Abnormity detection method of optical distribution network and optical distribution network system

Abstract

The invention relates to an optical distribution network technology, and discloses an anomaly detection method and device for an optical distribution network and an optical distribution network system, and the method comprises the steps: monitoring an eye pattern signal in the optical distribution network based on PON equipment; when it is detected that the abnormal optical eye diagram signal exists, the PON device sends the abnormal optical eye diagram signal to a far-end operation module through an optical distribution network; the far-end operation module calls a pre-training model to detect whether the abnormal light eye diagram signal belongs to a seriously degraded signal; wherein the pre-training model detects whether the abnormal light eye diagram signal belongs to a seriously degraded signal or not by detecting whether a key parameter of the abnormal light eye diagram signal exceeds a nominal value range or not; and if yes, sending a corresponding adjustment command to the optical network unit which triggers the abnormal optical eye diagram signal. The invention further discloses a computer readable storage medium. The invention aims to timely discover and process the abnormal condition of the optical distribution network so as to improve the reliability and the stability of an optical communication system.

Description

Technical Field

[0001] This application relates to the field of optical distribution network technology, and in particular to an anomaly detection method for optical distribution networks, an optical distribution network system, and a computer-readable storage medium. Background Technology

[0002] Optical Distribution Network (ODN) is based on PON (Passive Optical Network) technology and enables fiber optic connections between Optical Line Terminals (OLTs) and Optical Network Units (ONUs). ODNs are characterized by a wide variety of components, complex models, and concealed, dispersed components, making their operation and maintenance a complex and tedious task for operators.

[0003] Currently, anomaly detection in PON equipment (including optical line terminals and optical network units) relies on traditional indicators (such as optical power at the transmitting end, line loss, and sensitivity settings at the receiving end). Furthermore, after a fault occurs (e.g., a large-scale loss of optical network units), technical maintenance personnel go to the site, use instruments (such as optical power meters) to locate the problem, and then provide corresponding solutions (e.g., if the problem is determined to be in the downlink direction (i.e., the OLT-ONU direction), and the problem is located at the source OLT, then replacing the OLT's optical modules can be used). This method is not only time-consuming and labor-intensive, but also makes it even more difficult to locate intermittent problems because relevant information cannot be captured in real time.

[0004] The above content is only used to help understand the technical solution of this application and does not represent an admission that the above content is prior art. Summary of the Invention

[0005] The main objective of this application is to provide an anomaly detection method, an optical distribution network system, and a computer-readable storage medium for optical distribution networks, aiming to enable timely detection and handling of anomalies in optical distribution networks, thereby improving the reliability and stability of optical communication systems.

[0006] To achieve the above objectives, this application provides an anomaly detection method for optical distribution networks, comprising the following steps: Eye diagram signals in an optical distribution network are monitored based on PON equipment; wherein, the PON equipment includes each optical line terminal and / or an optical network unit associated with each optical line terminal; When an abnormal optical eye diagram signal is detected, the PON device sends the abnormal optical eye diagram signal to the remote operation module through the optical distribution network; The remote operation module calls the pre-trained model to detect whether the abnormal optical eye diagram signal is a severely degraded signal. The pre-trained model detects whether the abnormal optical eye diagram signal is a severely degraded signal by detecting whether the key parameters of the abnormal optical eye diagram signal exceed the nominal value range. If so, a corresponding adjustment command is sent to the optical network unit that triggered the abnormal optical eye diagram signal.

[0007] To achieve the above objectives, this application also provides an optical distribution network system, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements the steps of the above-described anomaly detection method for the optical distribution network.

[0008] To achieve the above objectives, this application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described anomaly detection method for optical distribution networks.

[0009] The anomaly detection method, optical distribution network system, and computer-readable storage medium provided in this application enable timely detection and handling of anomalies in the optical distribution network through real-time monitoring of PON devices, transmission of abnormal signals, analysis of pre-trained models, and adjustment of optical network units, thereby improving the reliability and stability of the optical communication system.

[0010] This approach enables the immediate detection of problems in the optical distribution network, moving beyond passive detection after a fault occurs. It involves continuous monitoring of the optical eye diagram signal, allowing for immediate analysis and processing upon detecting any anomalies. This real-time capability allows the system to respond promptly to potential problems, preventing further escalation and significantly reducing the impact of faults on the optical distribution network.

[0011] Meanwhile, analyzing abnormal optical eye diagram signals using pre-trained models can quickly and accurately determine whether the signal has severely degraded and the possible causes of the problem. By learning from a large amount of data, the pre-trained model has grasped the influence of various factors on optical eye diagram signals and can comprehensively consider multiple key parameters to provide more accurate analysis results. This precise analysis helps to quickly locate problems, reduces unnecessary troubleshooting steps, and improves problem-solving efficiency.

[0012] In summary, by monitoring eye diagrams in real time and using pre-defined models, degradation prediction of optical distribution networks was achieved, significantly improving the efficiency of anomaly detection in optical distribution networks and effectively helping operators to respond quickly to network degradation. Attached Figure Description

[0013] Figure 1This is a schematic diagram of the steps of an anomaly detection method for an optical distribution network in one embodiment of this application; Figure 2 This is a schematic diagram of the internal architecture of an optical distribution network system according to an embodiment of this application.

[0014] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0015] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0016] Furthermore, descriptions using terms such as "first" and "second" in this application are for descriptive purposes only (e.g., to distinguish identical or similar features) and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, technical solutions from different embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If a combination of technical solutions is contradictory or impossible to implement, such a combination should be considered nonexistent and not within the scope of protection claimed in this application.

[0017] Reference Figure 1 In one embodiment, the anomaly detection method for the optical distribution network includes: Step S10: Monitor eye diagram signals in the optical distribution network based on PON equipment; wherein, the PON equipment includes each optical line terminal and / or optical network unit associated with each optical line terminal; Step S20: When an abnormal optical eye diagram signal is detected, the PON device sends the abnormal optical eye diagram signal to the remote operation module through the optical distribution network. Step S30: The remote operation module calls the pre-trained model to detect whether the abnormal optical eye diagram signal is a severely degraded signal; wherein, the pre-trained model detects whether the key parameters of the abnormal optical eye diagram signal exceed the nominal value range, thereby detecting whether the abnormal optical eye diagram signal is a severely degraded signal. Step S40: If yes, send the corresponding adjustment command to the optical network unit that triggered the abnormal optical eye diagram signal.

[0018] In this embodiment, the execution terminal can be an optical distribution network system, or other devices or apparatuses (such as control devices) that control the optical distribution network system.

[0019] As described in step S10, the PON equipment includes each optical line terminal (OLT) and / or each optical line terminal associated with an optical network unit (ONU). The optical line terminal is usually located in the central equipment room and is the core control device of the entire passive optical network; the optical network unit is distributed near the user end and is responsible for connecting with the user equipment.

[0020] Eye diagrams are crucial indicators of signal quality in optical communication. An eye diagram is a pattern resembling an eye, formed by the superposition of multiple bit signals. By observing the shape and opening of the eye diagram, one can intuitively understand signal distortion, noise, and inter-symbol interference. In optical distribution networks, monitoring eye diagrams can promptly detect changes in signal quality, providing a basis for subsequent anomaly detection.

[0021] As described in step S20, the PON device determines whether the eye diagram signal is abnormal based on preset rules and thresholds. These rules and thresholds are determined according to the design standards and actual operating experience of the optical communication system. For example, an eye diagram opening threshold is set; if the actual measured eye diagram opening is less than this threshold, the eye diagram signal is considered abnormal. Alternatively, a jitter threshold is set; when the jitter of the eye diagram exceeds this threshold, it is determined to be an abnormal signal.

[0022] After acquiring the eye diagram signal, the PON device compares the characteristic parameters of the eye diagram signal with preset rules and thresholds in real time. Once one or more characteristic parameters are found to be outside the normal range, the eye diagram signal is immediately determined to be an abnormal optical eye diagram signal.

[0023] The optical distribution network adopts a point-to-multipoint topology, using optical splitters to distribute the optical signal from a main fiber to multiple branch fibers, enabling communication between multiple ONUs and OLTs. When transmitting abnormal optical eye diagram signals, the PON equipment utilizes the existing optical distribution network infrastructure to encode and modulate the abnormal signal as an optical signal, and then transmits it to the remote operation module via optical fiber.

[0024] Once an abnormal optical eye diagram signal is detected, the PON device will use the existing optical distribution network to send the abnormal signal to the remote operation module. The optical distribution network is a point-to-multipoint fiber optic transmission and access technology that can efficiently transmit signals from each PON device to the remote operation module. The remote operation module is typically a central node with data analysis and processing capabilities, responsible for further analysis and processing of the received abnormal signal.

[0025] As described in step S30, the remote operation module is the core analysis and decision-making unit in the entire optical distribution network anomaly detection system. It has powerful data processing and analysis capabilities, and can receive abnormal optical eye diagram signals sent from PON devices, and perform in-depth analysis and judgment on these signals.

[0026] The pre-trained model is trained on a large amount of historical data and is used to determine whether an abnormal eye diagram signal is a severely degraded signal. The model training process uses machine learning or deep learning algorithms, taking key parameters of the eye diagram signal as input and the signal severity as output.

[0027] Optionally, before model training, a large amount of optical eye diagram signal data needs to be collected. This data includes normal optical eye diagram signals and abnormal optical eye diagram signals with varying degrees of degradation. The data source can be monitoring records from the actual optical distribution network during long-term operation, or it can be generated through simulation experiments.

[0028] Optionally, key features can be extracted from the collected optical eye diagram signal data, such as eye height, eye width, eye rise time, eye fall time, jitter, and extinction ratio. These features can reflect the quality and performance of the optical signal and are important criteria for judging whether the signal has been severely degraded.

[0029] Optionally, a suitable machine learning or deep learning model, such as a convolutional neural network or support vector machine, can be selected. The extracted feature data is used as input, and the corresponding label indicating whether the signal is severely degraded is used as output to train the model. By continuously adjusting the model's parameters, the model can accurately learn the relationship between features and the degree of signal degradation.

[0030] Optionally, during model training, the initial nominal value range of the initial input is optimized and adjusted using prior knowledge learned during model training to obtain the final nominal value range.

[0031] The initial nominal value range can be determined based on the design standards, industry specifications, and theoretical performance indicators of the optical communication system. For example, in optical communication, the eye diagram opening has a theoretically optimal range under ideal noise-free and interference-free conditions, depending on factors such as the modulation method and transmission rate used by the system. This range is a fundamental condition for ensuring accurate and error-free transmission of optical signals, and it is derived from system design and theoretical analysis.

[0032] During model training, a large amount of optical eye diagram (OEM) signal data is collected, covering normal and abnormal signals under different environmental conditions and device states. By analyzing this data, the model can learn the influence patterns of various factors on key parameters of the OEM signal. Through learning from a large amount of data, the model can extract the relationships between these influencing factors and changes in key parameters, summarizing these relationships as prior knowledge.

[0033] Finally, based on the initial nominal value range, the nominal value range is dynamically adjusted by combining the prior knowledge learned during model training.

[0034] Different optical distribution networks may exhibit varying signal characteristics and performance due to differences in equipment type, network topology, and operating environment. By utilizing prior knowledge learned through model training, a personalized nominal value range can be generated for each specific optical distribution network.

[0035] The trained model is stored in the remote operation module's storage device. When the remote operation module receives an abnormal optical eye signal, it calls the pre-trained model for analysis. The calling process can be implemented through software programs, where key features of the abnormal optical eye signal are input into the model, and the model performs calculations and judgments based on its learned knowledge.

[0036] After receiving the abnormal optical eye diagram signal, the remote operation module inputs the signal into the pre-trained model. The pre-trained model first processes the signal and extracts key parameters. For example, by analyzing the waveform of the eye diagram, parameters such as eye opening and jitter are calculated.

[0037] The pre-trained model compares the extracted key parameters with a pre-defined nominal value range. This nominal range is a normal parameter range determined according to the design standards and performance requirements of the optical communication system. If a key parameter exceeds this nominal range, the model considers the abnormal optical eye diagram signal to have a potential for degradation.

[0038] Optionally, the model will not determine whether a signal is severely degraded based solely on the deviation of a single key parameter, but will consider the situation of multiple key parameters comprehensively. For example, when multiple key parameters simultaneously exceed their nominal range, or when a key parameter exceeds its nominal range by a significant degree, the model will determine that the abnormal optical eye diagram signal is a severely degraded signal. Simultaneously, the model will also consider the interrelationships between the various key parameters to improve the accuracy of the judgment.

[0039] After the pre-trained model completes the detection, it outputs the detection results to the remote operation module. The detection results are usually presented in a simple labeling form, such as "severely degraded" or "not severely degraded," to facilitate subsequent processing by the remote operation module.

[0040] Optionally, the remote operation module will take different processing measures based on the detection results. If the detection results indicate that the abnormal optical eye diagram signal is a severely degraded signal, the remote operation module will proceed to the next processing step; if the detection results indicate that the signal is not a severely degraded signal, the remote operation module may continue to monitor the signal or take some minor adjustment measures to improve the signal quality.

[0041] As described in step S40, when the pre-trained model determines that the abnormal optical eye diagram signal is a severely degraded signal, corresponding measures will be taken to process it.

[0042] Optionally, adjustment commands can be sent to the optical network unit that triggered the abnormal optical eye diagram signal. These adjustment commands may include adjusting the optical transmit power, adjusting the signal modulation method, recalibrating the clock, etc., to improve the signal quality of the optical network unit and restore it to normal operating condition. In this way, abnormal problems occurring in the optical distribution network can be resolved in a timely manner, ensuring the stable operation of the network.

[0043] In one embodiment, by real-time monitoring of PON devices, transmission of abnormal signals, analysis of pre-trained models, and adjustment of optical network units, timely detection and handling of abnormal situations in the optical distribution network are achieved, thereby improving the reliability and stability of the optical communication system.

[0044] This approach enables the immediate detection of problems in the optical distribution network, moving beyond passive detection after a fault occurs. It involves continuous monitoring of the optical eye diagram signal, allowing for immediate analysis and processing upon detecting any anomalies. This real-time capability allows the system to respond promptly to potential problems, preventing further escalation and significantly reducing the impact of faults on the optical distribution network.

[0045] Meanwhile, analyzing abnormal optical eye diagram signals using pre-trained models can quickly and accurately determine whether the signal has severely degraded and the possible causes of the problem. By learning from a large amount of data, the pre-trained model has grasped the influence of various factors on optical eye diagram signals and can comprehensively consider multiple key parameters to provide more accurate analysis results. This precise analysis helps to quickly locate problems, reduces unnecessary troubleshooting steps, and improves problem-solving efficiency.

[0046] In summary, by monitoring eye diagrams in real time and using pre-defined models, degradation prediction of optical distribution networks was achieved, significantly improving the efficiency of anomaly detection in optical distribution networks and effectively helping operators to respond quickly to network degradation.

[0047] In one embodiment, based on the above embodiments, the step of sending a corresponding adjustment command to the optical network unit that triggers the abnormal optical eye diagram signal includes: The remote operation module uploads the abnormal optical eye diagram signal, which is a severely degraded signal, to the cloud so that the operator of the network management system associated with the cloud can determine whether the relevant optical network unit needs to be adjusted. When the cloud-based network management system receives an adjustment command, it sends the corresponding adjustment command to the relevant optical network unit through the downlink channel of the optical distribution network.

[0048] In this embodiment, when the remote operation module calls the pre-trained model to detect abnormal optical eye diagram signals, it detects the key parameters of the abnormal optical eye diagram signals and finds that these key parameters exceed the nominal value range, thereby determining that the abnormal optical eye diagram signals are severely degraded signals.

[0049] Once a severely degraded signal is identified, the remote operation module will upload the abnormal optical eye diagram signal to the cloud. The abnormal optical eye diagram signal uploaded to the cloud is available for use by the operators of the network management system associated with the cloud. The operators can view detailed information about these abnormal optical eye diagram signals through the interface of the network management system, including the signal waveform, key parameters, and comparison with the nominal value range.

[0050] Upon receiving an abnormal optical eye diagram signal, the network management system operator will conduct a detailed review and analysis. The operator will consider factors such as the overall operational status of the optical distribution network, historical fault records, and current service requirements to determine whether the abnormal optical eye diagram signal will severely impact the normal operation of optical network units and the performance of the entire optical distribution network.

[0051] If the operator believes that the abnormal optical eye diagram signal, although a severely degraded signal, will not currently cause substantial harm to the network, or can be handled through other means (such as subsequent routine maintenance), the operator may choose not to send an adjustment command. Conversely, if the operator determines that the abnormal optical eye diagram signal has seriously affected the normal operation of the optical network unit, or may cause more serious network failures, the operator will issue an adjustment command in the network management system.

[0052] When the cloud-based network management system receives an adjustment command from an operator, the system parses and processes the command. The adjustment command includes the parameters to be adjusted (such as transmit power, wavelength, etc.) and the specific value or range to be adjusted.

[0053] The network management system sends adjustment commands to the relevant optical network units (ONUs) via the downlink channel of the optical distribution network. The downlink channel is the channel in the optical distribution network used to transmit data and control signals from optical line terminals (OLTs) to ONUs. During transmission, the commands are also encrypted to ensure their security and integrity.

[0054] Upon receiving the adjustment command, the relevant optical network unit (ONU) decodes and verifies the command to confirm its legality and validity. Then, the ONU adjusts its relevant parameters according to the command's requirements to improve the quality of the optical eye diagram signal and restore it to normal operating condition.

[0055] In one embodiment, a manual judgment step is introduced, which fully leverages the operator's experience and expertise, avoiding potential errors that might arise from relying solely on pre-trained models for automatic adjustments. Simultaneously, centralized management through the cloud and network management system enables unified monitoring and adjustment of multiple optical network units within the optical distribution network, improving the efficiency and accuracy of network management.

[0056] In one embodiment, based on the above embodiment, when the pre-trained model confirms a severe degradation signal, it also outputs key parameters in the severe degradation signal that exceed the nominal value range, so as to generate corresponding adjustment commands based on the output key parameters.

[0057] In this embodiment, when the pre-trained model detects abnormal optical eye diagram signals, it compares each key parameter in the signal with a pre-set nominal value range. Once it finds that one or more key parameters exceed the nominal value range, the model will not only determine that the signal is a severely degraded signal, but also accurately output these out-of-range key parameters and their specific values.

[0058] Optionally, after receiving key parameters from the pre-trained model that exceed the nominal range, the remote operation module will perform a detailed analysis of these parameters. Based on the analysis results, it will search for the corresponding adjustment strategy in a pre-established mapping relationship (for example, if the output key parameter is an excessively low eye height, and the deviation from the range is significant, then an adjustment strategy to increase the transmit power will be matched). Based on the matched adjustment strategy, a specific adjustment command will be generated. The adjustment command will clearly specify the name of the parameter to be adjusted, the direction of adjustment (increase or decrease), and the specific value or range of adjustment. For example, the generated adjustment command could be "Increase the transmit power of optical network unit X by 2dBm".

[0059] In this way, the remote operation module can output key parameters that exceed the nominal range through the pre-trained model and generate adjustment commands accordingly, which can achieve precise adjustment of optical network units and improve the reliability and performance of the optical distribution network.

[0060] Alternatively, the remote operation module can synchronously upload the judgment results of severely degraded signals output by the pre-trained model and key parameters exceeding the nominal value range to the cloud, allowing operators of the cloud-based network management system to make further judgments. Operators of the network management system can use the data analysis tools provided by the cloud to conduct in-depth analysis of abnormal key parameters. Based on the analysis and visualization results of abnormal key parameters, operators can make more scientific and reasonable decisions. For example, if multiple key parameters of a certain optical network unit are found to be frequently abnormal and significantly exceeding the range, the operator can decide to conduct a more in-depth inspection or adjustment of that optical network unit; if it is found that the abnormality of certain parameters is related to a specific time period or environmental factors, the operator can take corresponding preventive measures.

[0061] By synchronously uploading critical abnormal parameters to the cloud, operators can obtain more comprehensive and detailed optical network operation data, thereby making more accurate decisions and improving the management efficiency and reliability of the optical distribution network.

[0062] In one embodiment, based on the above embodiments, the anomaly detection method for the optical distribution network further includes: When the cloud-based network management system receives the adjustment command, it will also send the adjustment command to the remote operation module; The remote operation module associates adjustment commands with corresponding severe degradation signals and updates the pre-trained model based on the association results.

[0063] In this embodiment, when the operator of the network management system in the cloud confirms that the abnormal optical eye diagram signal is a severely degraded signal, he / she can issue a corresponding adjustment command to the network management system.

[0064] When the cloud-based network management system receives the adjustment command, in addition to sending the corresponding adjustment command to the relevant optical network unit through the downlink channel of the optical distribution network, it will also send the command to the remote operation module through a specific communication link.

[0065] After receiving an adjustment command, the remote operation module associates it with previously detected severe degradation signals. Specifically, during anomaly detection, the pre-trained model analyzes anomalous optical eye diagram signals to identify which signals constitute severe degradation. When the cloud sends adjustment commands based on these signals, the remote operation module establishes a mapping between the adjustment commands and the corresponding severe degradation signals. For example, for a severe degradation signal with a specific characteristic, the corresponding adjustment command might be to increase the transmit power of the optical network unit. Through this association, the system can record effective adjustment measures for different types of severe degradation signals.

[0066] The remote operation module uses the aforementioned correlation results to update the pre-trained model. The pre-trained model is typically built based on machine learning or deep learning algorithms and contains a large number of parameters and rules. During the update process, new correlation information is incorporated into the model's training data, readjusting the model's parameters to enable it to better identify severely degraded signals and learn effective adjustment strategies for these signals. For example, in a deep learning model, backpropagation can be used to update the weights of the neural network to improve the model's accuracy in classifying severely degraded signals and its ability to generate adjustment commands.

[0067] As the model is continuously correlated and updated, its ability to identify severely degraded signals gradually improves. The model can learn more about the characteristics and patterns of severely degraded signals from new correlated information, thus enabling it to more accurately determine whether a signal is severely degraded in subsequent detections. For example, the original model might only be able to determine whether a signal is severely degraded based on some key parameters; after updates, it can integrate more signal features for judgment, improving the accuracy of identification.

[0068] When the number of such iterative updates is sufficient, the pre-trained model will gradually learn the ability to autonomously generate corresponding adjustment commands. This is because, during the continuous learning process, the model has accumulated a large amount of correlation information between severely degraded signals and adjustment commands, and can automatically infer appropriate adjustment measures based on newly detected severely degraded signals. For example, when the model detects that a key parameter of the optical eye diagram signal exceeds a certain range, it can generate adjustment commands such as increasing or decreasing the transmit power of optical network units or adjusting the signal modulation method based on previously learned experience.

[0069] Once the pre-trained model has the ability to autonomously generate adjustment commands, it can be responsible for outputting adjustment commands for relevant optical network units.

[0070] Optionally, to ensure the accuracy and security of the adjustments, the remote operation module can first upload the adjustment commands generated by the model to the cloud for operator confirmation. The cloud operator can review the model-generated adjustment commands based on their experience and understanding of the entire network. If the command is deemed reasonable, it is sent to the corresponding optical network unit; if the command is deemed problematic, it can be modified or rejected to avoid network failures due to model misjudgment.

[0071] In this way, the improved recognition capabilities of the pre-trained model enable it to detect severely degraded signals more quickly and accurately, and autonomously generate adjustment commands, reducing the time and workload of manual intervention and improving the efficiency of anomaly handling in the optical distribution network. Through continuous learning and iteration, the model can generate more appropriate adjustment commands based on actual conditions, promptly resolving anomalies in the network, thereby enhancing the stability and reliability of the optical distribution network. Simultaneously, it reduces the need for manual judgment and adjustment, lowering labor costs and the risk of human error, while also improving network resource utilization and reducing operational costs.

[0072] In summary, by continuously updating and optimizing the pre-trained model, a more intelligent and efficient solution is provided for anomaly detection and adjustment of the light allocation network.

[0073] In one embodiment, based on the above embodiments, the pre-trained model is further trained to generate corresponding adjustment commands based on key parameters in the severely degraded signal that exceed the nominal value range; When the pre-trained model identifies a severe degradation signal, it also generates corresponding adjustment commands and outputs them based on the key parameters in the severe degradation signal that exceed the nominal value range.

[0074] In this embodiment, a large amount of historical data on optical eye diagram signals in the optical distribution network is collected in advance. This data includes normal signals and various abnormal optical eye diagram signals, especially severely degraded signals. For each severely degraded signal, the specific deviations of its key parameters from the nominal range are recorded in detail, such as the specific deviations of parameters like signal amplitude, frequency, and phase.

[0075] Simultaneously, the adjustment commands implemented for these severely degraded signals and their corresponding effects are collected. These adjustment commands may include adjusting the transmit power of optical network units, changing the modulation scheme, and adjusting the signal bandwidth.

[0076] Key features are extracted from the collected severely degraded signal data, and critical parameters exceeding the nominal value range are quantified as important features. For example, the proportion of signal amplitude exceeding the nominal value and frequency offset are converted into numerical features. Each severely degraded signal sample is labeled with a corresponding adjustment command, forming a training data pair, i.e., (severely degraded signal feature, adjustment command).

[0077] Optionally, the model can be trained using labeled training data. By continuously adjusting the model's parameters, the model can learn the mapping relationship between the key parameters of the severely degraded signal and the adjustment commands. During training, an appropriate loss function is used to evaluate the difference between the model's predictions and the actual adjustment commands, such as cross-entropy loss function or mean squared error loss function. Optimization algorithms (such as stochastic gradient descent or Adam optimizer) are used to minimize the loss function, thereby continuously optimizing the model's performance.

[0078] Optionally, when the PON device detects an abnormal optical eye diagram signal and sends it to the remote operation module, the pre-trained model first analyzes the signal to determine whether it is a severely degraded signal. This process is achieved by detecting whether the key parameters of the signal exceed the nominal value range. The model compares the key parameters of the signal with the pre-set nominal values ​​and determines whether it is a severely degraded signal according to certain rules (such as parameter deviation exceeding a certain threshold).

[0079] Optionally, once the signal is confirmed to be severely degraded, the pre-trained model will generate corresponding adjustment commands based on the key parameters in the signal that exceed the nominal range, using the mapping relationships learned during training. For example, if the amplitude of the optical eye diagram signal is detected to be lower than the nominal value by a certain proportion, the model may generate an adjustment command to increase the transmit power of the optical network unit; if the frequency offset of the signal exceeds the allowable range, the model may generate a command to adjust the modulation frequency of the signal.

[0080] Optionally, the generated adjustment commands will be output and can be uploaded to the cloud for operator confirmation. After confirmation, the adjustment commands are sent to the corresponding optical network units via the downlink channel of the optical distribution network to achieve adjustment and optimization of the optical network.

[0081] In one embodiment, when a severely degraded signal is detected, the pre-trained model can immediately generate adjustment commands based on the signal characteristics without manual intervention, which greatly shortens the time for anomaly handling and improves the response speed and stability of the optical distribution network.

[0082] By learning from a large amount of historical data, the model can more accurately grasp the relationship between the characteristics of severe degradation signals and adjustment commands, and the generated adjustment commands are more targeted and effective, reducing the problem of improper adjustment caused by human judgment errors.

[0083] In this way, the pre-trained model has the ability to autonomously generate adjustment commands, making the operation and maintenance of the optical distribution network more intelligent, reducing reliance on human experience, and improving operation and maintenance efficiency and quality.

[0084] In one embodiment, based on the above embodiments, the pre-trained model, based on the initial nominal value range, is trained and learns the nominal value range corresponding to each network configuration for different network configurations of the optical distribution network. In the process of the pre-trained model to identify severely degraded signals, the corresponding nominal value range is obtained based on the network configuration of the current optical distribution network to identify abnormal optical eye diagram signals.

[0085] In this embodiment, different network configurations in the optical distribution network significantly affect the transmission characteristics of optical signals. For example, factors such as network topology (e.g., tree, ring), fiber length, and optical splitting ratio can all lead to variations in attenuation and dispersion of optical signals during transmission, thus affecting key parameters of the eye diagram signal. Therefore, a single initial nominal value range may not be sufficient to accurately determine whether the optical eye diagram signal is severely degraded under different network configurations. To improve the accuracy of anomaly detection, the pre-trained model can also learn corresponding nominal value ranges for different network configurations.

[0086] First, a detailed classification of the possible network configurations for optical distribution networks is conducted. This can include a combination of factors such as network topology (e.g., tree, ring, star), optical splitting ratio (e.g., 1:16, 1:32, 1:64), and fiber length range (e.g., short distance <10km, medium distance 10-20km, long distance >20km).

[0087] For each network configuration, a large amount of eye diagram signal data was collected from the normally operating optical distribution network. This data can be obtained through long-term monitoring using PON devices (optical line terminals and optical network units). For example, in a network configuration with a tree topology, an optical split ratio of 1:32, and fiber lengths of 15-20km, continuous monitoring for one month collected thousands of eye diagram signal samples.

[0088] Feature extraction was performed on the collected eye diagram signal data to identify key parameters for judging whether the signal is severely degraded. The distribution of these key parameters under different network configurations was analyzed. For example, in short-range, low optical split-ratio network configurations, the eye height may generally be higher, while in long-range, high optical split-ratio network configurations, the eye height may be significantly lower.

[0089] Optionally, the pre-trained model can be trained using machine learning or deep learning algorithms. The network configuration is used as the input feature, and the range of values ​​for key parameters is the output target. For example, a neural network model can be used, taking relevant information about the network configuration (such as topology encoding, optical splitting ratio, fiber length, etc.) as input to the input layer. After calculation through multiple hidden layers, the output is the nominal range of values ​​for key parameters under the corresponding network configuration.

[0090] During training, the model's parameters are continuously adjusted to ensure accurate prediction of the nominal range of key parameters under different network configurations. Loss functions such as mean squared error can be used to measure the difference between the model's predicted values ​​and the actual nominal range, and the value of the loss function can be continuously reduced through optimization algorithms (such as stochastic gradient descent).

[0091] When the pre-trained model receives an abnormal optical eye diagram signal, it first determines the current network configuration of the optical distribution network. This can be obtained by querying the network management system or the configuration information of the PON device. For example, the current network topology, optical splitting ratio, and other information can be obtained through the configuration file of the optical line terminal, and the actual length of the optical fiber can be obtained through an optical fiber length measuring device.

[0092] Optionally, based on the identified current network configuration, the nominal value range of the corresponding key parameters can be obtained from the pre-trained model. For example, if the current network configuration is a tree topology, the optical splitting ratio is 1:64, and the fiber length is 25km, the model will output the nominal value range of key parameters such as eye height, eye width, and extinction ratio under this network configuration.

[0093] The key parameters of the abnormal eye diagram signal are compared with the obtained nominal value range. If a key parameter exceeds the nominal value range, the abnormal eye diagram signal is determined to be a severely degraded signal. For example, if the nominal value range of eye height is 0.8 - 1.2V under the current network configuration, and the measured eye height value of the abnormal eye diagram signal is 0.6V, then the signal can be determined to be severely degraded.

[0094] In this way, the pre-trained model can take into account the impact of different network configurations on optical signal transmission, improving the accuracy and specificity of anomaly detection. It avoids misjudgments that might occur when using a single nominal value range, making anomaly detection in optical distribution networks more intelligent and reliable.

[0095] In one embodiment, based on the above embodiments, the key parameters include at least one of eye height, eye width, eye ascent time, and eye descent time.

[0096] In this embodiment, eye height refers to the vertical height of the eye diagram opening, which reflects the amplitude characteristics of the signal. In an ideal optical communication system, the signal amplitude should be stable and meet design requirements.

[0097] The size of the eye height directly affects the accuracy of signal determination. If the eye height is too small, it means that the amplitude variation range of the signal is narrowed, making it more susceptible to noise and interference during signal determination at the receiver, thus increasing the bit error rate. When the pre-trained model detects an eye height exceeding the nominal range, it may indicate problems such as optical signal attenuation, amplifier failure, or modulator malfunction during transmission. If the eye height is below the lower limit of the nominal range, the receiver may be unable to accurately distinguish different signal states, resulting in a severe deterioration in signal quality. In this case, abnormal optical eye diagram signals are likely to be severely degraded signals.

[0098] Eye width refers to the horizontal width of the eye diagram opening, and it is closely related to inter-symbol interference (ISI). ISI occurs because the signal is affected by channel characteristics during transmission, causing the waveforms of consecutive symbols to overlap. A larger eye width indicates less ISI and better signal transmission quality. When the pre-trained model detects an eye width exceeding the nominal range, it may be due to fiber dispersion, excessively high transmission rates, or improper equalizer adjustments, leading to increased ISI. If the eye width is too small, the receiver will have difficulty accurately distinguishing different symbols when deciding the signal, resulting in a sharp increase in the bit error rate. In this case, the abnormal optical eye diagram signal is likely a severely degraded signal.

[0099] Eye rise time refers to the time required for a signal to rise from a low level to a high level. It reflects the rise edge characteristics of the signal and is related to the high-frequency components of the signal. A suitable eye rise time is crucial for fast and accurate signal transmission. If the eye rise time is too long, it indicates that the signal rise speed is too slow, which may narrow the effective bandwidth of the signal, affecting the transmission rate and quality. When the pre-trained model detects that the eye rise time exceeds the nominal range, it may be due to factors such as the drive circuit of the optical transmitter, the modulator, or the high-frequency characteristics of the optical fiber.

[0100] Excessive eye rise time can distort the signal waveform, affecting the receiver's correct judgment of the signal. When it exceeds a certain range, abnormal optical eye diagram signals may be judged as severely degraded signals.

[0101] Eye fall time refers to the time required for a signal to fall from a high level to a low level, reflecting the falling edge characteristics of the signal. Similar to eye rise time, eye fall time also affects signal transmission quality. If the eye fall time is too long, the falling edge of the signal will be slower, which will also narrow the effective bandwidth of the signal and increase inter-symbol interference. When the pre-trained model detects that the eye fall time exceeds the nominal range, it may indicate a problem with the relevant characteristics of the optical transmitter or transmission channel.

[0102] An eye descent time exceeding the nominal range will distort the signal waveform, affecting the receiver's accurate identification of the signal. In severe cases, the abnormal optical eye diagram signal may be a severely degraded signal.

[0103] Optionally, the pre-trained model will examine each of these key parameters (eye height, eye width, eye rise time, and eye fall time) in the abnormal optical eye diagram signal. When one or more of these parameters exceed the nominal range, the model will comprehensively determine whether the abnormal optical eye diagram signal is a severely degraded signal. For example, if the eye height is too low and the eye width is too small, the abnormal optical eye diagram signal is likely to be judged as a severely degraded signal. Subsequently, the remote operation module will send corresponding adjustment commands to the optical network unit that triggered the anomaly to improve the performance of the optical distribution network.

[0104] In one embodiment, based on the above embodiments, the PON chip in the PON device monitors the eye diagram signal in the optical distribution network through real-time on-chip data collection and preprocessing.

[0105] In this embodiment, the PON chip collects various data related to the optical signal for eye diagram signal analysis. This includes optical signal intensity information, as changes in optical signal intensity directly affect the amplitude of the eye diagram and thus the eye height parameter. For example, a weak optical signal may result in a smaller eye height. Simultaneously, it collects signal timing information, such as the rising and falling edges of the signal, which is crucial for calculating eye rise and fall times. Additionally, the chip records signal jitter, which affects the horizontal opening of the eye diagram, i.e., eye width.

[0106] The raw data collected may contain noise and interference. The PON chip cleans this data, removing noise points that clearly do not conform to normal signal characteristics. For example, external electromagnetic interference may cause sudden spikes in the collected optical signal intensity data. The chip uses specific algorithms to identify and remove these outliers, making the data for subsequent analysis purer and more reliable.

[0107] Optionally, the PON chip extracts key features relevant to eye diagram signal monitoring from the cleaned data. For example, it extracts features such as eye rise time and eye fall time from the collected signal time series data. These features are important criteria for subsequent judgment of whether the eye diagram signal is abnormal. Through feature extraction, a large amount of raw data can be transformed into a few key parameters, reducing the amount of data and improving the efficiency of analysis.

[0108] Optionally, to facilitate subsequent analysis and comparison, the PON chip will normalize the extracted feature data. Different feature parameters may have different value ranges, and normalization can unify these parameters into a relatively standard range. For example, the original value ranges of eye height and eye width may differ greatly. After normalization, they can be analyzed on the same scale, enabling the pre-trained model to more accurately judge the eye diagram signal.

[0109] After data collection and preprocessing, the PON chip can construct an eye diagram model based on the processed data and compare it with a normal eye diagram template. If the constructed eye diagram shows significant differences from the normal template in key parameters such as eye height, eye width, eye rise time, or eye fall time, it is determined to be an abnormal optical eye diagram signal. For example, when the eye width of the constructed eye diagram is significantly smaller than that of the normal template, it indicates that the signal may have significant inter-symbol interference, which is an abnormal situation.

[0110] In one embodiment, on-chip data collection and preprocessing using a PON chip offers advantages in terms of efficiency and integration. Compared to transmitting large amounts of raw data to external devices for processing, on-chip processing reduces data transmission latency and bandwidth requirements, while also lowering system complexity. Furthermore, the PON chip can be directly and tightly integrated with optical line terminals and optical network units, more closely mimicking the actual transmission environment of optical signals and improving monitoring accuracy and real-time performance.

[0111] Furthermore, this application embodiment also provides an optical distribution network system, the internal architecture of which can be as follows: Figure 2 As shown, the system includes a processor, memory, communication interface, and input interface connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database stores data called by the computer programs. The communication interface is used for data communication with external terminals. The input interface is used to receive signals input from external devices. When the computer program is executed by the processor, it implements an anomaly detection method for an optical distribution network as described in the above embodiment.

[0112] Those skilled in the art will understand that Figure 2 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the optical distribution network system to which the present application is applied.

[0113] Furthermore, this application also proposes a computer-readable storage medium comprising a computer program that, when executed by a processor, implements the steps of the anomaly detection method for the optical distribution network as described in the above embodiments. It is understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0114] In summary, the optical distribution network anomaly detection method, optical distribution network system, and computer-readable storage medium provided in this application embodiment, through real-time monitoring of PON devices, transmission of abnormal signals, analysis of pre-trained models, and adjustment of optical network units, achieve timely detection and handling of optical distribution network anomalies, thereby improving the reliability and stability of the optical communication system.

[0115] This approach enables the immediate detection of problems in the optical distribution network, moving beyond passive detection after a fault occurs. It involves continuous monitoring of the optical eye diagram signal, allowing for immediate analysis and processing upon detecting any anomalies. This real-time capability allows the system to respond promptly to potential problems, preventing further escalation and significantly reducing the impact of faults on the optical distribution network.

[0116] Meanwhile, analyzing abnormal optical eye diagram signals using pre-trained models can quickly and accurately determine whether the signal has severely degraded and the possible causes of the problem. By learning from a large amount of data, the pre-trained model has grasped the influence of various factors on optical eye diagram signals and can comprehensively consider multiple key parameters to provide more accurate analysis results. This precise analysis helps to quickly locate problems, reduces unnecessary troubleshooting steps, and improves problem-solving efficiency.

[0117] In summary, by monitoring eye diagrams in real time and using pre-defined models, degradation prediction of optical distribution networks was achieved, significantly improving the efficiency of anomaly detection in optical distribution networks and effectively helping operators to respond quickly to network degradation.

[0118] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media provided in this application and in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-speed SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0119] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0120] The above description is only a preferred embodiment of this application and does not limit the patent scope of this application. Any equivalent structural or procedural changes made based on the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. An anomaly detection method for an optical distribution network, characterized in that, include: Eye diagram signals in an optical distribution network are monitored based on PON equipment; wherein, the PON equipment includes each optical line terminal and / or an optical network unit associated with each optical line terminal; When an abnormal optical eye diagram signal is detected, the PON device sends the abnormal optical eye diagram signal to the remote operation module through the optical distribution network; The remote operation module calls the pre-trained model to detect whether the abnormal optical eye diagram signal is a severely degraded signal. The pre-trained model detects whether the abnormal optical eye diagram signal is a severely degraded signal by detecting whether the key parameters of the abnormal optical eye diagram signal exceed the nominal value range. If so, a corresponding adjustment command is sent to the optical network unit that triggered the abnormal optical eye diagram signal.

2. The anomaly detection method for an optical distribution network as described in claim 1, characterized in that, The step of sending a corresponding adjustment command to the optical network unit that triggers the abnormal optical eye diagram signal includes: The remote operation module uploads the abnormal optical eye diagram signal, which is a severely degraded signal, to the cloud so that the operator of the network management system associated with the cloud can determine whether the relevant optical network unit needs to be adjusted. When the cloud-based network management system receives an adjustment command, it sends the corresponding adjustment command to the relevant optical network unit through the downlink channel of the optical distribution network.

3. The anomaly detection method for an optical distribution network as described in claim 2, characterized in that, The anomaly detection method for the optical distribution network also includes: When the cloud-based network management system receives the adjustment command, it will also send the adjustment command to the remote operation module; The remote operation module associates adjustment commands with corresponding severe degradation signals and updates the pre-trained model based on the association results.

4. The anomaly detection method for an optical distribution network as described in claim 1, characterized in that, Based on the initial nominal value range, the pre-trained model is trained and learns the nominal value range corresponding to each network configuration for different network configurations of the optical distribution network. In the process of the pre-trained model to identify severely degraded signals, the corresponding nominal value range is obtained based on the network configuration of the current optical distribution network to identify abnormal optical eye diagram signals.

5. The anomaly detection method for an optical distribution network as described in claim 1, characterized in that, When the pre-trained model confirms a severe degradation signal, it also outputs key parameters in the severe degradation signal that exceed the nominal value range, so as to generate corresponding adjustment commands based on the output key parameters.

6. The anomaly detection method for an optical distribution network as described in any one of claims 1-4, characterized in that, The pre-trained model is also trained to generate corresponding adjustment commands based on key parameters in severely degraded signals that exceed the nominal range. When the pre-trained model identifies a severe degradation signal, it also generates corresponding adjustment commands and outputs them based on the key parameters in the severe degradation signal that exceed the nominal value range.

7. The anomaly detection method for an optical distribution network as described in any one of claims 1-4, characterized in that, The key parameters include at least one of eye height, eye width, eye ascent time, and eye descent time.

8. The anomaly detection method for an optical distribution network as described in claim 1, characterized in that, The PON chip in the PON device monitors the eye diagram signal in the optical distribution network through real-time on-chip data collection and preprocessing.

9. An optical distribution network system, characterized in that, The optical distribution network system includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When executed by the processor, the computer program implements the steps of the anomaly detection method for the optical distribution network as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the anomaly detection method for an optical distribution network as described in any one of claims 1 to 8.

Images