CN114598581A - Training method, recognition method and device for two-stage detection model of probability shaping signal - Google Patents
Training method, recognition method and device for two-stage detection model of probability shaping signal Download PDFInfo
- Publication number
- CN114598581A CN114598581A CN202210077206.9A CN202210077206A CN114598581A CN 114598581 A CN114598581 A CN 114598581A CN 202210077206 A CN202210077206 A CN 202210077206A CN 114598581 A CN114598581 A CN 114598581A
- Authority
- CN
- China
- Prior art keywords
- laser
- signal
- training
- phase noise
- noise ratio
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L27/00—Modulated-carrier systems
- H04L27/0012—Modulated-carrier systems arrangements for identifying the type of modulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B10/00—Transmission systems employing electromagnetic waves other than radio-waves, e.g. infrared, visible or ultraviolet light, or employing corpuscular radiation, e.g. quantum communication
- H04B10/07—Arrangements for monitoring or testing transmission systems; Arrangements for fault measurement of transmission systems
- H04B10/075—Arrangements for monitoring or testing transmission systems; Arrangements for fault measurement of transmission systems using an in-service signal
- H04B10/079—Arrangements for monitoring or testing transmission systems; Arrangements for fault measurement of transmission systems using an in-service signal using measurements of the data signal
- H04B10/0795—Performance monitoring; Measurement of transmission parameters
- H04B10/07953—Monitoring or measuring OSNR, BER or Q
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Theoretical Computer Science (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Electromagnetism (AREA)
- Optical Communication System (AREA)
Abstract
The invention provides a method and a device for training and identifying a two-stage detection model of a probability shaping signal. Meanwhile, because the phase noise of the signal laser and the local oscillator laser in the Lorentz line type is a micro-nano process and the variance of the phase noise increment and the line width of the laser are in a linear relation, the root mean square and the average value of the phase noise increment at different symbol intervals are constructed in a machine learning mode and are mapped to the line width of the laser, so that the line width of the laser is identified, and the generalization capability of the identification is effectively improved.
Description
Technical Field
The invention relates to the technical field of optical communication, in particular to a method and a device for training and identifying a two-stage detection model of a probability shaped signal.
Background
In recent years, new technologies such as cloud equipment and intelligent equipment are rapidly developed, and data traffic is continuously increased. With the development of optical communications, the capacity of single-mode optical fibers has gradually approached the shannon limit, and in order to cope with the rapidly increasing data traffic demand, the resources of the physical layer need to be more effectively utilized. The elastic optical network can adaptively adjust the modulation format and the data rate according to the channel condition and the capacity requirement, and can better allocate physical resources. In order to enable flexible control of optical transmission systems by a flexible optical network, optical performance monitoring and modulation format identification are widely studied.
Parameters of optical signal to noise ratio (OSNR) and laser linewidth are extremely important for the design of transmission systems in elastic optical networks, and monitoring of laser linewidth also helps to diagnose sudden anomalies in lasers. In addition, Modulation Format Identification (MFI) is particularly important for the selection of receiver algorithms in elastic optical networks. With the development of Digital Signal Processing (DSP) technology, optical performance monitoring is shifted from the optical domain to the electrical domain, which further reduces monitoring costs. In order to realize high-precision performance monitoring, machine learning techniques are widely used.
In a coherent optical communication system, the OSNR and the laser linewidth are two key indexes for measuring amplitude noise and phase noise, but the prior art does not have a detection method for a probability shaped wave, and cannot identify the laser linewidth at the same time.
Disclosure of Invention
The embodiment of the invention provides a method, a method and a device for training a two-stage detection model of a probability shaped signal, which are used for eliminating or improving one or more defects in the prior art and solving the problems that the prior art cannot effectively identify a probability shaped wave and cannot identify the line width of a laser.
The technical scheme of the invention is as follows:
in one aspect, the present invention provides a method for training a two-stage detection model of a probability shaped signal, including:
acquiring a training sample set, wherein the training sample set comprises a plurality of samples, and each sample comprises an amplitude histogram which is obtained by processing a DSP module at a receiving end of an elastic optical network through a blind equalization algorithm under a plurality of signal modulation formats; each sample also includes phase noise; marking the optical signal to noise ratio, the modulation format and the laser line width corresponding to each sample as a label; the elastic optical network adopts a Lorentz line type signal laser and a local oscillator laser;
acquiring a first initial model and a second initial model, wherein the first initial model and the second initial model are ANN (Artificial Neural Network) networks; the first initial model comprises an input layer, a shared hidden layer, a first specific hidden layer and a second specific hidden layer, wherein the first specific hidden layer and the second specific hidden layer are respectively connected with the shared hidden layer;
training the first initial model by using the training sample set by taking the amplitude histogram of each sample as input and the optical signal-to-noise ratio and the modulation format as output to obtain an optical signal-to-noise ratio and modulation format identification model;
and training the second initial model by adopting the training sample set by taking a sequence formed by the root-mean-square value and the average value of the phase noise increment of each sample at different symbol intervals as input and the line width of the laser as output to obtain a laser line width identification model.
In some embodiments, the samples in the training sample set are generated under five modulation formats, QPSK, 16QAM, PS-16QAM, 64QAM, and PS-64 QAM.
In some embodiments, the training sample set performs amplitude histogram sampling according to the step length of 50KHz in the range of 50KHz to 500KHz of the laser line width, and simultaneously performs amplitude histogram sampling on each parameter in the range of 10 to 25dB of the optical signal to noise ratio for the elastic optical network adopting the QPSK modulation format; for the elastic optical network adopting 16QAM and PS-16QAM modulation formats, sampling an amplitude histogram of each parameter within the range of 15-30 dB of an optical signal-to-noise ratio; for the elastic optical network adopting 64QAM and PS-64QAM modulation formats, sampling an amplitude histogram of each parameter within the range of 20-35 dB of an optical signal-to-noise ratio; a first set number of samples are respectively collected for each combination of laser linewidth, modulation format and optical signal to noise ratio.
In some embodiments, the amplitude histogram comprises an amplitude histogram of a dual polarization state signal.
In some embodiments, the training sample set performs phase noise acquisition in a range from 50KHz to 500KHz in line width of a laser, with the step length of 50KHz, and simultaneously performs phase noise acquisition on each parameter in a range from 20 dB to 30dB in optical signal to noise ratio for an elastic optical network adopting QPSK, 16QAM, and PS-16QAM modulation formats; for the elastic optical network adopting PS-64QAM and 64QAM modulation formats, phase noise collection is respectively carried out on each parameter within the range of 30-40 dB of optical signal to noise ratio; and respectively collecting a second set number of samples for the combination of the line width, the modulation format and the optical signal to noise ratio of each laser.
In some embodiments, the inputting a sequence formed by the root mean square value and the average value of the phase noise increment of each sample under different symbol intervals comprises:
for a Lorentzian-line-type signal laser and local oscillator laser, the phase noise is a wiener process, expressed as:
wherein w (n) is a sequence obeying a Gaussian distribution, the mean of w (n) is 0, and the variance isThe expression of (a) is:
where Δ v is the sum of the signal and the full width at half maximum of the local oscillator laser, τsIs the time interval of adjacent signals;
wherein abs (·) is an absolute value function, N represents a symbol interval number, and N represents a time-series number;
calculating phase noise increments corresponding to a plurality of set symbol intervals, and acquiring a root mean square value and an average value corresponding to each symbol interval;
and forming a sequence by the root mean square value and the average value of the phase noise increment corresponding to each symbol interval number as the input of the second initial model.
In some embodiments, the plurality of set symbol interval numbers includes 2, 5, and 10 in calculating the phase noise rise corresponding to the plurality of set symbol interval numbers.
In another aspect, the present invention further provides a two-stage detection method for probability shaped signals, including:
acquiring an amplitude to be detected histogram of a signal to be detected, which is obtained by processing through a blind equalization algorithm, inputting the amplitude to be detected histogram into an optical signal-to-noise ratio and modulation format recognition model in the two-stage detection model training method of the probability shaped signal, and outputting an optical signal-to-noise ratio and modulation format recognition result of the signal to be detected;
acquiring the root mean square value and the average value of phase noise increment of a signal to be detected at different symbol intervals, forming a sequence to be detected, inputting the sequence to be detected into a laser line width recognition model in the two-stage detection model training method of the probability shaping signal, and outputting a laser line type recognition result.
In another aspect, the present invention further provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the steps of the method are implemented.
In another aspect, the present invention also provides a computer readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of the above method.
The invention has the beneficial effects that:
in the method and the device for training and identifying the two-stage detection model of the probability shaped signal, the electric domain signal is obtained by processing the signal receiving end through the DSP module, and mapping from the amplitude histogram to the optical signal-to-noise ratio and the modulation format is established in a machine learning mode based on the amplitude histogram processed by the blind equalization algorithm so as to realize identification of the optical signal-to-noise ratio and the modulation format of the probability shaped signal. Meanwhile, because the phase noise of the signal laser and the local oscillator laser in the Lorentz line type is a micro-nano process and the variance of the phase noise increment and the line width of the laser are in a linear relation, the root mean square of the phase noise increment and the average value are constructed in a machine learning mode and mapped to the line width of the laser, so that the line width of the laser is identified.
Furthermore, a sequence is formed by introducing root mean square values and average values of phase noise increments under different symbol intervals, and a mapping relation between the sequence and the laser line width is established in a machine learning mode, so that the generalization capability of identification can be effectively improved.
Additional advantages, objects, and features of the invention will be set forth in part in the description which follows and in part will become apparent to those having ordinary skill in the art upon examination of the following or may be learned from practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.
It will be appreciated by those skilled in the art that the objects and advantages that can be achieved with the present invention are not limited to the specific details set forth above, and that these and other objects that can be achieved with the present invention will be more clearly understood from the detailed description that follows.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the invention and together with the description serve to explain the principles of the invention. In the drawings:
FIG. 1 is a schematic flow chart of a method for training a two-stage detection model of a probability shaped signal according to an embodiment of the present invention;
FIG. 2 is a logic diagram illustrating a flow of a method for training a two-stage detection model for probability-shaped signals according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of a first initial model in a two-stage detection model training method for probability shaped signals according to an embodiment of the present invention;
FIG. 4 is a schematic structural diagram of a second initial model in the method for training a two-stage detection model of a probability shaped signal according to an embodiment of the present invention;
FIG. 5 is a histogram of the amplitude distribution of signals under different modulation with different SNR;
FIG. 6 is a graph of phase noise curves for different laser linewidths;
fig. 7 is a result of estimating the snr of a probability shaped signal according to a two-stage snr and a laser linewidth monitoring scheme in an embodiment of the present invention;
fig. 8 is a result of estimating the laser linewidth by the two-stage osnr of the probability shaped signal and the laser linewidth monitoring scheme according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the following embodiments and accompanying drawings. The exemplary embodiments and descriptions of the present invention are provided to explain the present invention, but not to limit the present invention.
It should be noted that, in order to avoid obscuring the present invention with unnecessary details, only the structures and/or processing steps closely related to the solution according to the present invention are shown in the drawings, and other details not so related to the present invention are omitted.
It should be emphasized that the term "comprises/comprising" when used herein, is taken to specify the presence of stated features, elements, steps or components, but does not preclude the presence or addition of one or more other features, elements, steps or components.
It is also noted herein that the term "coupled," if not specifically stated, may refer herein to not only a direct connection, but also an indirect connection in which an intermediate is present.
In order to efficiently detect the modulation format, the optical signal-to-noise ratio and the laser line width of the probability shaping signal, the optical performance detection is switched from an optical domain to an electrical domain through a digital signal processing module (DSP module), so that the detection cost is reduced, and the performance is improved.
Specifically, the present invention provides a method for training a two-stage detection model of a probability shaped signal, as shown in fig. 1, including steps S101 to S104:
step S101: acquiring a training sample set, wherein the training sample set comprises a plurality of samples, and each sample comprises an amplitude histogram which is obtained by processing a DSP module at a receiving end of an elastic optical network through a blind equalization algorithm under a plurality of signal modulation formats; each sample also includes phase noise; marking the optical signal to noise ratio, the modulation format and the laser line width corresponding to each sample as a label; the elastic optical network adopts a Lorentz line type signal laser and a local oscillator laser;
step S102: acquiring a first initial model and a second initial model, wherein the first initial model and the second initial model are ANN networks; the first initial model comprises an input layer, a shared hidden layer, a first specific hidden layer and a second specific hidden layer, wherein the first specific hidden layer and the second specific hidden layer are respectively connected with the shared hidden layer, the first specific hidden layer is used for recognizing a modulation format and is connected with a first output layer, and the second specific hidden layer is used for recognizing an optical signal-to-noise ratio and is connected with a second output layer.
Step S103: and training the first initial model by using the training sample set by taking the amplitude histogram of each sample as input and the optical signal to noise ratio and the modulation format as output to obtain an optical signal to noise ratio and modulation format identification model.
Step S104: and training the second initial model by adopting a training sample set by taking a sequence formed by the root mean square value and the average value of the phase noise increment of each sample at different symbol intervals as input and the line width of the laser as output to obtain a line width identification model of the laser.
In the embodiment, the optical signal-to-noise ratio, the modulation format and the laser line width are identified by adopting the electric signal obtained by processing through the DSP module in the optical network, and the mapping relation is constructed by introducing machine learning, so that the high-efficiency identification is realized. Specifically, in this embodiment, a two-stage identification model is constructed, and the osnr, the modulation format, and the laser linewidth are identified by constructing two sets of mapping relationships.
In step S101, an optical signal obtained by probability shaping at the transmitting end is transmitted through an optical fiber, amplified by an optical fiber amplifier, introduced with amplifier spontaneous emission noise (ASE noise), received by an integrated coherent receiver at the receiving end, and subjected to digital signal processing. And obtaining an amplitude histogram processed by a blind equalization algorithm as a parameter for identifying the optical signal-to-noise ratio and the modulation format. Specifically, the histogram of the amplitude distribution of the signal after the constant modulus algorithm processing can be adopted. Meanwhile, the phase noise obtained by blind phase search algorithm is adopted as a parameter for identifying the laser line width.
And screening the existing data or constructing an experiment platform to collect sample data and forming a training sample set. Exemplarily, an experimental platform can be set up, and the transmitting end comprises a laser, a double-bias IQ modulator and a plurality of lengths of single-mode optical fibers; at a receiving end, the data optical signals are amplified by an erbium-doped fiber amplifier (EDFA), received by an integrated coherent receiver, processed by an analog-to-digital converter, processed by digital signals, and processed by upsampling, dispersion compensation, IQ equalization and constant modulus algorithm to obtain an amplitude histogram. And further calculating and acquiring phase noise through a blind phase search algorithm.
In the process of constructing the sample data set, in order to improve generalization capability and identification precision, data acquisition is carried out under multiple modulation modes, and in some embodiments, the samples in the training sample set are generated under five modulation formats, namely QPSK, 16QAM, PS-16QAM, 64QAM and PS-64 QAM. It is also possible to provide single mode fibers of various lengths, for example, 80KM, 160KM and 240 KM.
In some embodiments, the training sample set performs amplitude histogram sampling in a range of 50KHz to 500KHz of laser line width according to a step length of 50KHz, and simultaneously performs amplitude histogram sampling on each parameter in a range of 10 to 25dB of optical signal to noise ratio for an elastic optical network adopting a QPSK modulation format; for the elastic optical network adopting 16QAM and PS-16QAM modulation formats, respectively sampling an amplitude histogram of each parameter within the range of 15-30 dB of an optical signal-to-noise ratio; for the elastic optical network adopting 64QAM and PS-64QAM modulation formats, sampling an amplitude histogram of each parameter within the range of 20-35 dB of an optical signal-to-noise ratio; a first set number of samples are respectively collected for each combination of laser linewidth, modulation format and optical signal to noise ratio. Further, the amplitude histogram is an amplitude histogram containing the dual polarization state signal. Thus, there are at least 8000(5 × 16 × 10 × 5 × 2) amplitude histogram samples in the training sample set. Here 8000 samples (5 × 16 × 10 × 5 × 2) in the data set were collected under 5 modulation formats, 16 signal-to-noise ratios, 10 line widths, and 2 permutations of polarization states, and 5 sets of data were collected in each case.
In some embodiments, the training sample set may be further divided into a training set and a test set according to a set proportion, the training set is used for training and updating the model parameters, and the test set is used for testing and optimizing the model parameters. Illustratively, 90% of the amplitude histogram samples are randomly selected as the training set, and another 10% of the amplitude histogram samples are selected as the test set.
Further, for phase noise used for laser line width identification, a training sample set is in the range of 50KHz to 500KHz of laser line width, phase noise collection is carried out according to 50KHz as step length, and meanwhile, for an elastic optical network adopting QPSK, 16QAM and PS-16QAM modulation formats, phase noise collection is carried out on each parameter in the range of 20 dB to 30dB of optical signal to noise ratio; for the elastic optical network adopting PS-64QAM and 64QAM modulation formats, phase noise collection is respectively carried out on each parameter within the range of 30-40 dB of optical signal to noise ratio; and respectively collecting a second set number of samples for the combination of the line width, the modulation format and the optical signal to noise ratio of each laser. Thus, at least 2200(4 × 5 × 11 × 10) phase noises are collected as a data set, and in some embodiments, 70% of the data is randomly selected as a training set, and the remainder as a test set.
And finally, marking the corresponding modulation format, the optical signal-to-noise ratio and the laser line width as a label for each sample.
In step S102, the first initial model and the second initial model are defined as an ANN network, which is a widely parallel interconnected network composed of simple units with adaptability and whose organization can simulate the interaction of the biological nervous system with real-world objects. In this embodiment, the structure of the first initial model based on the multitask learning is specifically defined, and since the first initial model needs to use the amplitude histogram to identify the osnr and the modulation format at the same time, the structure of the first initial model constructs a specific hidden layer and an output layer for two identification tasks after sharing the hidden layer. While the second initial model may contain multiple hidden layers in addition to the input and output layers.
In step S103, for training the first initial model, network updating may be performed based on error back propagation, specifically using a loss function and a random gradient descent method to perform processing.
In step S104, when the signal laser and the local oscillator laser are of the Lorentzian type, phase noise occursThe method is a wiener process, and the wiener process is an independent increment process, and the variance of the phase noise increment and the laser line width are in a linear relation, so that the key for monitoring the laser line width is to monitor the variance of the phase noise increment.
In some embodiments, the inputting a sequence formed by the root mean square value and the average value of the phase noise increment of each sample under different symbol intervals comprises:
for a Lorentz line type signal laser and a local oscillator laser, the phase noise is a wiener process expressed as:
wherein w (n) is a sequence obeying a Gaussian distribution, the mean of w (n) is 0, and the variance isThe expression of (a) is:
where Δ v is the sum of the signal and the full width at half maximum of the local oscillator laser, τsIs the time interval of adjacent signals.
where abs (·) is an absolute value function, N represents the number of symbol intervals, and N represents the time series.
Calculating phase noise increments corresponding to a plurality of set symbol intervals, and acquiring a root mean square value and an average value corresponding to each symbol interval; and forming a sequence by the root mean square value and the average value of the phase noise increment corresponding to each symbol interval number as the input of the second initial model.
Specifically, in this case, the phase noise in the integrated coherent receiver can be estimated from the result of Carrier Phase Recovery (CPR), and can be specifically processed by means of blind phase search, and the phase noise is obtained and then calculated by equation (3)Absolute value of (a).I.e. a superposition of N w (N), w (N) being a sequence obeying a gaussian distribution, w (N) having a mean value of 0,variance ofUnder the action of the absolute value function, the parameters after variance superposition are obtained.
Due to the fact that under the condition of different laser line widths, N values needed when the line widths are accurately predicted are different. In order to enable the scheme to accurately predict the line width under multiple line widths, the embodiment needs to take multiple N values, and the data of the phase noise increment under different symbol intervals is used as the input of the ANN network to improve the robustness of the scheme. After parameter optimization, the value of N is 2, 5, and 10, that is, in calculating the phase noise increment corresponding to a plurality of set symbol interval numbers, the plurality of set symbol interval numbers include 2, 5, and 10. Since the increment is a gaussian distribution with a mean value of zero, only absolute value information needs to be obtained when calculating the increment.
In the training process of the second initial model, network updating can be performed based on error back propagation, and a method of loss function and stochastic gradient descent is specifically used for processing.
On the other hand, the invention also provides a two-stage detection method of the probability shaped signal, which comprises the following steps S201 to S202:
step S201: and acquiring an amplitude to be detected histogram of the signal to be detected, which is obtained by processing through a blind equalization algorithm, inputting the amplitude to be detected histogram into an optical signal-to-noise ratio and modulation format recognition model in the probability shaping signal two-stage detection model training method in the steps S101 to S104, and outputting an optical signal-to-noise ratio and modulation format recognition result of the signal to be detected.
Step S202: and acquiring the root mean square value and the average value of the phase noise increment of the signal to be detected at different symbol intervals, forming a sequence to be detected, inputting the sequence to be detected into a laser line width recognition model in the probability shaping signal two-stage detection model training method in the steps S101-S104, and outputting a laser line type recognition result.
In this embodiment, after a signal to be detected is received by the integrated coherent receiver at the receiving end, the signal to be detected is processed by the DSP, and the histogram of the amplitude to be detected obtained by processing with the blind equalization algorithm is input to the optical signal-to-noise ratio and modulation format identification model obtained by the pre-training in steps S101 to S104, so as to detect the optical signal-to-noise ratio and the modulation format. Further, according to the input parameter format and requirements of the laser line width recognition model pre-trained in the steps S101 to S104.
In another aspect, the present invention further provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the steps of the method are implemented.
In another aspect, the present invention also provides a computer readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of the above method.
The invention is illustrated below with reference to a specific example:
the embodiment provides a scheme for monitoring the double-stage optical signal-to-noise ratio and the laser line width of a probability shaping signal.
The dual-stage osnr and laser linewidth monitoring scheme is shown in fig. 2. In the first stage, the ANN network based on multitask learning is utilized to realize the estimation of the optical signal to noise ratio and the identification of the modulation format. After the first stage is located in the blind equalization algorithm, the histogram of the amplitude of the equalized signal is input into the neural network based on the multitask learning. The structure of the neural network is shown in fig. 3, wherein the shared layer can simultaneously learn the commonality of the two tasks of Modulation Format (MFI) and optical signal to noise ratio (OSNR) estimation, so that the accuracy of the result can be improved and the complexity of the network structure can be reduced. The second stage is located after Carrier Phase Recovery (CPR) and the laser linewidth estimation is achieved using a simple ANN network, the structure of which is shown in fig. 4.
Specifically, in the first stage, a multitask learning ANN network is used to map the amplitude histogram to the SNR and modulation format. In the second stage, an ANN network is used to map the RMS mean and the RMS mean of the phase noise increments to the laser linewidth.
In the training process, a training sample set is first established. In this embodiment, screening and establishment can be performed according to existing data in an existing database, and an experimental platform can also be set up to perform simulation operation and collect data. Illustratively, establishing an experimental platform includes: the method comprises the steps of setting an erbium-doped fiber amplifier (EDFA) at a receiving end, introducing amplifier spontaneous emission noise (ASE noise), acquiring an optical signal through an integrated coherent receiver, performing analog-to-digital conversion, performing digital processing after up-sampling, at least comprising dispersion compensation, IQ equalization and constant modulus algorithm processing, and obtaining an amplitude histogram of a dual-polarization signal after the constant modulus algorithm processing. And then phase noise is obtained through blind phase search.
For the amplitude histogram, the training sample set samples the amplitude histogram in the range of 50KHz to 500KHz of the laser line width according to the step length of 50KHz, and simultaneously, for the elastic optical network adopting QPSK modulation format, samples the amplitude histogram for each parameter in the range of 10 to 25dB of the optical signal-to-noise ratio; for the elastic optical network adopting 16QAM and PS-16QAM modulation formats, sampling an amplitude histogram of each parameter within the range of 15-30 dB of an optical signal-to-noise ratio; for the elastic optical network adopting 64QAM and PS-64QAM modulation formats, sampling an amplitude histogram of each parameter within the range of 20-35 dB of an optical signal-to-noise ratio; a first set number of samples are respectively collected for each combination of laser linewidth, modulation format and optical signal to noise ratio. Further, the amplitude histogram is an amplitude histogram containing the dual polarization state signal. Thus, there are at least 8000(5 × 16 × 10 × 5 × 2) amplitude histogram samples in the training sample set. Randomly select 90% of the amplitude histogram samples as the training set and another 10% of the amplitude histogram samples as the test set. And adding the corresponding optical signal-to-noise ratio and modulation format as labels to the amplitude histogram of each sample.
For phase noise used for laser line width identification, a training sample set is in the range of 50KHz to 500KHz of laser line width, phase noise collection is carried out according to the step length of 50KHz, and meanwhile, for an elastic optical network adopting QPSK, 16QAM and PS-16QAM modulation formats, phase noise collection is carried out on each parameter in the range of 20 dB to 30dB of optical signal-to-noise ratio; for the elastic optical network adopting PS-64QAM and 64QAM modulation formats, phase noise collection is respectively carried out on each parameter within the range of 30-40 dB of optical signal to noise ratio; and respectively collecting a second set number of samples for the combination of the line width, the modulation format and the optical signal to noise ratio of each laser. Thus, at least 2200(4 × 5 × 11 × 10) phase noises are collected as a data set, and in some embodiments, 70% of the data is randomly selected as a training set, and the remainder as a test set. In this embodiment, the adopted laser and the local oscillator laser are of a lorentz line type, so that the phase noise is a wiener process, and with reference to the descriptions of the above formulas 1 to 3, a sequence consisting of a root mean square value and an average value of phase noise increments corresponding to different symbol intervals is calculated and used as an input of the second-stage ANN network, and the corresponding laser line width is used as an output.
And respectively training the ANN in the first stage and the ANN in the second stage by utilizing the training sample set to obtain an optical signal-to-noise ratio and modulation format identification model and a laser line width identification model.
Further, in order to study the performance of the scheme for monitoring the double-stage optical signal-to-noise ratio of the probability shaped signal and the line width of the laser, which is provided by the embodiment, an 80km transmission coherent optical communication system simulation is established, five signals, namely QPSK, 16QAM, Probability Shaping (PS) -16QAM, 64QAM and PS-64QAM, are transmitted, the root mean square value and the mean value of an amplitude distribution histogram and a phase noise increment are collected, and the pre-trained optical signal-to-noise ratio, modulation format identification model and laser line width identification model are input according to corresponding formats.
As shown in fig. 5, the shape of the amplitude histogram of the signal is related to both the modulation format and the signal-to-noise ratio. As shown in fig. 6, the variation of the phase noise is related to the line width size. Simulation results showed that the accuracy of modulation format identification was 100%, and the Root Mean Square Error (RMSE) of OSNR and line width monitoring error was 0.2dB and 18KHz, respectively. Wherein curves of the predicted OSNR and the actual OSNR are shown in fig. 7, and values of the predicted linewidth and the actual linewidth are shown in fig. 8.
In the method and the device for training and identifying the two-stage detection model of the probability shaped signal, the electric domain signal is obtained by processing the signal receiving end through the DSP module, and mapping from the amplitude histogram to the optical signal-to-noise ratio and the modulation format is established in a machine learning mode based on the amplitude histogram processed by the blind equalization algorithm so as to realize identification of the optical signal-to-noise ratio and the modulation format of the probability shaped signal. Meanwhile, because the phase noise of the signal laser and the local oscillator laser in the Lorentz line type is a micro-nano process and the variance of the phase noise increment and the line width of the laser are in a linear relation, the root mean square of the phase noise increment and the average value are constructed in a machine learning mode and mapped to the line width of the laser, so that the line width of the laser is identified.
Furthermore, a sequence is formed by introducing root mean square values and average values of phase noise increments under different symbol intervals, and a mapping relation between the sequence and the laser line width is established in a machine learning mode, so that the generalization capability of identification can be effectively improved.
Those of ordinary skill in the art will appreciate that the various illustrative components, systems, and methods described in connection with the embodiments disclosed herein may be implemented as hardware, software, or combinations of both. Whether this is done in hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention. When implemented in hardware, it may be, for example, an electronic circuit, an Application Specific Integrated Circuit (ASIC), suitable firmware, plug-in, function card, or the like. When implemented in software, the elements of the invention are the programs or code segments used to perform the required tasks. The program or code segments may be stored in a machine-readable medium or transmitted by a data signal carried in a carrier wave over a transmission medium or a communication link. A "machine-readable medium" may include any medium that can store or transfer information. Examples of a machine-readable medium include electronic circuits, semiconductor memory devices, ROM, flash memory, Erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, Radio Frequency (RF) links, and so forth. The code segments may be downloaded via computer networks such as the internet, intranet, etc.
It should also be noted that the exemplary embodiments mentioned in this patent describe some methods or systems based on a series of steps or devices. However, the present invention is not limited to the order of the above-described steps, that is, the steps may be performed in the order mentioned in the embodiments, may be performed in an order different from the order in the embodiments, or may be performed simultaneously.
Features that are described and/or illustrated with respect to one embodiment may be used in the same way or in a similar way in one or more other embodiments and/or in combination with or instead of the features of the other embodiments in the present invention.
The above description is only a preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes may be made to the embodiment of the present invention by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A method for training a two-stage detection model of a probability shaped signal is characterized by comprising the following steps:
acquiring a training sample set, wherein the training sample set comprises a plurality of samples, and each sample comprises an amplitude histogram which is obtained by processing a DSP module at a receiving end of an elastic optical network through a blind equalization algorithm under a plurality of signal modulation formats; each sample also includes phase noise; marking the optical signal to noise ratio, the modulation format and the laser line width corresponding to each sample as a label; the elastic optical network adopts a Lorentz line type signal laser and a local oscillator laser;
acquiring a first initial model and a second initial model, wherein the first initial model and the second initial model are ANN networks; the first initial model comprises an input layer, a shared hidden layer, a first specific hidden layer and a second specific hidden layer, wherein the first specific hidden layer and the second specific hidden layer are respectively connected with the shared hidden layer;
training the first initial model by using the training sample set by taking the amplitude histogram of each sample as input and the optical signal-to-noise ratio and the modulation format as output to obtain an optical signal-to-noise ratio and modulation format identification model;
and training the second initial model by adopting the training sample set by taking a sequence formed by the root-mean-square value and the average value of the phase noise increment of each sample at different symbol intervals as input and the line width of the laser as output to obtain a laser line width identification model.
2. The method of training a two-stage detection model for probability-shaped signals according to claim 1, wherein the samples in the training sample set are generated under five modulation formats of QPSK, 16QAM, PS-16QAM, 64QAM and PS-64 QAM.
3. The method of claim 2, wherein the training sample set is used for sampling an amplitude histogram in a range of 50KHz to 500KHz laser linewidth according to a step length of 50KHz, and simultaneously, for an elastic optical network adopting a QPSK modulation format, the method is used for sampling an amplitude histogram for each parameter in a range of 10 to 25dB optical signal to noise ratio; for the elastic optical network adopting 16QAM and PS-16QAM modulation formats, sampling an amplitude histogram of each parameter within the range of 15-30 dB of an optical signal-to-noise ratio; for the elastic optical network adopting 64QAM and PS-64QAM modulation formats, sampling an amplitude histogram of each parameter within the range of 20-35 dB of an optical signal-to-noise ratio; a first set number of samples are respectively collected for each combination of laser linewidth, modulation format and optical signal to noise ratio.
4. The method of training a two-stage detection model for probability-shaped signals of claim 3, wherein the magnitude histogram comprises a magnitude histogram of dual polarization state signals.
5. The method for training the two-stage detection model of the probability shaped signal according to claim 2, wherein the training sample set is used for collecting the phase noise according to the step length of 50KHz in the range of 50KHz to 500KHz of the laser line width, and simultaneously, for the elastic optical network adopting QPSK, 16QAM and PS-16QAM modulation formats, the phase noise collection is respectively carried out on each parameter in the range of 20 dB to 30dB of the optical signal to noise ratio; for the elastic optical network adopting PS-64QAM and 64QAM modulation formats, phase noise collection is respectively carried out on each parameter within the range of 30-40 dB of optical signal to noise ratio; and respectively collecting a second set number of samples for the combination of the line width, the modulation format and the optical signal to noise ratio of each laser.
6. The method of training a two-stage detection model for probability-shaped signals according to claim 1, wherein the method comprises the following steps of taking a sequence formed by a root mean square value and an average value of phase noise increment of each sample at different symbol intervals as input:
for a Lorentzian-line-type signal laser and local oscillator laser, the phase noise is a wiener process, expressed as:
wherein w (n) is a sequence obeying a Gaussian distribution, the mean of w (n) is 0, and the variance is The expression of (a) is:
where Δ v is the full width at half maximum of the signal and local oscillator laserAnd, τsIs the time interval of adjacent signals;
wherein abs (·) is an absolute value function, N represents a symbol interval number, and N represents a time-series number;
calculating phase noise increments corresponding to a plurality of set symbol intervals, and acquiring a root mean square value and an average value corresponding to each symbol interval;
and forming a sequence by the root mean square value and the average value of the phase noise increment corresponding to each symbol interval number as the input of the second initial model.
7. The method of claim 6, wherein the plurality of symbol interval numbers comprises 2, 5 and 10 in calculating the phase noise rise corresponding to the plurality of symbol interval numbers.
8. A method for two-stage detection of a probability shaped signal, comprising:
acquiring an amplitude histogram to be detected of a signal to be detected, which is obtained by processing through a blind equalization algorithm, inputting the amplitude histogram to be detected into an optical signal-to-noise ratio and modulation format recognition model in the two-stage detection model training method for probability shaped signals according to any one of claims 1 to 7, and outputting an optical signal-to-noise ratio and a modulation format recognition result of the signal to be detected;
acquiring a root mean square value and an average value of phase noise increment of a signal to be detected at different symbol intervals, forming a sequence to be detected, inputting the sequence to be detected into a laser line width recognition model in the two-stage detection model training method for the probability shaped signal according to any one of claims 1 to 7, and outputting a laser line type recognition result.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the steps of the method according to any of claims 1 to 8 are implemented when the processor executes the program.
10. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method according to any one of claims 1 to 8.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210077206.9A CN114598581B (en) | 2022-01-24 | 2022-01-24 | Training method, recognition method and device for double-stage detection model of probability shaping signal |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210077206.9A CN114598581B (en) | 2022-01-24 | 2022-01-24 | Training method, recognition method and device for double-stage detection model of probability shaping signal |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114598581A true CN114598581A (en) | 2022-06-07 |
CN114598581B CN114598581B (en) | 2023-08-04 |
Family
ID=81806343
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210077206.9A Active CN114598581B (en) | 2022-01-24 | 2022-01-24 | Training method, recognition method and device for double-stage detection model of probability shaping signal |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114598581B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115208462A (en) * | 2022-07-14 | 2022-10-18 | 上海交通大学 | Optical module control parameter optimization method and system of optical communication system |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2006102572A1 (en) * | 2005-03-22 | 2006-09-28 | Qualcomm Incorporated | Detection of signal modulation format type |
WO2016202923A1 (en) * | 2015-06-19 | 2016-12-22 | Danmarks Tekniske Universitet | Optical signal-to-noise ratio estimator and estimation methodology for optical communication |
CN109450836A (en) * | 2018-12-14 | 2019-03-08 | 北京理工大学 | A kind of light digital communication modulation format recognition methods of state transition probability distribution |
US20190149387A1 (en) * | 2017-10-04 | 2019-05-16 | Infinera Corporation | Distribution matching for probabilistic constellation shaping with an arbitrary input/output alphabet |
CN111935042A (en) * | 2020-07-20 | 2020-11-13 | 武汉邮电科学研究院有限公司 | Probability shaping identification system and method based on machine learning and receiving end |
CN112367284A (en) * | 2020-11-02 | 2021-02-12 | 华南师范大学 | Probability distribution identification method, device, equipment and medium under probability shaping constellation |
CN112543070A (en) * | 2019-09-23 | 2021-03-23 | 上海诺基亚贝尔股份有限公司 | On-line extraction of channel characteristics |
US11012187B1 (en) * | 2020-03-04 | 2021-05-18 | Fujitsu Limited | Error correction in optical networks with probabilistic shaping and symbol rate optimization |
-
2022
- 2022-01-24 CN CN202210077206.9A patent/CN114598581B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2006102572A1 (en) * | 2005-03-22 | 2006-09-28 | Qualcomm Incorporated | Detection of signal modulation format type |
WO2016202923A1 (en) * | 2015-06-19 | 2016-12-22 | Danmarks Tekniske Universitet | Optical signal-to-noise ratio estimator and estimation methodology for optical communication |
US20190149387A1 (en) * | 2017-10-04 | 2019-05-16 | Infinera Corporation | Distribution matching for probabilistic constellation shaping with an arbitrary input/output alphabet |
CN109450836A (en) * | 2018-12-14 | 2019-03-08 | 北京理工大学 | A kind of light digital communication modulation format recognition methods of state transition probability distribution |
CN112543070A (en) * | 2019-09-23 | 2021-03-23 | 上海诺基亚贝尔股份有限公司 | On-line extraction of channel characteristics |
US11012187B1 (en) * | 2020-03-04 | 2021-05-18 | Fujitsu Limited | Error correction in optical networks with probabilistic shaping and symbol rate optimization |
CN111935042A (en) * | 2020-07-20 | 2020-11-13 | 武汉邮电科学研究院有限公司 | Probability shaping identification system and method based on machine learning and receiving end |
CN112367284A (en) * | 2020-11-02 | 2021-02-12 | 华南师范大学 | Probability distribution identification method, device, equipment and medium under probability shaping constellation |
Non-Patent Citations (1)
Title |
---|
刘博;姜蕾;毛雅亚;吴翔宇;***;徐星;赵建业;韩顺;: "概率整形编码调制技术", 南京信息工程大学学报(自然科学版), no. 01 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115208462A (en) * | 2022-07-14 | 2022-10-18 | 上海交通大学 | Optical module control parameter optimization method and system of optical communication system |
Also Published As
Publication number | Publication date |
---|---|
CN114598581B (en) | 2023-08-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109361471B (en) | Optical communication signal modulation format identification method based on amplitude deviation analysis | |
US20210111794A1 (en) | Optical network performance evaluation using a hybrid neural network | |
CN109379132B (en) | Device and method for estimating optical fiber dispersion by low-speed coherent detection and neural network | |
CN113938188B (en) | Construction method and application of optical signal-to-noise ratio monitoring model | |
US10887009B2 (en) | G-OSNR estimation on dynamic PS-QAM channels using hybrid neural networks | |
CN109450836B (en) | Method for identifying modulation format of optical digital communication with state transition probability distribution | |
Zhao et al. | A modulation format identification method based on information entropy analysis of received optical communication signal | |
CN109347776B (en) | Method for identifying modulation format of optical communication signal with differential phase-to-amplitude ratio | |
CN106209222A (en) | A kind of method and device of optical signal-to-noise ratio monitoring | |
CN114598581B (en) | Training method, recognition method and device for double-stage detection model of probability shaping signal | |
Yu et al. | A modified PSO assisted blind modulation format identification scheme for elastic optical networks | |
Lu et al. | Modulation format identification assisted by sparse-fast-Fourier-transform for hitless flexible coherent transceivers | |
CN114553315B (en) | Optical fiber nonlinear equalization method and system based on CNN-biRNN | |
CN113364519B (en) | Optical performance monitoring method and device, electronic equipment and storage medium | |
CN113364527B (en) | Nonlinear damage compensation method suitable for high-speed coherent polarization multiplexing system | |
Kashi et al. | Artificial neural networks for fiber nonlinear noise estimation | |
Zhou et al. | Simultaneous baud rate/modulation format identification and multi-parameter optical performance monitoring using multi-task learning with enhanced picture of Radon transform | |
Wang et al. | Deep learning enabled simultaneous OSNR and CD monitoring for coherent transmission system | |
WO2016050756A1 (en) | A method for producing a quality of transmission estimator for optical transmissions | |
Kashef et al. | Non‐Gaussian statistics of CO‐OFDM signals after non‐linear optical fibre transmission | |
CN113285762B (en) | Modulation format identification method based on relative entropy calculation | |
Al-Shamayleh et al. | Performance Evaluation of Optical Transmission Based on Link Estimation by Using Deep Learning Techniques | |
CN113869227A (en) | Signal modulation mode identification method, device, equipment and readable storage medium | |
Fan et al. | Experimental Validation of CNN vs. FFNN for Time-and Energy-Efficient EVM Estimation in Coherent Optical Systems | |
Cui et al. | Optical fiber channel modeling method using multi-BiLSTM for PM-QPSK systems |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |