nonlinear regression, deep learning, or a combination of these. The signal processing unit 30 processes the output signal y(t) (=S) based on the state signal x(t) from the signal conversion unit 20. out The signal processing unit 30 restores the input signal u(t)(=S in ) and output signal y(t)(=S out The degree of agreement with ) is compared. The signal processing unit 30, during the learning process, processes the input signal S in and output signal S out The weights (coupling coefficients) applied to the signals output from each of the second signal conversion elements 24 are adjusted to increase the degree of agreement. During the learning process, multiple input signals S in This process is repeated using [this method] to derive the optimal values ​​for the weights (coupling coefficients).

[0048] Here, we will explain how to derive the optimal values ​​of the weights (coupling coefficients). For an input signal u(t) (0≦t≦T), the state signal from the i-th signal conversion element (a signal conversion element connected in parallel to the signal processing unit 30: for example, the second signal conversion element 24) is x i Let (t)(1≦i≦N).

[0049] In this case, the output signal y(t) is expressed by the following equation. y(t)=w T ·x(t) Here, w is w = [w1, w2, ..., w N Represented by ], state signal x i This is a weight vector that combines the weights applied to each of (t). Also, x(t) is the state signal x i This is a state signal vector formed by combining each of (t).

[0050] For example, if the broadband signal measuring instrument 100 satisfies the first condition, the weight update (derivation of the optimal value) may be performed based on the above formula using only the state at a certain time t.

[0051] For example, if the broadband signal measuring instrument 100 satisfies the second condition, the sampling time information may be added and the above equation may be evolved over time to obtain the following equation.

[0052]

number

[0053] In the above equation, Δt is the sampling time (1 / sampling rate), x(jΔt) is the state signal vector which combines each of the state signals, and w ij This is a weight matrix that combines the weight vectors for each time step.

[0054] As described above, by evolving the weight update formula over time, it is possible to recover information from times when sampling was not performed, even when the sampling rate is small.

[0055] Next, the inference process will be explained. During the inference process, the weights (coupling coefficients) applied to the signals output from each of the second signal conversion elements 24 are fixed based on the learning results.

[0056] During inference, the input signal S is sent to the broadband signal measuring instrument 100. in The input signal S is received. in This is the signal to be detected. Input signal S in For example, this is a broadband signal. Input signal S in This is, as with learning, for example, a light signal.

[0057] Even during inference, the input signal S inThe signal is input to the reservoir 10 and distributed to multiple signals. The signal output from the reservoir 10 is input to one of the elements of the signal conversion unit 20. The signal conversion in the signal conversion unit 20 is the same as during learning. For example, the signal output from the reservoir 10 is converted by the first signal conversion element 22. The signal converted by the first signal conversion element 22 may be further converted by the second signal conversion element 24. For example, the bandwidth of the first signal conversion element 22 and the second signal conversion element 24 is the input signal S in The bandwidth may be narrower than that. Narrowband first signal conversion element 22 and second signal conversion element 24 lose high-speed information when converting the input signal. Also, for example, the sampling rate of the second signal conversion element 24 is the input signal S in The sampling rate can be less than twice the bandwidth. The second signal conversion element 24 with a small sampling rate loses information about the timing that is not sampled when converting the input signal.

[0058] The signal processing unit 30 processes the output signal from the signal conversion unit 20. The signal processing unit 30 processes the signal based on the learning results. The signal processing unit 30 outputs the result of the signal processing as an output signal S out It outputs as follows. The input signal S is processed through learning. in and output signal S out Since the weights (coupling coefficients) are determined so that they match, the broadband signal measuring instrument 100 outputs the signal S out It can infer with high accuracy.

[0059] For example, if the first signal conversion element 22 converts the optical signal to an electrical signal, the electrical signal becomes the output signal S. out It outputs as an electrical signal S. out The input signal S is an optical signal. in The waveform is restored with high accuracy. Furthermore, even when the optical signal is replaced with an electrical signal by the first signal conversion element 22, the output signal S is restored after the electrical signal is converted back to an optical signal. out It may also be output as follows. In this case, the input signal S in and output signal S out These are all optical signals and show a high degree of agreement.

[0060] The broadband signal measuring instrument 100 according to this embodiment receives an input signal S in Even when using narrowband first signal conversion elements 22 and second signal conversion elements 24, machine learning can be used to convert a wideband input signal S in It can accurately restore the input signal S. in Even when using a second signal conversion element 24 with a small sampling rate, machine learning can be used to obtain a wideband input signal S in It can accurately restore the signal. Although wideband elements or elements with a high sampling rate are expensive, the wideband signal measuring instrument 100 according to this embodiment can use relatively inexpensive and versatile elements, thus offering excellent versatility.

[0061] Although preferred embodiments of the present invention have been described in detail above, the present invention is not limited to any particular embodiment, and various modifications and changes are possible within the scope of the gist of the present invention as described in the claims.

[0062] For example, up to this point, the input signal S in Although the example given was that the input signal S is an optical signal, in This can also be an electrical signal.

[0063] Figure 4 is a schematic diagram illustrating the function of another example of a broadband signal measuring instrument according to this embodiment. The broadband signal measuring instrument 101 shown in Figure 4 has an input signal source 40. The input signal source 40 has, for example, a light source 41 and a modulator 42. The light source 41 is, for example, a laser light source. The wavelength of the light output from the laser light source may be a single wavelength or multiple wavelengths. The modulator 42 receives the input signal S, which is an electrical signal. in The input signal S is converted into an optical signal. The modulator 42 is, for example, a high-speed optical phase modulator, a Mach-Zehnder optical modulator, etc. In the broadband signal measuring instrument 101, the configuration of the reservoir 10, the signal conversion unit 20, and the signal processing unit 30 is the same as in the broadband signal measuring instrument 100. The broadband signal measuring instrument 101 takes an input signal S, which is an electrical signal, as an optical signal. in It can be restored with high accuracy.

[0064] In Figure 4, the electrical signal is converted to an optical signal before being input to the reservoir 10, but the electrical signal may also be input directly. In this case, the reservoir 10 is an electrical element in which the concept of reservoir computing is physically implemented. Also in this case, since the signal conversion unit 20 does not need to replace the optical signal with an electrical signal, the signal conversion unit 20 does not need to have the first signal conversion element 22. In this case, the second signal conversion element 24 only needs to satisfy at least one of the first and second conditions. [Examples]

[0065] (Example 1) First, a silicon photonic reservoir chip was prepared as reservoir 10. Seven 1MHz bandwidth photodiodes were prepared as the first signal conversion element 22. A 1MHz bandwidth analog-to-digital converter was prepared as the second signal conversion element 24.

[0066] A high-speed signal in the 10 GHz band was input to reservoir 10. Reservoir 10 has one input waveguide and multiple output waveguides. A photodiode is installed in each output waveguide. The electrical signal converted by the photodiode is converted into a digital signal by an analog-to-digital converter and input to the signal processing unit 30.

[0067] The signal processing unit 30 output an inference result based on the input signal and the weights determined during training. As a result, it was confirmed that the input signal and the inferred output signal matched with high accuracy.

[0068] (Example 2) Example 2 involved a similar experiment using the system shown in Figure 4. A laser beam with a wavelength of 1550 nm was output from the light source 41. A high-speed optical phase modulator (AWG70002 (25 GS / s)) was used as the modulator 42.

[0069] Figure 5 shows the results of Example 2. The top graph in Figure 5 is the waveform of the input signal, and the middle graph is the waveform of the inferred output signal. The bottom graph in Figure 5 shows the result of converting a 10 GHz band input signal with a 1 MHz band narrowband photodiode.

[0070] As shown in the bottom graph of Figure 5, it is difficult to reconstruct the input signal when a broadband signal is received by a narrowband element. On the other hand, the broadband signal measuring instrument according to Example 2 was able to reconstruct the input signal with high accuracy despite using a narrowband element, as shown in the middle graph of Figure 5.

[0071] Furthermore, using the same system, the number of signals output from reservoir 10 and the number of photodiodes were virtually increased, and the accuracy of signal reconstruction was measured. Figure 6 shows the accuracy of the output signal from the broadband signal measuring instrument of Example 2 with respect to the number of photodiodes. The vertical axis is the normalized mean squared error (NMSE), and the horizontal axis is the number of signals output from reservoir 10 and the number of photodiodes (PD). NMSE is the normalized mean squared error between the original signal and the reconstructed signal. This experiment was also conducted with three patterns: the noise ratio in the original signal was 10%, 50%, and 100%. The higher the noise ratio in the original signal, the more difficult it is to predict the behavior of the original signal and the more difficult it is to reconstruct it. As shown in Figure 6, in all cases, the accuracy of output signal reconstruction improved as the number of signals output from reservoir 10 and the number of photodiodes increased.

[0072] (Example 3) Example 3 was conducted using the experimental setup shown in Figure 7. The light source 41 is a laser capable of emitting multiple different wavelengths. The laser outputs three wavelengths: 1550 nm, 1550.1 nm, and 1549.9 nm. This differs from Example 2 in that the laser light output from the light source 41 has multiple wavelengths. In Example 3, the signal output from the reservoir 10 was decoupled by wavelength using an array waveguide grating (AWG) 26, and each decoupled signal was input to a photodetector. The bandwidth of the photodetector was set to 1 GHz.

[0073] Figure 8 shows the results of Example 3. The top graph in Figure 8 is the waveform of the input signal, and the middle graph is the waveform of the inferred output signal. The bottom graph in Figure 8 shows the result of converting a 10 GHz band input signal with a 1 GHz band narrowband photodiode.

[0074] As shown in the bottom graph of Figure 8, it is difficult to reconstruct the input signal when a broadband signal is received by a narrowband element. On the other hand, the broadband signal measuring instrument according to Example 3 was able to reconstruct the input signal with high accuracy despite using a narrowband element, as shown in the middle graph of Figure 8.

[0075] Furthermore, in Example 3, since light of multiple wavelengths is used, higher-dimensional dispersion can be achieved with a single reservoir. If higher-dimensional dispersion can be achieved with a single reservoir, the size of the reservoir can be suppressed. For example, when it is necessary to recover a signal with a higher noise level, it is necessary to distribute more signals with the reservoir, which can lead to a larger reservoir size. In such cases, the configuration of Example 3 is useful.

[0076] (Example 4) Example 4 differs from Example 1 in that the bandwidths of the first signal conversion element 22 and the second signal conversion element 24 are made equivalent to the bandwidth of the input signal, and the sampling rate of the second signal conversion element 24 is changed. The number of first signal conversion elements 22 and second signal conversion elements 24 is set to 50 each.

[0077] f is twice the maximum frequency of the signal being measured. nyq The sampling rate f of the second signal conversion element 24 is set as follows. sample f nyq This was set to 1 / 50th of the original value. In other words, the second signal conversion element 24 performs sampling only once every 50 points for the input signal.

[0078] FIG. 9 shows the results of Example 4. The "Target" in Example 4 is the waveform of an input signal to be measured. "Reconstructed" is a signal obtained by reconstructing the input signal using the wideband signal measuring instrument of Example 4. "Sampled" is f nyq which is a signal obtained by reconstructing the input signal without performing learning or the like, using only one of the 1 / 50 signal conversion elements 24.

[0079] As can be seen from the graph indicated by "Sampled" in FIG. 9, even when a wideband signal is received by an element with a low sampling rate, it is difficult to reconstruct the input signal. On the other hand, the graphs of "Target" and "Reconstructed" substantially match. That is, it was confirmed that the wideband signal measuring instrument according to Example 4 can reconstruct the input signal with high accuracy despite the use of an element having a low sampling rate. Further, when using the second signal conversion element 24 having a 1 / n sampling rate of f nyq , it was confirmed that if n pieces of the second signal conversion elements 24 are used, the input signal can be reconstructed with high accuracy. [Industrial Applicability]

[0080] The wideband signal measuring instrument according to the present embodiment is expected to be applied to scientific fields requiring high-speed signal detection for optical communication, B5G and 6G applications, detection of high-speed phenomena that are difficult to measure (e.g., high-speed sensing such as chemical reaction processes), and used in industrial fields requiring high-speed oscilloscopes. [Description of Symbols]

[0081] 10 Reservoir 11 Node 20 Signal Conversion Unit 22 First Signal Conversion Element 24 Second Signal Conversion Element 26 Arrayed Waveguide Grating 30 Signal Processing Unit 31 Input Unit 32 Output Unit 33 Storage Unit 33A Teacher Data Storage Unit 33B Coupling Coefficient Storage Unit 34 Control Unit 34A Processor 34B memory 40 Input signal sources 41 Light source 42 Modulators 100, 101 Broadband signal measuring instrument S in Input signal S out Output signal

Claims

1. The system comprises a reservoir, a plurality of signal conversion elements connected to the reservoir, and a signal processing unit that processes signals output from the plurality of signal conversion elements. The reservoir converts the input signal input to the reservoir into multiple signals, Each of the plurality of signal conversion elements receives one of the plurality of signals, A broadband signal measuring instrument in which each of the plurality of signal conversion elements satisfies at least one of the following two conditions: a first condition that the bandwidth is narrower than the bandwidth of the input signal, and a second condition that the sampling rate is less than twice the bandwidth of the input signal.

2. The broadband signal measuring instrument according to claim 1, wherein each of the plurality of signal conversion elements satisfies the first condition.

3. The broadband signal measuring instrument according to claim 1, wherein each of the plurality of signal conversion elements satisfies the second condition.

4. The broadband signal measuring instrument according to claim 1, wherein the bandwidth of each of the plurality of signal conversion elements is 1 / 10 or less of the bandwidth of the input signal.

5. The broadband signal measuring instrument according to claim 1, wherein the sampling rate of each of the plurality of signal conversion elements is 1 / 10 or less of the bandwidth of the input signal.

6. The broadband signal measuring instrument according to claim 1, wherein each of the plurality of signal conversion elements includes a photoelectric conversion element.

7. The broadband signal measuring instrument according to claim 1, wherein each of the plurality of signal conversion elements includes an analog-to-digital converter.

8. The broadband signal measuring instrument according to claim 1, wherein the input signal is an optical signal.

Images