DETAILED ACTION
Response to Arguments
Applicant's arguments filed on 10/22/2025 have been fully considered but they are not persuasive. The examiner has thoroughly reviewed Applicant’s amendment and arguments but firmly believes that the cited references reasonably and properly meet the claimed limitation as rejected.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1, 4, 7 and 8 are rejected under 35 U.S.C. 103 as being unpatentable over Tanaka et al (Tanaka et al: “Field Demonstration of Real-time Optical Network Diagnosis Using Deep Neural Network and Telemetry”, OFC 2019, paper number Tu2E.5, pages 1-3. Hereinafter Tanaka NPL) in view of Tanaka et al (US 2024/0413903. Hereinafter Tanaka ‘903) and Zhang et al (CN 113536919 A. English machine translation provided).
1). With regard to claim 1, Tanaka NPL discloses a learning apparatus (Figures 1 and 4) that trains a learning model (page 1, deep neural network DNN model) which estimates a state of an optical signal (e.g., a state related to fiber bending etc., “Detecting fiber bending helps to avoid not only degradation in transmission quality but also the possibility of fiber break in near-term.”),
said learning apparatus comprising at least one processor (e.g., processing unit in the Network controller etc.), the at least one processor carrying out:
(i) a process for acquiring a constellation (Figure 1, e.g., the constellation shown in Figure 1 for the X-Pol and Y-Pol) of a known state optical signal (e.g., pages 2-3, known bending state) transmitted through optical fiber (e.g., Figure 4, fiber between transmission side and receiving side), and the constellation being a definition of a signal point constellation that indicates a combination of a phase and an amplitude of an in-phase channel (I channel) and a quadrature channel (Q channel) in a digital quadrature modulation method (it is a basic property of a QAM (Quadrature Amplitude Modulation) signals. As shown in Figure 1, the constellation for the X-Pol and Y-Pol, and the coherent receiver structure and page 1 “The captured data is converted into 2 (X- and Y-polarization) 64 x 64 2D histogram data”, Tanaka uses 16-ary QAM, or 16QAM digital quadrature modulation method. According to the basic/fundamental definition/property of a digital QAM, the signal point constellation indicates a combination of a phase and an amplitude of an in-phase channel (I channel) and a quadrature channel (Q channel) in the digital quadrature modulation method);
(ii) a process for generating a corrected constellation obtained by changing the bending (“Fiber bending was added at transmitter side of the fiber” and “we conducted classification of straight vs. bend status in an off-line estimation, which corresponds to the training step in Fig. 1. Bending diameter assumed were 9 mm, 15 mm, and 20 mm. The estimation result are shown in Fig. 5(a)”); and
(iii) a process for using, as training data (used as training data in “training step”), the constellation, and the known state (e.g., pages 2-3, known bending state), the corrected constellation to train the learning model (page 1, “In the training step, the system first captures data output from the analog-to-digital converter (ADC) in the digital signal processor (DSP) in the receiver and stores it in an in-memory database (Redis). The network controller at remote site then pulls the captured data from the Redis server and trains the DNN. The training is performed not in the transponder, but in the network controller as it has graphic processing units (GPUs); this eases the computing loads of the transponder. The captured data is converted into 2 (X- and Y-polarization) 64 x 64 2D histogram data and input to CNN, which is the most common DNN method for analyzing visual imagery. We adopt the already proven AlexNet [6] CNN implementation and use the Chainer framework [7] for CNN training and actual estimations. AlexNet consists of five convolution layers and three fully connected layers (Fig. 2). The number of sampled data used for each training dataset is about 300-500. After training, the trained CNN model is stored in the transponder”).
But, Tanaka NPL does not expressly state that the known state including a noise ratio, crosstalk, or band narrowing, and the process for generating a corrected constellation including a first corrected constellation obtained by rotating the constellation by 90 degrees on a complex plane, a second corrected constellation obtained by rotating the constellation by 180 degrees on the complex plane, and a third corrected constellation obtained by rotating the constellation by 270 degrees on the complex plane.
First, the constellation, in Figure 1, Tanaka NPL shows X-Pol and Y-Pol constellations; therefore, it is obvious to one skilled in the art that the constellation of a known state optical signal is obtained. Tanaka ‘903, the same person of the Tanaka NPL, discloses “NPL 1 proposes to use a neural network with a constellation of digital modulation signals obtained by a receiver of a digital coherent optical transceiver visualized by a two-dimensional plot of real/imaginary axes as input parameters. More specifically, a method of using such a neural network to identify distortion of a constellation caused by state changes due to bending, pressure, or the like applied to an optical fiber has been proposed” ([0003]. Note: NPL 1 is the Tanaka NPL); and in Figure 16, “In learning of such a neural network, a constellation with known OSNR and PMD is input to the neural network. In such learning, a process of tuning weights of the neural network on the basis of the output result of the neural network and the OSNR and PMD in the input constellation is repeated”, and “In the system of FIG. 17, a constellation is acquired from an optical receiver, learning of the neural network is performed in an optical transmission line state learning unit, and an obtained model is used for estimation”.
Regarding rotating the constellation on a complex plane, Zhang et al discloses a signal modulation identification algorithm based on data enhancement and convolutional neural network, and a data enhancement scheme is implemented (Abstract, and Figure 4 etc.), as shown in Figure 2, “Taking the I/Q representation of the signal as an example, rotating a modulated signal by a certain angle around its origin, one can obtain enhanced signal samples I'/Q' as defined by equation (1) ([0011]), and “Rotating the constellation diagram of the QPSK signal, the effect is shown in fig. 2, and the data set enhanced by the rotation is expanded to 4 times of the original data set”, Figure 2 shows the constellation is rotated by 90 degrees (p/2), 180 degrees (p), and 270 degrees (3p/2), respectively (Figure 2 and [0008]-[0012]). That is, Zhang et al discloses a process for generating a corrected constellation including a first corrected constellation obtained by rotating the constellation by 90 degrees on a complex plane, a second corrected constellation obtained by rotating the constellation by 180 degrees on the complex plane, and a third corrected constellation obtained by rotating the constellation by 270 degrees on the complex plane (Zhang: Figure 2(b), a first corrected constellation obtained by rotating the constellation by 90 degrees (p/2) on the complex plane; Figure 2(c), a second corrected constellation obtained by rotating the constellation by 180 degrees (p) on the complex plane; and Figure 2(d), a third corrected constellation obtained by rotating the constellation by 270 degrees (3p/2) on the complex plane).
By rotating the constellation, data set is expanded (e.g., 4 times) and enhanced, and the generalization capability of the model is improved.
Second, regarding the known state including a noise ratio, crosstalk, or band narrowing, Tanaka ‘903 discloses “using such a neural network to identify distortion of a constellation caused by state changes due to bending, pressure, or the like applied to an optical fiber has been proposed” ([0003]), and “it is possible to estimate quantities representing transmission characteristics such as unknown wavelength dispersion and an optical signal-to-noise ratio” ([0004]), “an example of learning and estimation using a neural network for the purpose of estimating an optical signal-to-noise ratio (OSNR) and polarization mode dispersion (PMD) at a receiver of a digital coherent optical transceiver will be described with reference to FIG. 16” and “In learning of such a neural network, a constellation with known OSNR and PMD is input to the neural network. In such learning, a process of tuning weights of the neural network on the basis of the output result of the neural network and the OSNR and PMD in the input constellation is repeated. As a result, the neural network learns regularity for each label. A label means a set of OSNR and PMD”. Therefore, the combination of Tanaka NPL and Tanaka ‘903 and Zhang et al further discloses wherein the state is a noise ratio.
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to apply the teachings of Tanaka ‘903 and Zhang et al to the system/method of Tanaka NPL so that a known state including a nose ratio etc., can be acquired, and the corrected constellations are obtained by rotating the constellation on a complex plane, and the data set is expanded and the generalization capability of the model is improved, and the learning model can be better trained.
2). With regard to claim 4, Tanaka NPL discloses an optical signal state estimation apparatus (Figure 1 and 4) that estimates a state (e.g., a state related to fiber bending etc., “Detecting fiber bending helps to avoid not only degradation in transmission quality but also the possibility of fiber break in near-term.”) of an optical signal transmitted through optical fiber (e.g., fiber between transmission side and receiving side, Figure 4),
said optical signal state estimation apparatus comprising at least one processor (e.g., processing unit in the “CNN (estimation)” of Figure 1), the at least one processor carrying out:
(a) a process for acquiring a constellation of the optical signal (Figure 1, e.g., the constellation shown in Figure 1 for the X-Pol and Y-Pol, which is also input to demodulation and CNN model), the constellation being a definition of a signal point constellation that indicates a combination of a phase and an amplitude of an in-phase channel (I channel) and a quadrature channel (Q channel) in a digital quadrature modulation method (it is a basic property of a QAM (Quadrature Amplitude Modulation) signals. As shown in Figure 1, the constellation for the X-Pol and Y-Pol, and the coherent receiver structure and page 1 “The captured data is converted into 2 (X- and Y-polarization) 64 x 64 2D histogram data”, Tanaka uses 16-ary QAM, or 16QAM digital quadrature modulation method. According to the basic/fundamental definition/property of a digital QAM, the signal point constellation indicates a combination of a phase and an amplitude of an in-phase channel (I channel) and a quadrature channel (Q channel) in the digital quadrature modulation method); and
(b) a process for using a learned model (“model”) to estimate the state of the optical signal from the acquired constellation (pages 1-2, “After training, the trained CNN model is stored in the transponder. In the estimation step, the CNN estimation (also called inference) function in the transponder loads the trained CNN model”), the learned model having been trained with use of, as training data (used as training data in “training step”), a constellation of a known state optical signal (e.g., pages 2-3, known bending state), a corrected constellation obtained by changing the bending (“Fiber bending was added at transmitter side of the fiber” and “we conducted classification of straight vs. bend status in an off-line estimation, which corresponds to the training step in Fig. 1. Bending diameter assumed were 9 mm, 15 mm, and 20 mm. The estimation result are shown in Fig. 5(a)”).
But, Tanaka NPL does not expressly state that the known state including a noise ratio, crosstalk, or band narrowing, and the corrected constellation including a first corrected constellation obtained by rotating the constellation by 90 degrees on a complex plane, a second corrected constellation obtained by rotating the constellation by 180 degrees on the complex plane, and a third corrected constellation obtained by rotating the constellation by 270 degrees on the complex plane.
First, the constellation, in Figure 1, Tanaka NPL shows X-Pol and Y-Pol constellations; therefore, it is obvious to one skilled in the art that the constellation of a known state optical signal is obtained. Tanaka ‘903, the same person of the Tanaka NPL, discloses “NPL 1 proposes to use a neural network with a constellation of digital modulation signals obtained by a receiver of a digital coherent optical transceiver visualized by a two-dimensional plot of real/imaginary axes as input parameters. More specifically, a method of using such a neural network to identify distortion of a constellation caused by state changes due to bending, pressure, or the like applied to an optical fiber has been proposed” ([0003]. Note: NPL 1 is the Tanaka NPL); and in Figure 16, “In learning of such a neural network, a constellation with known OSNR and PMD is input to the neural network. In such learning, a process of tuning weights of the neural network on the basis of the output result of the neural network and the OSNR and PMD in the input constellation is repeated”, and “In the system of FIG. 17, a constellation is acquired from an optical receiver, learning of the neural network is performed in an optical transmission line state learning unit, and an obtained model is used for estimation”.
Regarding rotating the constellation on a complex plane, Zhang et al discloses a signal modulation identification algorithm based on data enhancement and convolutional neural network, and a data enhancement scheme is implemented (Abstract, and Figure 4 etc.), as shown in Figure 2, “Taking the I/Q representation of the signal as an example, rotating a modulated signal by a certain angle around its origin, one can obtain enhanced signal samples I'/Q' as defined by equation (1) ([0011]), and “Rotating the constellation diagram of the QPSK signal, the effect is shown in fig. 2, and the data set enhanced by the rotation is expanded to 4 times of the original data set”, Figure 2 shows the constellation is rotated by 90 degrees (p/2), 180 degrees (p), and 270 degrees (3p/2), respectively (Figure 2 and [0008]-[0012]). That is, Zhang et al discloses the corrected constellation including a first corrected constellation obtained by rotating the constellation by 90 degrees on a complex plane, a second corrected constellation obtained by rotating the constellation by 180 degrees on the complex plane, and a third corrected constellation obtained by rotating the constellation by 270 degrees on the complex plane (Zhang: Figure 2(b), a first corrected constellation obtained by rotating the constellation by 90 degrees (p/2) on the complex plane; Figure 2(c), a second corrected constellation obtained by rotating the constellation by 180 degrees (p) on the complex plane; and Figure 2(d), a third corrected constellation obtained by rotating the constellation by 270 degrees (3p/2) on the complex plane).
By rotating the constellation, data set is expanded (e.g., 4 times) and enhanced, and the generalization capability of the model is improved.
Second, regarding the known state including a noise ratio, crosstalk, or band narrowing, Tanaka ‘903 discloses “using such a neural network to identify distortion of a constellation caused by state changes due to bending, pressure, or the like applied to an optical fiber has been proposed” ([0003]), and “it is possible to estimate quantities representing transmission characteristics such as unknown wavelength dispersion and an optical signal-to-noise ratio” ([0004]), “an example of learning and estimation using a neural network for the purpose of estimating an optical signal-to-noise ratio (OSNR) and polarization mode dispersion (PMD) at a receiver of a digital coherent optical transceiver will be described with reference to FIG. 16” and “In learning of such a neural network, a constellation with known OSNR and PMD is input to the neural network. In such learning, a process of tuning weights of the neural network on the basis of the output result of the neural network and the OSNR and PMD in the input constellation is repeated. As a result, the neural network learns regularity for each label. A label means a set of OSNR and PMD”. Therefore, the combination of Tanaka NPL and Tanaka ‘903 and Zhang et al further discloses wherein the state is a noise ratio.
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to apply the teachings of Tanaka ‘903 and Zhang et al to the system/method of Tanaka NPL so that a known state including a nose ratio etc., can be acquired, and the corrected constellations are obtained by rotating the constellation on a complex plane, and the data set is expanded and the generalization capability of the model is improved, and the learning model can be better trained.
3). With regard to claim 7, Tanaka NPL and Tanaka ‘903 and Zhang et al discloses all of the subject matter as applied to claim 4 above. And the combination of Tanaka NPL and Tanaka ‘903 and Zhang et al further discloses an optical signal multiplexing apparatus (Tanaka NPL: Figure 1, a coherent receiver) comprising an optical signal state estimation apparatus (CNN (estimation) with the trained “Model”) according to claim 4.
4). With regard to claim 8, Tanaka NPL discloses a learning method for training a learning model (page 1, deep neural network DNN model) that estimates a state of an optical signal (e.g., a state related to fiber bending etc., “Detecting fiber bending helps to avoid not only degradation in transmission quality but also the possibility of fiber break in near-term.”),
said learning method comprising:
acquiring a constellation (Figure 1, e.g., the constellation shown in Figure 1 for the X-Pol and Y-Pol) of a known state optical signal (e.g., pages 2-3, known bending state) transmitted through optical fiber (e.g., Figure 4, fiber between transmission side and receiving side), and the constellation being a definition of a signal point constellation that indicates a combination of a phase and an amplitude of an in-phase channel (I channel) and a quadrature channel (Q channel) in a digital quadrature modulation method (it is a basic property of a QAM (Quadrature Amplitude Modulation) signals. As shown in Figure 1, the constellation for the X-Pol and Y-Pol, and the coherent receiver structure and page 1 “The captured data is converted into 2 (X- and Y-polarization) 64 x 64 2D histogram data”, Tanaka uses 16-ary QAM, or 16QAM digital quadrature modulation method. According to the basic/fundamental definition/property of a digital QAM, the signal point constellation indicates a combination of a phase and an amplitude of an in-phase channel (I channel) and a quadrature channel (Q channel) in the digital quadrature modulation method);
generating a corrected constellation obtained by changing the bending (“Fiber bending was added at transmitter side of the fiber” and “we conducted classification of straight vs. bend status in an off-line estimation, which corresponds to the training step in Fig. 1. Bending diameter assumed were 9 mm, 15 mm, and 20 mm. The estimation result are shown in Fig. 5(a)”); and
using, as training data (used as training data in “training step”), the constellation and the corrected constellation, and the known state e.g., pages 2-3, known bending state), to train the learning model (page 1, “In the training step, the system first captures data output from the analog-to-digital converter (ADC) in the digital signal processor (DSP) in the receiver and stores it in an in-memory database (Redis). The network controller at remote site then pulls the captured data from the Redis server and trains the DNN. The training is performed not in the transponder, but in the network controller as it has graphic processing units (GPUs); this eases the computing loads of the transponder. The captured data is converted into 2 (X- and Y-polarization) 64 x 64 2D histogram data and input to CNN, which is the most common DNN method for analyzing visual imagery. We adopt the already proven AlexNet [6] CNN implementation and use the Chainer framework [7] for CNN training and actual estimations. AlexNet consists of five convolution layers and three fully connected layers (Fig. 2). The number of sampled data used for each training dataset is about 300-500. After training, the trained CNN model is stored in the transponder”).
But, Tanaka NPL does not expressly state: the known state including a noise ratio, crosstalk, or band narrowing, and the corrected constellation including a first corrected constellation obtained by rotating the constellation by 90 degrees on a complex plane, a second corrected constellation obtained by rotating the constellation by 180 degrees on the complex plane, and a third corrected constellation obtained by rotating the constellation by 270 degrees on the complex plane.
First, the constellation, in Figure 1, Tanaka NPL shows X-Pol and Y-Pol constellations; therefore, it is obvious to one skilled in the art that the constellation of a known state optical signal is obtained. Tanaka ‘903, the same person of the Tanaka NPL, discloses “NPL 1 proposes to use a neural network with a constellation of digital modulation signals obtained by a receiver of a digital coherent optical transceiver visualized by a two-dimensional plot of real/imaginary axes as input parameters. More specifically, a method of using such a neural network to identify distortion of a constellation caused by state changes due to bending, pressure, or the like applied to an optical fiber has been proposed” ([0003]. Note: NPL 1 is the Tanaka NPL); and in Figure 16, “In learning of such a neural network, a constellation with known OSNR and PMD is input to the neural network. In such learning, a process of tuning weights of the neural network on the basis of the output result of the neural network and the OSNR and PMD in the input constellation is repeated”, and “In the system of FIG. 17, a constellation is acquired from an optical receiver, learning of the neural network is performed in an optical transmission line state learning unit, and an obtained model is used for estimation”.
Regarding rotating the constellation on a complex plane, Zhang et al discloses a signal modulation identification algorithm based on data enhancement and convolutional neural network, and a data enhancement scheme is implemented (Abstract, and Figure 4 etc.), as shown in Figure 2, “Taking the I/Q representation of the signal as an example, rotating a modulated signal by a certain angle around its origin, one can obtain enhanced signal samples I'/Q' as defined by equation (1) ([0011]), and “Rotating the constellation diagram of the QPSK signal, the effect is shown in fig. 2, and the data set enhanced by the rotation is expanded to 4 times of the original data set”, Figure 2 shows the constellation is rotated by 90 degrees (p/2), 180 degrees (p), and 270 degrees (3p/2), respectively (Figure 2 and [0008]-[0012]). That is, Zhang et al discloses to generate a corrected constellation including a first corrected constellation obtained by rotating the constellation by 90 degrees on a complex plane, a second corrected constellation obtained by rotating the constellation by 180 degrees on the complex plane, and a third corrected constellation obtained by rotating the constellation by 270 degrees on the complex plane (Zhang: Figure 2(b), a first corrected constellation obtained by rotating the constellation by 90 degrees (p/2) on the complex plane; Figure 2(c), a second corrected constellation obtained by rotating the constellation by 180 degrees (p) on the complex plane; and Figure 2(d), a third corrected constellation obtained by rotating the constellation by 270 degrees (3p/2) on the complex plane).
By rotating the constellation, data set is expanded (e.g., 4 times) and enhanced, and the generalization capability of the model is improved.
Second, regarding the known state including a noise ratio, crosstalk, or band narrowing, Tanaka ‘903 discloses “using such a neural network to identify distortion of a constellation caused by state changes due to bending, pressure, or the like applied to an optical fiber has been proposed” ([0003]), and “it is possible to estimate quantities representing transmission characteristics such as unknown wavelength dispersion and an optical signal-to-noise ratio” ([0004]), “an example of learning and estimation using a neural network for the purpose of estimating an optical signal-to-noise ratio (OSNR) and polarization mode dispersion (PMD) at a receiver of a digital coherent optical transceiver will be described with reference to FIG. 16” and “In learning of such a neural network, a constellation with known OSNR and PMD is input to the neural network. In such learning, a process of tuning weights of the neural network on the basis of the output result of the neural network and the OSNR and PMD in the input constellation is repeated. As a result, the neural network learns regularity for each label. A label means a set of OSNR and PMD”. Therefore, the combination of Tanaka NPL and Tanaka ‘903 and Zhang et al further discloses wherein the state is a noise ratio.
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to apply the teachings of Tanaka ‘903 and Zhang et al to the system/method of Tanaka NPL so that a known state including a nose ratio etc., can be acquired, and the corrected constellations are obtained by rotating the constellation on a complex plane, and the data set is expanded and the generalization capability of the model is improved, and the learning model can be better trained.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to LI LIU whose telephone number is (571)270-1084. The examiner can normally be reached 9 am - 8 pm.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Kenneth Vanderpuye can be reached at (571)272-3078. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/LI LIU/Primary Examiner, Art Unit 2634 November 14, 2025