Communication system, communication method, and program

Abstract

An aspect of the present invention is a communication system comprising a branching unit that branches an optical signal transmitted by a transmitter configured to transmit the optical signal, and a learning unit that updates an estimation model on the basis of one of optical signals branched by the branching unit, the estimation model serving as a mathematical model for estimating a state of the system on the basis of information indicating an object characteristic that is a predetermined characteristic of the optical signal.

Description

Technical Field

[0001] The present invention relates to a communication system, a communication method, and a program.

Background Art

[0002] In recent years, attempts have been made to apply machine learning and deep learning techniques to the monitoring of transmission characteristics and the detection of abnormal occurrences in optical fiber communication for digital modulation signals obtained by the receivers of digital coherent optical transceivers.

[0003] For example, in Non-Patent Document 1, it has been proposed to use a neural network that uses a constellation visualized by two-dimensional plotting of the real axis / imaginary axis of a digital modulation signal obtained by the receiver of a digital coherent optical transceiver as an input parameter. More specifically, a method has been proposed to identify the distortion of the constellation generated by a state change due to bending, pressure, etc. applied on an optical fiber using such a neural network.

[0004] There have also been attempts to utilize a neural network that has learned the shape of the constellation for monitoring transmission characteristics. For example, in Non-Patent Document 2, it has been shown that by having a neural network learn feature amounts calculated from the constellation, it is possible to estimate quantities representing transmission characteristics such as unknown wavelength dispersion and optical signal-to-noise ratio.

[0005] Here, an example of learning and estimation of a neural network aimed at estimating the optical signal-to-noise ratio (OSNR) and polarization mode dispersion (PMD) by the receiver of a digital coherent optical transceiver will be described with reference to FIG. 16.

[0006] In the learning of such a neural network, a constellation for which the OSNR and PMD are known is input to the neural network. In such learning, the process of tuning the weights of the neural network is repeatedly performed based on the output result of the neural network and the OSNR and PMD in the input constellation. As a result, the neural network learns the regularity for each label. The label means a pair of OSNR and PMD.

[0007] When an unknown constellation without a label is input to the neural network that has completed learning, the neural network outputs the most appropriate label based on the past regularity.

[0008] Thus, for the learning of the neural network, it is necessary to prepare data (i.e., constellation) and labels (i.e., a pair of OSNR and PMD).

[0009] FIG. 17 shows a general system for obtaining a data set that is a combination of data and labels. The system of FIG. 17 includes an emulation unit that can arbitrarily change the transmission characteristics according to the parameters to be learned for the optical transmission path between the optical nodes to be measured.

[0010] In the system of FIG. 17, a constellation is obtained from the optical receiver, the neural network is learned in the optical transmission path state learning unit, and estimation is performed using the obtained model.

Prior Art Documents

Non-Patent Documents

[0011]

Non-Patent Document 1

[0012] However, in an actual optical network operation environment, it is difficult to arbitrarily change the transmission characteristics of the transmission line because it affects the service quality. As another method, a method of acquiring data with changed transmission characteristics before the start of data transmission and reception can be considered, but it takes time to acquire the data.

[0013] Furthermore, in any method, only estimation using a learned model can be performed after the start of the service, and learning data cannot be acquired. In this case, for example, even when the laser characteristics change due to aging deterioration, only the model prepared at the start of operation can be used, so there is a possibility that the estimation accuracy may deteriorate.

[0014] Thus, it has been difficult to detect the state of optical fiber communication, such as monitoring transmission characteristics and detecting occurrence of abnormalities in optical fiber communication, after the start of the service, and there has been a case where the burden required for managing optical fiber communication is large. Further, this has been a problem common to communication systems, not limited to optical fiber communication.

[0015] In view of the above circumstances, an object of the present invention is to provide a technique for reducing the burden required for managing a communication system.

Means for Solving the Problems

[0016] One aspect of the present invention is a communication system including a branching unit that branches an optical signal transmitted by a transmitter that transmits an optical signal, and a learning unit that updates an estimation model, which is a mathematical model for estimating the state of its own system based on information indicating a target characteristic that is a predetermined characteristic of the optical signal, based on one of the optical signals after branching by the branching unit.

[0017] One aspect of the present invention is a communication method executed by a communication system including a branching unit that branches an optical signal transmitted by a transmitter that transmits an optical signal, and a learning unit that updates an estimation model, which is a mathematical model for estimating the state of its own system based on a target characteristic that is a predetermined characteristic of the optical signal, based on one of the optical signals after branching by the branching unit, the communication method including a learning step of updating the estimation model based on one of the optical signals after branching by the branching unit.

[0018] One aspect of the present invention is a program for causing a computer to function as the above communication system.

Effects of the Invention

[0019] According to the present invention, it is possible to reduce the burden required for managing a communication system.

Brief Description of the Drawings

[0020]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

Figure 17

Embodiments for Carrying Out the Invention

[0021] (Embodiment) FIG. 1 is a diagram showing an example of the configuration of a communication system 100 according to an embodiment. The communication system 100 is a communication system that achieves both communication and reduction of the burden required for self-device management. The communication system 100 is, for example, a communication system that performs optical fiber communication. The communication system 100 includes a transmitting-side optical node device 1, a learning device 2, and a receiving-side optical node device 3.

[0022] The transmitting-side optical node device 1 transmits an optical signal. The learning device 2 receives the optical signal and updates, by learning, a mathematical model for estimating the state of the communication system 100 based on the received optical signal. The receiving-side optical node device 3 receives the optical signal. Further, the receiving-side optical node device 3 estimates the state of the communication system 100 using the mathematical model obtained by the learning device 2.

[0023] Note that the mathematical model is a set including one or a plurality of processes whose execution conditions and order (hereinafter referred to as “execution rules”) are predetermined. Learning means updating the mathematical model by a machine learning method. Updating the mathematical model means suitably adjusting the values of the parameters in the mathematical model. Further, executing the mathematical model means executing each process included in the mathematical model according to the execution rules.

[0024] The update of the mathematical model by learning is performed until a predetermined end condition for learning (hereinafter referred to as “learning end condition”) is satisfied. The learning end condition is, for example, a condition that learning has been performed a predetermined number of times.

[0025] The transmitting-side optical node device 1 and the learning device 2 are connected by an optical fiber 4, and an optical signal propagates from the transmitting-side optical node device 1 to the learning device 2 through the optical fiber 4. The transmitting-side optical node device 1 and the receiving-side optical node device 3 are connected by an optical fiber 4, and an optical signal propagates from the transmitting-side optical node device 1 to the receiving-side optical node device 3 through the optical fiber 4. The optical fiber 4 is a transmission path (i.e., an optical fiber) through which an optical signal propagates.

[0026] The transmitting - side optical node device 1 includes an optical transmitter 11 and an optical branching unit 12. The optical transmitter 11 transmits an optical signal. The optical branching unit 12 branches the optical signal. The optical branching unit 12 is, for example, an optical coupler.

[0027] The learning device 2 receives the branched optical signal branched by the optical branching unit 12. More specifically, the learning device 2 receives one of the optical signals branched into a plurality by the optical branching unit 12. The learning device 2 includes an emulation unit 21, an optical signal receiving unit 22, and a state learning unit 23.

[0028] <Emulation unit 21> The emulation unit 21 changes the characteristics of the received optical signal. That is, the emulation unit 21 is a functional unit that pseudo - gives to the received optical signal the changes that can occur in the optical signal due to events occurring in the communication system 100. The change in the characteristics of the optical signal by the emulation unit 21 is not the same for all optical signals. The more random - like the change in characteristics is, the more desirable it is. The change in the characteristics of the optical signal by the emulation unit 21 may also be a change that satisfies the conditions instructed by the user.

[0029] In this way, the emulation unit 21 generates a plurality of types of optical signals. The information indicating the characteristics of the optical signals generated by the emulation unit 21 is used for the learning of the mathematical model as described later. Therefore, the learning device 2 can perform learning using the information indicating the characteristics of a plurality of types of optical signals.

[0030] FIG. 2 is a diagram showing an example of the hardware configuration of the emulation unit 21 in the embodiment. The emulation unit 21 includes a control unit 211 including a processor 91 such as a CPU (Central Processing Unit) connected by a bus and a memory 92, and executes a program. The emulation unit 21 functions as a device including a control unit 211, a communication unit 212, a storage unit 213, an optical characteristic change unit 214, and an optical spectrum analyzer 215 by executing the program.

[0031] More specifically, the processor 91 reads out the program stored in the storage unit 213 and stores the read program in the memory 92. By executing the program stored in the memory 92, the emulation unit 21 functions as a device including a control unit 211, a communication unit 212, a storage unit 213, an optical characteristic changing unit 214, and an optical spectrum analyzer 215.

[0032] The control unit 211 controls the operations of various functional units included in the emulation unit 21. The control unit 211 controls, for example, the operation of the optical characteristic changing unit 214. The control unit 211 controls, for example, the operation of the optical spectrum analyzer 215 to obtain the result of the analysis by the optical spectrogram analyzer. The control unit 211 records various information in the storage unit 213, for example.

[0033] The communication unit 212 is configured to include a communication interface for connecting the emulation unit 21 to an external device. The communication unit 212 communicates with the external device via wired or wireless means. The external device is, for example, the state learning unit 23. The communication unit 212 transmits information indicating the content of the change by the optical characteristic changing unit 214 (hereinafter referred to as "change content information") to the state learning unit 23. Further, the communication unit 212 may be communicably connected to a computer operated by a user and receive a user operation on the emulation unit 21.

[0034] The storage unit 213 is configured using a computer-readable storage medium device such as a magnetic hard disk device or a semiconductor storage device. The storage unit 213 stores various information related to the emulation unit 21. The storage unit 213 may store, for example, the change content information.

[0035] An optical signal branched by the optical branching unit 12 is incident on the optical characteristic changing unit 214. That is, the reception of the optical signal by the learning device 2 means the reception of the optical signal by the optical characteristic changing unit 214.

[0036] The optical characteristic changing unit 214 changes the characteristics of the optical signal. The characteristics may be, for example, chromatic dispersion, polarization mode dispersion, or signal-to-noise ratio. In the example of FIG. 2, the optical characteristic changing unit 214 includes a chromatic dispersion generator 216, a polarization mode dispersion generator 217, and an ASE (Amplified Spontaneous Emission) light source 218. The chromatic dispersion generator 216 changes the chromatic dispersion. The polarization mode dispersion generator 217 changes the polarization mode dispersion (PMD). The ASE light source 218 adds noise to the signal.

[0037] Note that the configuration of the optical characteristic changing unit 214 in FIG. 2 is an example, and the optical characteristic changing unit 214 may have any configuration as long as it can change the desired characteristics of the optical signal. Therefore, the optical characteristic changing unit 214 may include only one of, for example, the chromatic dispersion generator 216, the polarization mode dispersion generator 217, or the ASE light source 218, or may include only any two of them. Further, the optical characteristic changing unit 214 may include, for example, a device that changes the characteristics of an optical signal other than the chromatic dispersion generator 216, the polarization mode dispersion generator 217, and the ASE light source 218.

[0038] The emulation unit 21 outputs the optical signal after conversion by the optical characteristic changing unit 214.

[0039] The control unit 211 acquires information indicating how the optical characteristic changing unit 214 has converted the optical signal from the optical characteristic changing unit 214. The information indicating how the optical characteristic changing unit 214 has converted the optical signal is the change content information.

[0040] More specifically, the control unit 211 acquires information indicating how each functional unit included in the optical characteristic changing unit 214, such as the chromatic dispersion generator 216, the polarization mode dispersion generator 217, or the ASE light source 218, has converted the optical signal from each functional unit included in the optical characteristic changing unit 214. The acquired change content information is used for learning the mathematical model in the state learning unit 23, which will be described in detail later.

[0041] The optical spectrum analyzer 215 analyzes the characteristics of the optical signal after being changed by the optical characteristic changing unit 214.

[0042] <Optical signal receiving unit 22> The optical signal receiving unit 22 receives the optical signal output by the emulation unit 21 and performs A / D conversion on the received optical signal. A / D conversion means a process of converting an analog signal into a digital signal. The optical signal receiving unit 22 is, for example, an analog-to-digital converter. The optical signal receiving unit 22 outputs the optical signal after A / D conversion.

[0043] FIG. 3 is a diagram showing an example of the hardware configuration of the optical signal receiving unit 22 in the embodiment. The optical signal receiving unit 22 includes a control unit 221 including a processor 903 such as a CPU and a memory 904 connected by a bus, and executes a program. The optical signal receiving unit 22 functions as a device including a control unit 221, a communication unit 222, a storage unit 223, a signal receiving unit 224, and a signal transmitting unit 225 by executing the program.

[0044] More specifically, the processor 903 reads out the program stored in the storage unit 233 and stores the read program in the memory 904. By executing the program stored in the memory 904 by the processor 903, the optical signal receiving unit 22 functions as a device including a control unit 221, a communication unit 222, a storage unit 223, a signal receiving unit 224, and a signal transmitting unit 225.

[0045] The control unit 221 controls the operations of various functional units included in the optical signal receiving unit 22. The control unit 221 records various information in the storage unit 223, for example. The control unit 221 executes A / D conversion on the optical signal received by the signal receiving unit 224, for example. The control unit 221 causes the optical signal after A / D conversion to be transmitted to the signal transmitting unit 225, for example. Note that the control unit 221 may perform signal processing other than A / D conversion such as sampling.

[0046] The communication unit 222 is configured to include a communication interface for connecting the optical signal receiving unit 22 to an external device. The communication unit 222 communicates with the external device via wired or wireless means. The external device is communicably connected to, for example, a computer operated by a user, and receives the user's operation on the optical signal receiving unit 22.

[0047] The storage unit 223 is configured using a computer-readable storage medium device such as a magnetic hard disk device or a semiconductor storage device. The storage unit 223 stores various information regarding the optical signal receiving unit 22.

[0048] The signal receiving unit 224 receives an optical signal. The signal transmitting unit 225 transmits the optical signal after A / D conversion by the signal receiving unit 224.

[0049] <State learning unit 23> The state learning unit 23 receives the change content information and the optical signal. The state learning unit 23 performs learning of the estimation model using the received change content information and optical signal. The estimation model is a mathematical model that estimates the difference between the optical signal to be estimated and the optical signal in the optical branching unit 12 for the target characteristic. The target characteristic is a predetermined characteristic of the optical signal.

[0050] That is, the estimation model is a mathematical model that estimates how the target characteristic of the optical signal to be estimated has changed from the target characteristic of the optical signal in the optical branching unit 12.

[0051] The target characteristics are, for example, the optical signal-to-noise ratio (OSNR) and polarization mode dispersion. Hereinafter, the communication system 100 will be described by taking the case where the target characteristics are the optical signal-to-noise ratio and polarization mode dispersion as an example.

[0052] The estimation model may be a mathematical model expressed in any form as long as it is a mathematical model that can be updated by learning. The estimation model is expressed by, for example, a neural network. The parameters of the neural network are suitably adjusted based on the value of the objective function (i.e., loss). The parameters of the neural network are the parameters of the mathematical model to be expressed.

[0053] FIG. 4 is a diagram showing an example of the hardware configuration of the state learning unit 23 in an embodiment. The state learning unit 23 includes a control unit 231 including a processor 905 such as a CPU and a memory 906 connected by a bus, and executes a program. The state learning unit 23 functions as a device including a control unit 231, a communication unit 232, a storage unit 233, and a signal reception unit 234 by executing the program.

[0054] More specifically, the processor 905 reads out the program stored in the storage unit 233 and stores the read program in the memory 906. By the processor 905 executing the program stored in the memory 906, the state learning unit 23 functions as a device including a control unit 231, a communication unit 232, a storage unit 233, and a signal reception unit 234.

[0055] The control unit 231 controls the operations of various functional units included in the state learning unit 23. The control unit 231 records various information in the storage unit 233, for example. The control unit 231 executes an estimation model, for example. The control unit 231 performs learning of the estimation model, for example.

[0056] The communication unit 232 is configured to include a communication interface for connecting the state learning unit 23 to an external device. The communication unit 232 communicates with the external device via wired or wireless means. The external device is, for example, the emulation unit 21. The communication unit 232 acquires change content information by communicating with the emulation unit 21. Further, the communication unit 232 may be communicably connected to a computer operated by a user and receive a user operation on the state learning unit 23.

[0057] The storage unit 233 is configured using a computer-readable storage medium device such as a magnetic hard disk device or a semiconductor storage device. The storage unit 233 stores various information regarding the state learning unit 23. The storage unit 233 stores, for example, change content information. The storage unit 233 stores, for example, an estimation model in advance. Note that storing a mathematical model means storing a value or a computer program that represents the mathematical model. The storage unit 233 stores, for example, the value of the parameters after the update of the estimation model. The storage unit 233 stores, for example, a learned estimation model. The learned estimation model means the estimation model at the time when the learning end condition is satisfied.

[0058] The signal reception unit 234 receives an optical signal. The signal reception unit 234 receives, for example, the signal output by the optical signal reception unit 22. Note that the optical signal reception unit 22 may be provided in the state learning unit 23. In such a case, the signal reception unit 234 receives the optical signal output by the emulation unit 21.

[0059] FIG. 5 is a diagram showing an example of the configuration of the control unit 231 included in the state learning unit 23 in the embodiment. The control unit 231 includes a received data processing unit 235 and a mathematical model learning unit 236.

[0060] The received data processing unit 235 acquires the change content information acquired by the communication unit 232 and the optical signal received by the signal reception unit 234. The received data processing unit 235 executes target characteristic acquisition processing on the acquired optical signal. The target characteristic acquisition processing is a signal processing for acquiring information indicating the target characteristic from the optical signal. Hereinafter, the information indicating the target characteristic is referred to as target characteristic information.

[0061] The target characteristic acquisition processing is, for example, a process that executes a Fourier transform on the optical signal and a process that acquires the optical signal-to-noise ratio and polarization mode dispersion based on the spectrogram obtained by the Fourier transform.

[0062] When the state learning unit 23 includes the optical signal receiving unit 22, more specifically, the optical signal receiving unit 22 is included in the received data processing unit 235. In such a case, A / D conversion is also performed in the target characteristic acquisition process.

[0063] The received data processing unit 235 also performs a training data generation process. The training data generation process is a process of generating data including the target characteristic information and the change content information by using the target characteristic information obtained by the target characteristic acquisition process and the change content information. The data generated in this way is used as training data for the learning of the estimation model. Therefore, the training data generation process is a process of generating training data including the target characteristic and the change content information by using the target characteristic information and the change content information.

[0064] The mathematical model learning unit 236 learns the estimation model by using the training data. More specifically, the mathematical model learning unit 236 first executes the estimation model for the target characteristic information included in the training data. Next, the mathematical model learning unit 236 updates the estimation model so as to reduce the difference between the execution result of the estimation model and the change content information included in the training data based on the execution result of the estimation model and the change content information included in the training data.

[0065] The estimation model estimates the difference between the optical signal having the target characteristic indicated by the target characteristic information included in the training data and the optical signal in the optical branching unit 12. Since the emulation unit 21 changes the optical signal in the optical branching unit 12 in the learning device 2, the higher the accuracy of the estimation model, the more the difference estimated by the estimation model is equal to the change given by the emulation unit 21 to the optical signal in the optical branching unit 12.

[0066] In this way, the estimation model is a mathematical model that estimates the difference between the optical signal to be estimated and the optical signal in the optical branching unit 12 based on the target characteristic information indicating the target characteristic of the optical signal to be estimated.

[0067] The mathematical model entertainment unit 236 includes a mathematical model execution unit 237 and an update unit 238. The mathematical model execution unit 237 executes an estimation model on the target characteristic information included in the training data. The mathematical model execution unit 237 estimates the difference between the optical signal having the target characteristic included in the training data and the optical signal in the optical branching unit 12 by executing the estimation model.

[0068] The update unit 238 updates the estimation model so as to reduce the difference between the execution result of the estimation model and the change content information based on the estimation result of the mathematical model execution unit 237 and the change content information included in the training data.

[0069] The update of the estimation model by the learning device 2 is performed at predetermined timings until the learning end condition is satisfied. Hereinafter, the predetermined timing is referred to as the learning timing. The learning timing is, for example, a periodic timing that arrives at a predetermined cycle. That is, the learning device 2 performs a series of processes of updating the estimation model from the start of learning until the learning end condition is satisfied at each learning timing. Therefore, the estimation model updated at least at the second and subsequent learning timings is the learned estimation model in which the learning end condition was satisfied at the immediately preceding learning timing.

[0070] Returning to the description of FIG. 1. The receiving-side optical node device 3 receives the branched optical signal branched by the optical branching unit 12. More specifically, the receiving-side optical node device 3 receives one of the optical signals branched into a plurality by the optical branching unit 12, which is different from the optical signal propagating to the learning device 2. The receiving-side optical node device 3 includes an optical receiver 31 and an estimation unit 32.

[0071] <Optical Receiver 31> FIG. 6 is a diagram showing an example of the hardware configuration of the optical receiver 31 in the embodiment. The optical receiver 31 includes a control unit 311 including a processor 907 such as a CPU connected by a bus and a memory 908, and executes a program. The optical receiver 31 functions as a device including a control unit 311, a communication unit 312, a storage unit 313, a signal reception unit 314, and an optical branching unit 315 by executing a program.

[0072] More specifically, the processor 907 reads out the program stored in the storage unit 313 and stores the read program in the memory 908. By executing the program stored in the memory 908, the optical receiver 31 functions as a device including the control unit 311, the communication unit 312, the storage unit 313, the signal reception unit 314, and the optical branching unit 315.

[0073] The control unit 311 controls the operations of various functional units included in the optical receiver 31. The control unit 311 records various information in the storage unit 313, for example. The control unit 311 executes target characteristic acquisition processing on the optical signal received by the signal reception unit 314, the details of which will be described later. By executing the target characteristic acquisition processing, the control unit 311 acquires information indicating the target characteristics of the optical signal received by the signal reception unit 314. Hereinafter, the target characteristic information indicating the target characteristics of the optical signal received by the optical receiver 31 is referred to as estimated target characteristic information.

[0074] The communication unit 312 is configured to include a communication interface for connecting the optical receiver 31 to an external device. The communication unit 312 communicates with the external device via wired or wireless means. The external device is, for example, the estimation unit 32.

[0075] The communication unit 312 transmits the estimated target characteristic information acquired by the control unit 311 to the estimation unit 32 through communication with the estimation unit 32. Further, the communication unit 312 may be communicably connected to a computer operated by a user and receive a user operation on the optical receiver 31.

[0076] The storage unit 313 is configured using a computer-readable storage medium device such as a magnetic hard disk device or a semiconductor storage device. The storage unit 313 stores various information related to the optical receiver 31. The storage unit 313 stores, for example, the estimated target characteristic information.

[0077] The signal receiving unit 314 receives an optical signal. The signal receiving unit 314 receives the signal output by the optical transmitter 11. The optical signal received by the signal receiving unit 314 is a part of the signal output by the optical transmitter 11 and is an optical signal branched by the optical branching unit 315. The optical branching unit 315 branches the optical signal. The optical branching unit 315 is, for example, an optical coupler. One of the optical signals branched by the optical branching unit 315 enters the signal receiving unit 314, and the other one is emitted from the optical receiver 31 and propagates toward a predetermined propagation destination.

[0078] <Estimation unit 32> FIG. 7 is a diagram showing an example of the hardware configuration of the estimation unit 32 in the embodiment. The estimation unit 32 includes a control unit 321 including a processor 909 such as a CPU connected by a bus and a memory 910, and executes a program. The estimation unit 32 functions as a device including the control unit 321, the communication unit 322, and the storage unit 323 by executing the program.

[0079] More specifically, the processor 909 reads out the program stored in the storage unit 323 and stores the read program in the memory 910. By the processor 909 executing the program stored in the memory 910, the estimation unit 32 functions as a device including the control unit 321, the communication unit 322, and the storage unit 323.

[0080] The control unit 321 controls the operations of various functional units included in the estimation unit 32. The control unit 321 records various information in the storage unit 323, for example. The control unit 321 executes a learned estimation model obtained by the learning device 2, for example.

[0081] The communication unit 322 is configured to include a communication interface for connecting the estimation unit 32 to an external device. The communication unit 322 communicates with the external device via wired or wireless means. The external device is, for example, the state learning unit 23. The communication unit 322 acquires a learned estimation model by communicating with the state learning unit 23. Further, the communication unit 322 may be communicably connected to a computer operated by a user and receive a user operation on the estimation unit 32.

[0082] The storage unit 323 is configured using a computer-readable storage medium device such as a magnetic hard disk drive or a semiconductor memory device. The storage unit 323 stores various information regarding the estimation unit 32. The storage unit 323 stores, for example, a pre-trained estimation model in advance.

[0083] The output unit 324 outputs various information. The output unit 324 includes, for example, a display device such as a CRT (Cathode Ray Tube) display, a liquid crystal display, or an organic EL (Electro-Luminescence) display. The output unit 324 may be configured as an interface that connects these display devices to the estimation unit 32. The output unit 324 outputs, for example, the result of executing the learned estimation model by the control unit 321 by a predetermined output method such as display.

[0084] FIG. 8 is a diagram showing an example of the configuration of the control unit 321 included in the estimation unit 32 in the embodiment. The control unit 321 includes a data acquisition unit 325, a state estimation unit 326, and an output control unit 327. The data acquisition unit 325 acquires the estimation target characteristic information received by the communication unit 322.

[0085] The state estimation unit 326 executes the learned estimation model on the estimation target characteristic information acquired by the data acquisition unit 325. By executing the learned estimation model on the estimation target characteristic information, the state estimation unit 326 acquires information indicating the difference between the optical signal having the target characteristic indicated by the estimation target characteristic information and the optical signal in the optical branching unit 12. Hereinafter, the information indicating the difference between the optical signal having the target characteristic indicated by the estimation target characteristic information and the optical signal in the optical branching unit 12 is referred to as estimation result information.

[0086] The output control unit 327 controls the operation of the output unit 324. The output control unit 327 controls the operation of the output unit 324 to cause the output unit 324 to output the estimation result information.

[0087] The communication unit 322 outputs the estimation result information to a predetermined output destination. The predetermined output destination is, for example, a device (hereinafter referred to as "abnormality determination device") that determines whether one or both of the optical transmitter 11 and the optical fiber 4 are normal based on the estimation result information.

[0088] Specifically, the result of executing the learned estimation model output by the output unit 324 is the estimation result information.

[0089] FIG. 9 is a diagram showing an example of the target characteristic information included in the training data in the embodiment. The target characteristic information is, for example, the constellation data of each of two polarization waves orthogonal to each other as shown in FIG. 9. The two polarization waves orthogonal to each other are, for example, the X polarization wave and the Y polarization wave.

[0090] <Example of processing flow> FIG. 10 is a flowchart showing an example of the processing flow executed by the learning device 2 in the embodiment. The learning device 2 determines whether it is learning timing (step S101). The determination of whether it is learning timing may be performed by any of the functional units provided in the learning device 2. For example, the control unit 211 provided in the emulation unit 21 performs the determination. The determination of whether it is learning timing may also be performed by the control unit 231 provided in the state learning unit 23.

[0091] If it is not learning timing (step S101: NO), the process of step S101 is repeated. On the other hand, if it is learning timing (step S101: YES), the learning device 2 receives the optical signal branched by the optical branching unit 12 (step S102). More specifically, the emulation unit 21 receives the optical signal branched by the optical branching unit 12. Next, the emulation unit 21 changes the characteristics of the received optical signal (step S103). Next, the optical signal receiving unit 22 converts the optical signal output by the emulation unit 21 into a digital signal and outputs it (step S104).

[0092] Next, the received data processing unit 235 generates training data including target characteristic information indicating the target characteristics of the optical signal output by the optical signal receiving unit 22 and change content information that is information indicating the content of the change in step S103 (step S105). Next, the mathematical model execution unit 237 executes the estimation model for the target characteristic information included in the training data obtained in step S105 (step S106).

[0093] By executing the estimation model, the mathematical model execution unit 237 obtains, as an estimation result, the difference between the optical signal having the target characteristics indicated by the target characteristic information included in the training data and the optical signal in the optical branching unit 12. Next, the update unit 238 updates the estimation model so as to reduce the difference between the difference estimated in step S106 and the change content information included in the training data (step S107).

[0094] Next, the update unit 238 determines whether the learning end condition is satisfied (step S108). If the learning end condition is satisfied (step S108: YES), the process ends. The estimation model at the time when the learning end condition is satisfied is the learned estimation model. The learned estimation model is used by the estimation unit 32. After the process ends, the process of step S101 is started again.

[0095] On the other hand, if the learning end condition is not satisfied (step S108: NO), the process returns to the process of step S102.

[0096] FIG. 11 is a flowchart showing an example of the flow of processing executed by the receiving-side optical node device 3 in the embodiment. The optical receiver 31 receives one of the optical signals branched by the optical branching unit 12 and other than the optical signal propagated to the learning device 2 (step S201). More specifically, the signal receiving unit 314 receives one of the optical signals branched by the optical branching unit 12 and other than the optical signal propagated to the learning device 2.

[0097] Next, the control unit 311 included in the optical receiver 31 executes target characteristic acquisition processing on the optical signal acquired in step S201 (step S202). The control unit 311 acquires estimated target characteristic information indicating the target characteristics of the optical signal acquired in step S201 by executing the target characteristic acquisition processing. Next, the estimation unit 32 acquires the estimated target characteristic information acquired in step S202 (step S203). More specifically, the data acquisition unit 325 acquires the estimated target characteristic information acquired in step S202 via the communication unit 322.

[0098] Next, the state estimation unit 326 executes the learned estimation model obtained by the learning device 2 on the estimated target characteristic information (step S204). By step S204, the state estimation unit 326 acquires estimation result information. Next, the output unit 324 displays the estimation result information (step S205).

[0099] FIG. 12 is a first flowchart showing an example of the processing flow executed by the communication system 100 in the embodiment. More specifically, FIG. 12 is a flowchart showing an example of the processing flow executed by the communication system 100 at the learning timing.

[0100] The optical transmitter 11 transmits an optical signal (step S301). Next, the optical signal is branched by the optical branching unit 12 (step S302). One of the optical signals branched by the optical branching unit 12 propagates to the learning device 2, and the other of the branched optical signals propagates to the optical receiver 31. (Step S303). The learning device 2 updates the estimation model using the propagated optical signal until the learning end condition is satisfied (step S304). The learning device 2 transmits the obtained learned estimation model to the estimation unit 32 by the communication unit 232 (step S305). Note that in step S304, for example, the processing from step S102 to step S108 is executed.

[0101] FIG. 13 is a second flowchart showing an example of the flow of processing executed by the communication system 100 in the embodiment. More specifically, FIG. 13 is a flowchart showing an example of the flow of processing executed by the communication system 100 at a timing other than the learning timing.

[0102] The optical transmitter 11 transmits an optical signal (step S401). Next, the optical signal is branched by the optical branching unit 12 (step S402). One of the optical signals branched by the optical branching unit 12 propagates to the learning device 2, and the other of the branched optical signals propagates to the optical receiver 31. (Step S403). The optical signal receiving unit 22 executes target characteristic acquisition processing on the optical signal that has propagated to the optical receiver 31, and acquires estimated target characteristic information (step S404).

[0103] Next, the signal transmission unit 225 transmits the optical signal that has propagated to the optical receiver 31 in step S403 (step S405). Next, the estimation unit 32 acquires the estimated target characteristic information acquired in step S404 (step S406).

[0104] Next, the estimation unit 32 executes the learned estimation model obtained by the learning device 2 on the estimated target characteristic information at the learning timing. By step S407, the state estimation unit 326 acquires the estimation result information. Next, the output unit 324 displays the estimation result information (step S408).

[0105] Note that the processing in step S405 does not necessarily have to be executed next to the processing in step S404, and may be executed at any timing as long as it is after the processing in step S403.

[0106] <Relationships and roles of the learning device 2, the optical branching unit 12, and the estimation model> Not only the communication system 100 but also communication systems are often assumed to be used for a long period of time and deteriorate over time. Therefore, it is desirable for the management of the communication system to estimate the state of the communication system in consideration of the effects of aging deterioration. The state of the communication system is, for example, the state of the optical transmitter 11. For the management of the system incorporating such changes over time, it is desirable to change the detection criteria and the rules for judging the state of the system according to the changes over time of the system. Therefore, the communication system 100 includes an optical branching unit 12 and a learning device 2.

[0107] As described above, the learning device 2 changes the optical signal branched by the optical branching unit 12 and learns how the optical signal changes in the optical branching unit 12 at the learning timing. The change is caused by the emulation unit 21.

[0108] The roles of the optical branching unit 12 and the emulation unit 21 will be described in order. In infrastructure facilities such as communication systems, it is difficult to install and remove devices after the start of the service. Specifically, the start of the service is the start of communication. Therefore, it is desirable that the learning device 2 be provided in the communication system 100 in a state where it does not need to be removed from the communication system 100.

[0109] Therefore, the communication system 100 includes an optical branching unit 12. Since the communication system 100 includes the optical branching unit 12, the optical signal transmitted by the optical transmitter 11 is branched by the optical branching unit 12, and along with the realization of communication, the transmission of the optical signal to the learning device 2 is realized. As a result, in the communication system 100, along with the realization of communication, it is also realized to update the mathematical model according to the change over time of the optical transmitter 11.

[0110] Next, the role of the emulation unit 21 will be described. The change in the optical signal by the emulation unit 21 corresponds to artificially causing a change that can occur in the optical signal due to an event occurring within the communication system 100. Therefore, the accuracy of the estimation by the estimation model is improved by learning using the optical signal after its characteristics have been changed by the emulation unit 21.

[0111] Incidentally, the target characteristics mainly change due to two factors with different time scales. One is the change over time in the state of the communication system 100. Hereinafter, the change over time in the state of the communication system 100 is referred to as the first factor. The other is the randomness included in natural phenomena represented by Heisenberg's uncertainty principle and noise. Hereinafter, the randomness included in natural phenomena represented by Heisenberg's uncertainty principle and noise is referred to as the second factor.

[0112] The change due to the first factor is a change with a longer time scale than the change due to the second factor. In the management of a system such as the communication system 100 assumed to be used for a long period of time, in order to grasp the state of the communication system 100, it is necessary to consider both of these changes.

[0113] Since the learning device 2 receives the optical signal branched from the optical branching unit 12, the estimation model can be updated even when the communication system 100 is in a communication state. If the change of the first factor is ignored, a mathematical model for estimating the state of the communication system using the target characteristics of the input optical signal only needs to be obtained once by machine learning. The mathematical model thus obtained can estimate the state of the communication system with high accuracy if the change of the communication system due to the first factor can be ignored.

[0114] However, when there is a change in the communication system due to the first factor, if learning is not performed according to the change over time of the communication system, the estimation accuracy will also decrease along with the change over time of the communication system.

[0115] As described above, the learning device 2 of the communication system 100 can update the estimation model even when the communication system 100 is in a communication state. Therefore, the learning device 2 can execute multiple times of learning according to the change over time of the communication system, not just once. Therefore, the learning device 2 can obtain a mathematical model for estimating the state of the communication system 100 under the condition that the first factor and the second factor exist.

[0116] More specifically, it will be explained how learning can be performed under the condition that the first factor and the second factor exist. As described above, the emulation unit 21 changes the optical signal branched by the optical branching unit 12. That is, the emulation unit 21 generates an optical signal obtained by changing the optical signal branched by the optical branching unit 12.

[0117] When there is a change over time in the state of the communication system 100, the optical signal branched by the optical branching unit 12 changes in a manner correlated with that change. Therefore, the emulation unit 21 generates an optical signal including information on the change over time in the state of the communication system 100.

[0118] On the other hand, the emulation unit 21 artificially gives a possible change to the optical signal due to an event occurring in the communication system 100 as described above. This artificial change mimics the change caused by the second factor. That is, the emulation unit 21 gives a change in the optical signal due to the randomness included in natural phenomena to the optical signal received by the learning device 2.

[0119] In this way, the emulation unit 21 generates an optical signal including information on the change caused by the first factor and information on the change caused by the second factor. Since the target characteristics of the optical signal generated by the emulation unit 21 are used for learning as part of the training data, the learning device 2 can perform learning under the condition that the first factor and the second factor exist.

[0120] The learning by the learning device 2 is executed for the estimation model. And, as described above, the target characteristics of the optical signal generated by the emulation unit 21 are used for learning as part of the training data. Therefore, although the estimation model is a mathematical model that estimates the difference between the optical signal to be estimated and the optical signal in the optical branching unit 12 for the target characteristics, that difference is information indicating the estimation result of the state of the communication system 100. That is, the estimation model is a mathematical model that estimates the state of the communication system 100 based on the target characteristics of the input optical signal.

[0121] As described above, the learned mathematical model obtained by the learning device 2 is used by the estimation unit 32 to estimate the state of the communication system 100. The learned estimation model is further updated at the learning timing, and a new learned estimation model is generated. Since the estimation unit 32 uses the estimation model obtained by the learning device 2, the learned estimation model used by the estimation unit 32 is updated to the learned estimation model at the time when the learning end condition is satisfied every time the learning end condition is satisfied. Therefore, the communication system 100 enables management of the communication system considering not only the second factor but also the first factor.

[0122] <Example of how to use the estimation result information> Here, an example of how to use the estimation result information will be described. As described above, the estimation result information is information indicating the difference between the optical signal having the target characteristic indicated by the estimation target characteristic information and the optical signal in the optical branching unit 12. And as described above, the difference is information indicating the estimation result of the state of the communication system 100. Therefore, the estimation result information is an example of information indicating the state of the communication system 100.

[0123] The difference is caused by the transmission loss caused by the transmission of the optical fiber 4 and the abnormality of the optical transmitter 11. If there is no abnormality in the optical transmitter 11, the difference is the difference due to the transmission loss. Therefore, if the difference exceeds a predetermined range, there is a possibility that it is an abnormality of the optical transmitter 11. Therefore, if the estimation result information is obtained, it is possible to determine whether or not an abnormality has occurred in the optical transmitter 11 by, for example, the above-described abnormality determination device. Also, by displaying the estimation result information on the output unit 324, the user can also determine whether or not an abnormality has occurred in the optical transmitter 11.

[0124] In addition, since the abnormality of the optical transmitter 11 may occur suddenly, if the difference is acquired in advance at the learning timing, when a sudden change in the difference occurs, for example, by the above-described abnormality determination device or the user, it is possible to determine whether or not an abnormality has occurred in the optical transmitter 11.

[0125] The communication system 100 can obtain the estimation result information and output the obtained estimation result information, thereby reducing the burden required for suppressing the occurrence of communication failures.

[0126] The communication system 100 of the embodiment configured as described above includes an optical branching unit 12 that branches the optical signal transmitted by the optical transmitter 11, and a learning device 2 that updates the estimation model based on the optical signal after branching by the optical branching unit 12. Therefore, the optical signal transmitted from the optical transmitter 11 is branched by the optical branching unit 12 and propagates to the learning device 2 and the optical receiver 31.

[0127] As a result, in the communication system 100, while realizing the transmission of the optical signal from the optical transmitter 11 to the optical receiver 31, the mathematical model (i.e., the estimation model) for estimating the state of the communication system 100 can be updated. Therefore, the communication system 100 can reduce the burden required for the management of its own system.

[0128] (Modification example) Note that the communication system 100 does not necessarily need to include only one optical transmitter 11 and one receiving-side optical node device 3 each. The communication system 100 may include a plurality of both the optical transmitter 11 and the receiving-side optical node device 3. Hereinafter, the communication system 100 including a plurality of both the optical transmitter 11 and the receiving-side optical node device 3 is referred to as the communication system 100a.

[0129] FIG. 14 is a diagram showing an example of the configuration of the communication system 100a in the modification example. Hereinafter, the same components as those included in the communication system 100 will be denoted by the same reference numerals as those in FIG. 1, and the description thereof will be omitted.

[0130] The communication system 100a differs from the communication system 100 in that it includes not just one but a plurality of sets of an optical transmitter 11, an optical branching unit 12, and a receiving-side optical node device 3. Further, the communication system 100a differs from the communication system 100 in that it includes an optical signal selection unit 5. The optical signal selection unit 5 determines which one of the optical signals from any of the plurality of optical branching units 12 included in the communication system 100a is to be transmitted to the learning device 2, and propagates only the determined optical signal to the learning device 2. Therefore, the optical signal selection unit 5 is a functional unit that includes, for example, an optical switch that switches the optical signal to be propagated to the learning device 2 according to a predetermined rule. The predetermined rule is, for example, a rule that switches the optical signal to be propagated to the learning device 2 in a predetermined order at a predetermined period.

[0131] In the communication system 100a, each estimation unit 32 receives the updated estimation model transmitted by the learning device 2. In the communication system 100a, each estimation unit 32 uses the received updated estimation model to estimate the state of the communication system 100a.

[0132] FIG. 15 is a diagram showing an example of the hardware configuration of the optical signal selection unit 5 in a modified example. The optical signal selection unit 5 includes a control unit 510 including a processor 911 such as a CPU connected by a bus and a memory 912, and executes a program. The optical signal selection unit 5 functions as a device including a control unit 510, a communication unit 520, a storage unit 530, and an optical switch 540 by executing the program.

[0133] More specifically, the processor 911 reads out the program stored in the storage unit 530 and stores the read program in the memory 912. By the processor 911 executing the program stored in the memory 912, the optical signal selection unit 5 functions as a device including a control unit 510, a communication unit 520, a storage unit 530, and an optical switch 540.

[0134] The control unit 510 controls the operations of various functional units included in the optical signal selection unit 5. The control unit 510 records various information in the storage unit 530, for example. The control unit 510 executes an optical signal determination process. The optical signal determination process is a process of determining which one of the plurality of optical branching units 12 included in the communication system 100a to transmit an optical signal to the learning device 2 according to a predetermined rule. The control unit 510 controls the operation of the optical switch 540 to transmit only the determined optical signal toward the learning device 2.

[0135] The communication unit 520 is configured to include a communication interface for connecting the optical signal selection unit 5 to an external device. The communication unit 520 is communicably connected to, for example, a computer operated by a user and may receive a user operation on the optical receiver 31.

[0136] The storage unit 530 is configured using a computer-readable storage medium device such as a magnetic hard disk device or a semiconductor storage device. The storage unit 530 stores various information related to the optical signal selection unit 5. The storage unit 530 stores, for example, a history of the results of the optical signal determination process.

[0137] The optical switch 540 receives optical signals transmitted from the plurality of optical branching units 12 included in the communication system 100a. The optical switch 540 transmits, under the control of the control unit 510, one of the received plurality of optical signals that is determined to be transmitted toward the learning device 2 by the optical signal determination process, to the learning device 2. The optical switch 540 may be, for example, a mechanical optical switch or a MEMS (Micro Electro Mechanical Systems) optical switch.

[0138] Note that the optical signal selection unit 5 may be provided in the transmission-side optical node device 1, may be provided in the learning device 2, or may be provided in both the transmission-side optical node device 1 and the learning device 2.

[0139] Note that the optical branching unit 12 does not necessarily have to be provided in the transmission-side optical node device 1, and the transmission-side optical node device 1 and the optical branching unit 12 may be mounted in different casings.

[0140] Note that each of the transmission-side optical node device 1 and the learning device 2 does not necessarily have to be mounted in a different casing, and they may be mounted in the same casing. Note that each of the transmission-side optical node device 1, the learning device 2, and the reception-side optical node device 3 does not necessarily have to be mounted in a different casing, and they may be mounted in the same casing.

[0141] Note that the learning device 2 is an example of a learning unit.

[0142] Note that each of the learning device 2 and the estimation unit 32 may be implemented using a plurality of information processing devices communicably connected via a network. Note that all or part of each function of the learning device 2 and the estimation unit 32 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 program may be recorded on a computer-readable recording medium. A computer-readable recording medium is, for example, a portable medium such as a flexible disk, a magneto-optical disk, a ROM, a CD-ROM, or a storage device such as a hard disk incorporated in a computer system. The program may be transmitted via a telecommunication line.

[0143] As described above, the embodiments of the present invention have been described in detail with reference to the drawings, but the specific configuration is not limited to this embodiment, and designs and the like within the scope not departing from the gist of the present invention are also included.

Explanation of Reference Numerals

[0144] 100, 100a... communication systems, 1... transmitting-side optical node device, 2... learning device, 3... receiving-side optical node device, 4... optical fiber, 5... optical signal selection unit, 11... optical transmitter, 12... optical branching unit, 21... emulation unit, 22... optical signal receiving unit, 23... state learning unit, 31... optical receiver, 32... estimation unit, 211... control unit, 212... communication unit, 213... memory unit, 214... optical characteristic modification unit, 215... optical spectrum analyzer, 216... wavelength dispersion generator, 217... polarization mode dispersion generator, 218... ASE light source, 221... control unit, 222... communication unit, 223... memory unit, 224... signal receiving unit, 225... signal transmitting unit, 231... control unit, 232... communication unit, 233... memory unit, 234... signal receiving unit, 235... received data processing unit, 236... mathematical model learning unit, 237... mathematical model execution unit, 238... update unit, 311... control unit, 312... communication unit, 313... memory unit, 314... signal receiving unit, 315... optical branching unit, 321... control unit, 322... communication unit, 323... memory unit, 324... output unit, 325... data acquisition unit, 326... state estimation unit, 327... output control unit, 510... control unit, 520... communication unit, 530... memory unit, 540... optical switch, 901, 903, 905, 907, 909, 911... processors, 902, 904, 906, 908, 910, 912... memories

Claims

1. A branching unit that branches the optical signal transmitted by a transmitter that transmits an optical signal, and a learning unit that updates an estimation model, which is a mathematical model for estimating the state of its own system based on information indicating a target characteristic that is a predetermined characteristic of the optical signal, a receiving-side optical node device that estimates the state of its own system using the mathematical model, comprising: the learning device receives the optical signal after branching by the branching unit, the update of the estimation model by the learning unit is performed based on information indicating a predetermined characteristic that is the result of a change in the characteristic of the optical signal received by the learning device as the information, and change content information that is information indicating the content of the change, the receiving-side optical node device receives an optical signal different from the optical signal received by the learning device among the optical signals branched by the branching unit, and estimates the state of its own system based on the target characteristic of the received optical signal, a communication system.

2. The learning unit performs a series of processes for updating the estimation model at predetermined timings until a predetermined end condition regarding learning is satisfied from the start of learning. The communication system according to Claim 1.

3. An estimation unit that estimates the state of its own system using the estimation model based on another target characteristic of the optical signal after branching by the branching unit, The communication system according to Claim 2, further comprising:

4. The estimation model used by the estimation unit is updated to the estimation model at the time when the end condition is satisfied each time the end condition is satisfied. The communication system according to Claim 3.

5. The target characteristics are optical signal-to-noise ratio (OSNR: Optical Signal-to-Noise Ratio) and polarization mode dispersion. The communication system according to any one of Claims 1 to 4.

6. A learning device comprising: a branching unit that branches an optical signal transmitted by a transmitter that transmits an optical signal; and a learning unit that updates an estimation model, which is a mathematical model for estimating the state of its own system based on information indicating a target characteristic that is a predetermined characteristic of the optical signal, based on one of the optical signals after branching by the branching unit; and a receiving-side optical node device that estimates the state of its own system using the mathematical model. The learning device receives the optical signal after branching by the branching unit, and the update of the estimation model by the learning unit is performed based on information indicating a predetermined characteristic as a result of a change in a characteristic of the optical signal received by the learning device as the information, and change content information that is information indicating the content of the change. The receiving-side optical node device receives an optical signal different from the optical signal received by the learning device among the optical signals branched by the branching unit, and estimates the state of its own system based on the target characteristic of the received optical signal. A communication method executed by a communication system, a learning step in which the learning unit updates the estimation model based on one of the optical signals after branching by the branching unit, having the communication method.

7. A program for causing a computer to function as the communication system according to any one of Claims 1 to 5.

Images