Optical transmitting / receiving device, optical communication method, and computer program
The proposed symbol determination unit with layered region selection and determination circuits optimizes machine learning for reduced calculation times in optical communication systems, addressing the inefficiencies in existing multilevel signal processing.
Patent Information
- Application Number
- JP2023568880
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2021-12-22
- Publication Date
- 2024-12-04
- Estimated Expiration
- 2041-12-22
AI Technical Summary
Existing optical communication systems using digital coherent technology face increased calculation times due to the increased number of internal variables in machine learning circuits, particularly in multilevel signal reception, despite reducing the number of SVM circuits.
A received symbol determination unit with multiple layers of region selection and determination circuits, utilizing machine learning to classify received signals into regions, reducing the number of internal variables and calculations by optimizing the number of machine learning circuits required.
Enables symbol determination of multilevel signals in less calculation time, improving processing efficiency in optical communication systems.
Smart Images

Figure 0007594224000003 
Figure 0007594224000004 
Figure 0007594224000005
Abstract
Description
[Technical field]
[0001] The present invention relates to an optical transmitting / receiving device, an optical communication method, and a computer program. [Background technology]
[0002] In the field of optical communications, large-capacity multi-level optical communications using digital coherent technology are being widely introduced (e.g., Non-Patent Document 1). Digital coherent technology is the application of elemental technologies of wireless communications to optical communications. Digital coherent technology is subject to the effects of various dispersions and nonlinear characteristics of optical fibers due to the unique condition of optical communications that optical fibers are used for the transmission path. This causes a problem of distortion of the received waveform.
[0003] To address this issue, methods of applying machine learning to compensate for distortion are being considered (for example, Non-Patent Document 2 and Non-Patent Document 3). Multi-class classification is required to determine whether a multi-value signal such as 16QAM or 64QAM is received. Therefore, as the number of values increases, the number of internal variables in the machine learning circuit increases, and the calculation time tends to become longer. Non-Patent Document 2 and Non-Patent Document 3 use a method that uses a support vector machine (hereinafter referred to as "SVM") for machine learning and reduces the number of SVM circuits used.
[0004] Fig. 12 is a diagram showing an example of a conventional circuit configuration using an SVM circuit. Fig. 13 is a diagram showing an example of a signal point arrangement in 16QAM. Note that Figs. 12 and 13 are drawings quoted from Non-Patent Document 2. In the following explanation, a configuration for determining a 16QAM signal consisting of signal points 0 to 15 will be explained as an example. Signal points 0 to 15 can be expressed in binary by 4 bits, 0000 to 1111. In the case of the signal point arrangement in Fig. 13, the upper 2 bits indicate the position in the x-axis direction, and the lower 2 bits indicate the position in the y-axis direction.
[0005] First, consider the most significant two bits. If the signal points are divided according to whether the most significant bit of the most significant two bits is "0" or "1", they will be divided into two regions. This is then classified into two using SVM. Next, if the signal points are divided according to whether the least significant bit of the most significant two bits is "0" or "1", they will similarly be divided into two regions. This is then classified into two using SVM. The same process is performed for the y-axis. As a result, if a total of four SVM classifiers are used on both the x and y-axes, 16 signal points can be determined. [Prior art documents] [Non-patent literature]
[0006] [Non-Patent Document 1] S. Tsukamoto et al., OFC2005, PDP29. [Non-Patent Document 2] M. Li et al., IEEE Photon. J., vol. 5, no. 6, 2013. [Non-Patent Document 3] Y. Han et al., IEEE Photon. J, vol6, no. 5, 2014. Summary of the Invention [Problem to be solved by the invention]
[0007] However, even if the number of machine learning circuits could be reduced, the amount of calculation in each machine learning circuit was not taken into consideration, which resulted in an increase in the calculation execution time.
[0008] In view of the above circumstances, an object of the present invention is to provide a technique that can realize symbol determination of a multilevel signal for a received optical signal in a shorter calculation time. [Means for solving the problem]
[0009] One aspect of the present invention includes a received symbol decision unit that performs symbol decision on a multilevel signal for a received optical signal and restores data based on a result of the symbol decision, the received symbol decision unit including region selection layers from 1st layer to Mth layer (M is an integer equal to or greater than 1) and one region decision layer, the mth layer (m is an integer equal to or greater than 1 and equal to or less than M) having the same number of region selection circuits as the number of region divisions in the m-1th layer region selection layer ("1" when m=1), the mth layer region selection circuit selecting the received multilevel signal from among a plurality of regions according to the number of layers (1 to M) an optical transmission / reception device that selects a region to which a symbol of a multi-value signal belongs based on a result of machine learning and outputs the region to a next layer in accordance with the selected region, wherein a region selection circuit of the m-1th layer performs processing according to the region selected by the region selection circuit of the m-1th layer, and the region determination layer includes region determination circuits in accordance with a total number of outputs of the region selection circuit in the region selection layer of the Mth layer, and the region determination circuit selects a region to which a symbol of the multi-value signal received from the region selection layer of the Mth layer belongs based on a result of machine learning and outputs a determination result indicating the symbol of the multi-value signal in accordance with the selected region.
[0010] One aspect of the present invention is an optical communication device comprising a received symbol decision unit that performs symbol decision on a multi-level signal for a received optical signal and restores data based on a result of the symbol decision, the received symbol decision unit comprising region selection layers 1 to M-th layers (M is an integer equal to or greater than 1) and one region decision layer, the m-th layer (m is an integer equal to or greater than 1 and equal to or less than M) comprising region selection circuits in the same number as the number of region divisions in the m-1th layer region selection layer ("1" when m=1), the region decision layer comprising region decision circuits corresponding to the total number of outputs of the region selection circuits in the M-th layer region selection layer, the area determination circuit selecting an area to which the symbol of the multi-value signal received from the area selection layer of the Mth layer belongs based on the result of machine learning, from among a plurality of areas according to the number of layers (1 to M), and outputting the area to a next layer according to the selected area, wherein the area selection circuit of the mth layer performs processing according to the area selected by the area selection circuit of the m-1th layer.
[0011] One aspect of the present invention is a computer program for causing a computer to function as the optical transmitting and receiving device described above. Effect of the Invention
[0012] According to the present invention, it is possible to realize symbol determination of a multilevel signal for a received optical signal in a shorter calculation time. [Brief description of the drawings]
[0013] [Figure 1] 1 is a diagram illustrating an example of a system configuration of an optical communication system 100 according to the present invention. [Diagram 2] 2 is a diagram illustrating a detailed configuration example of a transmission signal selection unit 121. FIG. [Diagram 3] 2 is a diagram showing a detailed configuration example of a received symbol decision unit 212. FIG. [Figure 4]2 is a diagram illustrating an example of a detailed configuration of a determination unit 2122. FIG. [Diagram 5] FIG. 2 is a diagram showing a specific example of the configuration of an area selection circuit 41. [Figure 6] FIG. 4 is a diagram showing an example of the configuration of area determination circuits 51 and 412 using SVM. [Figure 7] FIG. 4 is a diagram showing a configuration example of area determination circuits 51 and 412 using NN. [Figure 8] FIG. 13 is a diagram showing a configuration example in which two-branch area selection circuits 41 are combined in three layers to determine a 16QAM signal. [Figure 9] FIG. 13 is a diagram showing an outline of the process of a configuration in which two-branch area selection circuits 41 are combined in three layers to determine a 16QAM signal. [Figure 10] FIG. 1 is a diagram illustrating the principle of SVM. [Figure 11] FIG. 13 is a diagram showing a simulation result. [Figure 12] FIG. 1 is a diagram illustrating an example of a conventional circuit configuration using an SVM circuit. [Figure 13] FIG. 1 is a diagram illustrating an example of a signal point arrangement in 16QAM. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0014] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described in detail with reference to the drawings. 1 is a diagram showing an example of a system configuration of an optical communication system 100 according to the present invention. The optical communication system 100 includes a transmitting device 10, a receiving device 20, and a control device 30.
[0015] The transmitting device 10 includes a multiplexing device 11 and an optical modulation device 12. The optical modulation device 12 includes a transmission signal selection unit 121 and an optical modulation unit 122. Optical data signals transmitted from a plurality of terminal devices (not shown) are input to the multiplexing device 11 of the transmitting device 10. The multiplexing device 11 generates an optical multiplexed data signal by multiplexing the inputted plurality of optical data signals, and outputs the optical multiplexed data signal to the optical modulation device 12. The transmission signal selection unit 121 of the optical modulation device 12 outputs either the optical multiplexed data signal or a training signal to the optical modulation unit 122 according to the control of the control device 30. The optical modulation unit 122 modulates the optical signal output from the transmission signal selection unit 121 and transmits it.
[0016] The receiving device 20 includes an optical demodulator 21 and a demultiplexer 22. The optical demodulator 21 includes a dispersion compensator 211 and a received symbol decision unit 212. The receiving device 20 receives an optical signal transmitted from the transmitting device 10. The dispersion compensator 211 performs dispersion compensation processing and modulation dimension separation processing such as XY polarization separation on the received optical signal. The received symbol decision unit 212 performs symbol determination on the optical signal output from the dispersion compensator 211 and outputs the determination result to the demultiplexer 22. The received symbol decision unit 212 restores the received symbol to digital data. The received symbol decision unit 212 sends the restored data value to the demultiplexer 22 that is paired with the multiplexer 11 in the opposing transmitting device 10. The received symbol decision unit 212 uses the result of machine learning performed in advance when performing symbol determination. The demultiplexer 22 outputs a signal according to the determination result to a path according to the destination.
[0017] The control device 30 controls whether the received symbol decision unit 212 is in a mode in which machine learning is performed (hereinafter referred to as a "training mode") or in a mode in which communication is performed using the results of machine learning (hereinafter referred to as a "data decision mode"). When the control device 30 operates the optical communication system 100 in, for example, the training mode, it outputs a training control signal to the transmission signal selection unit 121 and the received symbol decision unit 212. The training control signal includes information indicating that communication is to be performed in the training mode and information indicating what kind of symbol signal is to be transmitted as the training signal. The control device 30 may be configured as a device different from the transmitting device 10 and the receiving device 20 as shown in FIG. 1, or may be configured as a part of the transmitting device 10 or as a part of the receiving device 20.
[0018] 2 is a diagram showing a detailed configuration example of the transmission signal selection unit 121. The transmission signal selection unit 121 includes a training control circuit 1211 and a selection switch 1212. The training control circuit 1211 receives a training control signal transmitted from the control device 30. The training control circuit 1211 outputs a signal of a symbol indicated by the received training control signal (hereinafter referred to as a "training signal") to the selection switch 1212. Furthermore, when the training control circuit 1211 receives the training control signal, it outputs a selection control signal indicating that the training signal is to be output to the selection switch 1212.
[0019] The selection switch 1212 inputs either one or both of a training signal and a data signal. The selection switch 1212 has a function of switching between the training signal and the data signal and sending it out. The selection switch 1212 sends out either the training signal or the data signal, whichever is indicated by the selection control signal.
[0020] 3 is a diagram showing a detailed example configuration of the received symbol decision unit 212. The received symbol decision unit 212 includes a training control circuit 2121 and a decision unit 2122. The training control circuit 2121 receives a training control signal transmitted from the control device 30. The training control circuit 2121 outputs, to the decision unit 2122, a training control signal including information of a symbol indicated by the received training control signal.
[0021] When the training control signal is received (when in training mode), the determination unit 2122 performs machine learning using the data signal output from the dispersion compensation unit 211 and information indicating the symbol pattern indicated by the training control signal as teacher data. That is, the determination unit 2122 performs learning processing on the premise that the pattern of the data signal output from the dispersion compensation unit 211 is the pattern indicated by the training signal. In this learning processing, the determination unit 2122 learns compensation parameters for nonlinear distortion. When the training control signal is not received (when in data determination mode), the determination unit 2122 performs nonlinear distortion compensation on the data signal output from the dispersion compensation unit 211 based on the result of machine learning and performs symbol determination. The determination unit 2122 outputs a restored data value based on the determination result.
[0022] 4 is a diagram showing a detailed configuration example of the determination unit 2122. In the machine learning training mode, the transmission signal selection unit 121 notifies the receiving device 20 as a training control signal that a training signal is being transmitted and the pattern of the training signal being transmitted. The training control circuit 2121 in the receiving device 20 generates a correct label according to the pattern of the training signal.
[0023] The determination unit 2122 includes a plurality of region selection layers 40, a region determination layer 50, and a multiplexing circuit 60. In the example of FIG. 4, N region selection layers 40 are provided, where N is an integer value of 2 or more. Each region selection layer 40 includes one or more region selection circuits 41. The region selection layer 40_1 in the first layer includes one region selection circuit 41. Each region selection layer 40 in the second layer to the Nth layer includes region selection circuits 41 corresponding to the total number of outputs in the region selection circuit 41 of the region selection layer 40 immediately preceding the first layer. The region determination layer 50 includes a plurality of region determination circuits 51.
[0024] A correct label is input to each region selection circuit 41 and each region determination circuit 51 as a training control signal. The region selection circuit 41 sets the internal variables of the machine learning circuit so as to classify the received symbol according to the correct label. The region selection circuit 41 outputs the received symbol to the corresponding region selection circuit 41 in the next region selection layer 40 according to the correct label. Similarly, in the next region selection layer 40, the region selection circuit 41 sets the internal variables of the machine learning circuit and outputs the received symbol to the corresponding region selection circuit 41 in the next region selection layer.
[0025] The number of layers of the judgment unit 2122 and the number of branches of each area selection circuit 41 are set in advance so that each area contains one type of symbol as a result of judgment by each area judgment circuit 51 in the final layer, that is, area judgment layer 50. By inputting training data signals one after another in this manner and learning all paths, the internal node states of the machine learning circuits provided in all area selection circuits 41 and area judgment circuits 51 are learned and determined.
[0026] In the data judgment mode, the region selection circuit 41 of the first region selection layer 40_1 receives the received symbol and performs region selection according to the machine learning result. The region selection circuit 41 outputs the received symbol to the corresponding region selection circuit 41 of the next region selection layer 40 (40_2). Similarly, in the region selection layer 40_2, the region selection circuit 41 outputs the received symbol to the corresponding region selection circuit 41 of the next region selection layer 40 (40_3) according to the machine learning result. The output of the region judgment layer 50, which is the final layer, determines one type of symbol. The multiplexing circuit 60 receives the judgment result from only one of the region judgment circuits 51 of the region judgment layer 50 via a connection. The number of the connection indicates the region where the last judgment was made. The input value from the connection indicates the final judgment result. The data value represented by the received symbol is uniquely determined based on the number of the connection and the judgment result. The multiplexing circuit 60 outputs the data value as a restored data value.
[0027] Next, the processing in each layer will be further described. The first region selection layer 40_1 includes one region selection circuit 41. In the data judgment mode, the region selection circuit 41 classifies a received symbol into one of L(1,1) regions by using the machine learning result. The region selection circuit 41 outputs the received symbol as it is only to the region selection circuit 41 in the next layer (second layer) that is responsible for the corresponding region. Here, L(m,n) is the number of output connections of the nth region selection circuit constituting the mth layer. If the total number of output connections from the mth layer is LS(m), the following formula 1 is established.
[0028]
number
[0029] The region selection layer 40_2 of the second layer includes L(1,1) region selection circuits 41. Of these multiple region selection circuits 41, only the region selection circuit 41 that receives an input of a received symbol from the first layer operates. For example, when an input is received only in the n-th region selection circuit 41 of the second layer, that region selection circuit 41 classifies the received symbol into one of the L(2,n) regions by using the machine learning result. The region selection circuit 41 outputs the received symbol as it is only to the region selection circuit 41 of the next layer (third layer) that is responsible for the corresponding region.
[0030] Similarly, region selection continues while sending received symbols to the next layer in sequence, and the following equation 2 holds in the Mth layer.
[0031]
number
[0032] The output of the region selection layer 40_M of the Mth layer is input to the region judgment layer 50. The region judgment layer 50 includes LS(M) region judgment circuits 51. Only the region judgment circuit 51 to which a received symbol is input from the region selection layer 40_M of the Mth layer operates. When only the nth region judgment circuit 51 receives an input, the region judgment circuit 51 classifies the received symbol into one of L(x, n) regions by using the machine learning result, and outputs only the judgment result to the downstream multiplexing circuit 60. Here, the symbol representing the region judgment layer 50 is "x".
[0033] The multiplexing circuit 60 outputs a restored data value that is uniquely determined from the position (number) of the area determination circuit 51 to which the input is made and its value.
[0034] The region selection circuit 41 and region determination circuit 51 of each layer described above may be configured to classify a received symbol into one of L(m,n) regions by machine learning using a neural network. The region selection circuit 41 and region determination circuit 51 of each layer described above may be configured to classify a received symbol into one of L(m,n) regions by one-versus-other type machine learning using a support vector machine. Note that both m and n are integers equal to or greater than 1.
[0035] 5 is a diagram showing a specific example of the configuration of the area selection circuit 41. The area selection circuit 41 includes a two-way branching circuit 411, an area determination circuit 412, and a 1xns switch 413. The two-way branching circuit 411 branches a training signal or a data signal into two, and outputs one to the area determination circuit 412 and the other to the 1xns switch 413. The area determination circuit 412 is a machine learning circuit. ns=L(m,n).
[0036] First, the operation in the training mode will be described. The area determination circuit 412 identifies whether the input signal is a training signal or a data signal based on the training control signal. If the input signal is a training signal, the correct label is also notified based on the training control signal. The area determination circuit 412 performs learning by machine learning using a neural network (hereinafter referred to as "NN"), SVM, or the like. Also, the area determination circuit 412 determines to which area selection circuit 41 or area determination circuit 51 in the subsequent layer (area selection layer 40 or area determination layer 50) the input signal is to be output according to the training control signal. The area determination circuit 412 outputs a control signal to the 1xns switch 413 according to the determination.
[0037] Next, the operation in the data judgment mode will be described. Based on the learning result of machine learning performed in advance, the region judgment circuit 412 decides to which region selection circuit 41 or region judgment circuit 51 in the subsequent layer (region selection layer 40 or region judgment layer 50) the input signal should be output to. In accordance with the decision, the region judgment circuit 412 outputs a control signal to the 1xns switch 413. The 1xns switch 413 switches the output destination of the received symbol to one of 1 to ns according to the output of the region judgment circuit 412.
[0038] 6 is a diagram showing an example of the configuration of an area determination circuit 51 using SVM. The area determination circuit 51 constitutes part of the area selection circuit 41 (area determination circuit 412), and is also used in the area determination layer 50. A training signal or a data signal is input to the area determination circuit 51. The area determination circuit 51 includes an SVM circuit 511. The area determination circuit 51 may be configured as a single SVM circuit 511 itself. The area determination circuit 51 may be configured as an exhaustive determination circuit of a plurality of SVM circuits 511.
[0039] The training control signal specifies whether the input signal is a training signal or a data signal. If the input signal is a training signal, its correct label is also notified at the same time. In this case, the SVM circuit 511 performs machine learning by SVM using the training signal and the correct label. If the input signal is a data signal, a judgment is made using the results of the machine learning and the judgment result is output. If the area judgment circuit 51 is part of the area selection circuit 41 (area judgment circuit 412), the judgment result is used as a control signal for the 1xns switch 413.
[0040] 7 is a diagram showing an example of the configuration of a region determination circuit 51 using a neural network. The region determination circuit 51 constitutes part of the region selection circuit 41 (region determination circuit 412), and is also used in the region determination layer 50. A training signal or a data signal is input to the region determination circuit 51. The region determination circuit 51 includes an neural network circuit 512. The region determination circuit 51 may be configured as a single neural network circuit 512 itself.
[0041] The training control signal specifies whether the input signal is a training signal or a data signal. If the input signal is a training signal, its correct label is also notified at the same time. In this case, the NN circuit 512 performs machine learning using the NN, using the training signal and the correct label. If the input signal is a data signal, a judgment is made using the results of the machine learning, and the judgment result is output. If the area judgment circuit 51 is part of the area selection circuit 41 (area judgment circuit 412), the judgment result is used as a control signal for the 1xns switch 413.
[0042] Fig. 8 is a diagram showing a configuration example in which 16QAM signals are determined by combining three layers of two-branch area selection circuits 41. Fig. 9 is a diagram showing an outline of processing in a configuration in which 16QAM signals are determined by combining three layers of two-branch area selection circuits 41. In the area selection layer 40_1, as shown in the first layer of Fig. 9, the 16QAM constellation is divided into two areas, left and right, and machine learning training is performed.
[0043] In the region selection layer 40_2, as shown in the second layer of Fig. 9, each of the left and right regions is divided into upper and lower regions, and machine learning training is performed. In the same manner, such region division is repeated until the region is divided into 16 regions. That is, in the specific example of Fig. 8, L(m,n)=2 is set in all the region selection circuits 41, and L(x,n)=2 is set in all the region determination circuits 51. Moreover, the region determination circuit 51 classifies the received symbol into one of the two regions by one-to-one machine learning using a support vector machine.
[0044] In this embodiment, the number of required machine learning circuits (the number of area determination circuits 412 and area determination circuits 51) is 15, which is an increase compared to the four required in the conventional method. However, since the internal variables of each machine learning circuit are reduced, the total number of internal variables is about the same. On the other hand, when training (machine learning processing) is completed and a symbol is determined, for example, in the first layer (area selection layer 40_1), if the received symbol is determined to belong to the right area, the left area is not calculated in the subsequent calculations. Therefore, the amount of calculation and the calculation execution time can be reduced compared to the conventional method. In this example, it is assumed that the area is divided into two parts repeatedly, but it may be divided into multiple parts or more, and the number of divisions may be different for each layer. The number of divisions may be different for different areas in the same layer.
[0045] Figure 10 is a diagram showing the principle of SVM. We want to draw a boundary line that can distinguish between received symbols indicated by circles and received symbols indicated by triangles with as low an error rate as possible. During machine learning training of SVM, margin areas are set inside each of the circles and triangles from the boundary line. Then, only received symbols outside the margin area as seen from each area are used for judgment. Hereinafter, these received symbols are called support vectors. Received symbols inside the margin area are not used for judgment. Therefore, if the number of received symbols outside the margin area is reduced during training, the amount of calculations during data judgment can be reduced. By performing such processing, the matrix coefficients that can be ignored in judgment increase, so the number of parameters decreases and the amount of calculations during data judgment can be reduced. In the case of NN, the amount of calculations during data judgment can be reduced by deliberately deleting internal variables close to 0.
[0046] Figure 11 shows the simulation results. The existing method in the figure is the nonlinear compensation by the conventional technology. Proposed method 1 performs nonlinear compensation by dividing each layer into two parts using the region division procedure explained in Figure 8. Proposed method 2 performs nonlinear compensation by dividing each layer into four parts using a brute force method using SVM. Simulations were performed with three patterns, 16QAM, 64QAM, and 256QAM, and the total number of SVs used in the calculation until the decision of one symbol was completed was counted and the average value was calculated. At the same time, the Q value was also confirmed. The RBF kernel was used as the kernel function of the SVM, and the hyperparameters were set to the optimal value that minimizes the number of SVs for each simulation condition. Nonlinear distortion was simulated by rotating each symbol point of the QAM signal by an average of 10 degrees in proportion to its intensity. A decrease in the number of SVs and an increase in the Q value were observed under all conditions, confirming the effectiveness of this proposal.
[0047] Each circuit in this embodiment may be configured using a processor such as a CPU (Central Processing Unit) and a memory. All or part of each circuit in this embodiment may be realized using hardware such as an ASIC (Application Specific Integrated Circuit), a PLD (Programmable Logic Device), or an FPGA (Field Programmable Gate Array). The above program may be recorded on a computer-readable recording medium. Examples of computer-readable recording media include portable media such as flexible disks, magneto-optical disks, ROMs, CD-ROMs, and semiconductor storage devices (e.g., SSDs: Solid State Drives), and storage devices such as hard disks and semiconductor storage devices built into computer systems. The above program may be transmitted via an electric communication line.
[0048] Although an embodiment of the present invention has been described in detail above with reference to the drawings, the specific configuration is not limited to this embodiment, and designs that do not deviate from the gist of the present invention are also included. [Industrial Applicability]
[0049] The present invention is applicable to optical communications. [Explanation of symbols]
[0050] 100...optical communication system, 10...transmitting device, 11...multiplexing device, 12...optical modulation device, 121...transmitting signal selection section, 122...optical modulation section, 20...receiving device, 21...optical demodulation device, 22...demultiplexing device, 211...dispersion compensation section, 212...receiving symbol judgment section, 30...control device, 1211...training control circuit, 1212...selection switch, 2121...training control circuit, 2122...judgment section, 40...region selection layer, 41...region selection circuit, 50...region judgment layer, 51...region judgment circuit, 60...multiplexing circuit, 411...2-branching circuit, 412...region judgment circuit, 413...1xns switch, 511...SVM circuit, 512...NN circuit
Claims
1. a reception symbol decision unit that performs symbol decision on a multilevel signal for a received optical signal and restores data based on a result of the symbol decision, the received symbol decision unit includes first to M-th region selection layers (M is an integer equal to or greater than 1) and one region decision layer; The m-th region selection layer (m is an integer between 1 and M) includes region selection circuits in the same number as the number of region divisions in the m-1-th region selection layer ("1" when m=1), The m-th layer region selection circuit selects, for the received multi-level signal, a region to which the symbol of the received multi-level signal belongs from among a plurality of regions according to the number of layers (1 to M) based on the result of machine learning, and outputs the selected region to the next layer; The mth layer region selection circuit performs processing according to the region selected by the m-1th layer region selection circuit, the region determination layer includes region determination circuits corresponding to the total number of outputs of the region selection circuits in the Mth region selection layer, The area determination circuit selects an area to which the symbol of the multi-value signal received from the Mth area selection layer belongs based on a result of machine learning, and outputs a determination result indicating the symbol of the multi-value signal according to the selected area.
2. When operating in a training mode, the area selection circuit and the area determination circuit perform machine learning using the received multi-value signal and a correct answer label indicating a symbol of the multi-value signal, thereby learning which area to select from the multiple areas in accordance with the received multi-value signal; 2. The optical transmitting / receiving device according to claim 1, wherein the area selection circuit and the area determination circuit, when operating in a data determination mode, select the area for the received multi-level signal based on a learning result obtained in the training mode.
3. 3. The optical transmitting and receiving device according to claim 1, wherein the machine learning in the area selection circuit and the area determination circuit is machine learning using a neural network, and the area selection circuit and the area determination circuit classify the symbol into one of a plurality of areas.
4. 3. The optical transmission and reception device according to claim 1, wherein the machine learning in the area selection circuit and the area determination circuit is machine learning using a support vector machine, and the area selection circuit and the area determination circuit classify the symbol into one of a plurality of areas by one-to-other type machine learning.
5. a reception symbol decision unit that performs symbol decision on a multilevel signal for a received optical signal and restores data based on a result of the symbol decision, the received symbol decision unit includes first to M-th region selection layers (M is an integer equal to or greater than 1) and one region decision layer; The m-th region selection layer (m is an integer between 1 and M) includes region selection circuits in the same number as the number of region divisions in the m-1-th region selection layer ("1" when m=1), In an optical communication device, the area determination layer includes area determination circuits corresponding to the total number of outputs of the area selection circuits in the Mth area selection layer, A step in which an area selection circuit of an m-th layer selects an area to which a symbol of the received multi-level signal belongs from among a plurality of areas according to the number of layers (1 to M) based on a result of machine learning for the received multi-level signal, and outputs the selected area to a next layer; the region determination circuit selects a region to which the symbol of the multi-level signal received from the region selection layer of the Mth layer belongs based on a result of machine learning, and outputs a determination result indicating the symbol of the multi-level signal according to the selected region, An optical communication method, wherein the mth layer region selection circuit performs processing according to the region selected by the m-1th layer region selection circuit.
6. A computer program for causing a computer to function as the optical transmitting and receiving device according to any one of claims 1 to 4.
Citation Information
Patent Citations
Carrier reproduction circuit
JP1997214578A
Soft judgment system and reception device
JP1998136046A
Optical receiver and optical symbol label identification method in optical receiver
JP2017163485A
Symbol determination device of input signal, method, and program
JP2021034816A
