Abnormal reason positioning method and device based on 5G private network behavior log backtracking

Through deep learning of LSTM model and neural network prediction model, the abnormal association relationship in the 5G private network behavior log is identified and tracked, and the problem of the existing technology being unable to identify and locate unknown faults is solved, and the association of undefined exception logs and faults is realized. Identification improves the accuracy and real-timeness of abnormal positioning.

CN119011375AActive Publication Date: 2024-11-22CHINA MOBILE GRP GUANGDONG CO LTD +2

Patent Information

Application Number
CN202410966738.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-18
Publication Date
2024-11-22
Estimated Expiration
2044-07-18

AI Technical Summary

Technical Problem

The existing 5G private network abnormal positioning technology cannot identify and locate unknown faults in real time. It relies on artificially defined abnormal checklists to identify unknown faults.

Method used

By obtaining 5G behavioral log flow, marking exception logs, and using deep learning LSTM model and neural network prediction model, identifying exception logs and tracking their association relationships, realizing the identification of unknown exception faults.

Benefits of technology

The identification of undefined exception logs and fault associations is realized, and the problem of the inability to identify and locate unknown faults in the prior art is solved, and the accuracy and real-time nature of abnormal positioning are improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119011375A_ABST
    Figure CN119011375A_ABST
Patent Text Reader

Abstract

The invention discloses an abnormal reason positioning method and device based on 5G private network behavior log backtracking. The method comprises the steps of obtaining a to-be-analyzed log based on a 5G behavior log stream; matching marked abnormal logs in the to-be-analyzed logs with the to-be-analyzed logs so as to perform log history backtracking according to the configuration information of the corresponding logs and output a service flow log sequence; training a deep learning LSTM model based on the business flow log sequence to identify and output an abnormal log; and obtaining fault data of the business flow of the abnormal log with the new identifier, the reasons of which are not positioned in the corresponding time period, and training a neural network prediction model by using the abnormal log with the new identifier to identify an association relationship between the abnormal log with the new identifier and the fault data. According to the method, the unpredefined exceptions can be effectively identified and positioned.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of core network technology, and in particular to a method and device for locating abnormal causes based on 5G private network behavior log backtracing. Background Art

[0002] With the expansion of China's 5G commercial market, the rapid development of 5G industry virtual private networks, and the development model of cloud-network integration, the application side has gradually prospered and enriched, which has also led to more dynamic and complex services, making network management more difficult. For example, with the introduction of microservice architecture technology, the network is deployed on demand, which has caused great difficulties in the location and demarcation of anomalies in the operation and maintenance of 5G private networks. At present, the industry has proposed a large number of solutions to the problem of abnormal location in the 5G network environment, such as the patent document "CN113762978A5G Slice User Complaint Demarcation Method, Device and Computing Equipment", which uses a complaint demarcation model based on a hybrid graph network to jointly learn the graph network that processes the slice instance topology and the convolutional neural network that processes the slice user complaint text, and predict the demarcation results, so as to accurately determine the source node of the slice user complaint, which is conducive to the subsequent targeted complaint handling.

[0003] The existing abnormal cause location solution based on user complaints is backward and cannot solve the problem of real-time abnormal location. In addition, due to the sporadic nature of complaints, it is difficult to form targeted abnormal cause location.

[0004] In actual management activities, operation and maintenance personnel mainly rely on the 5G private network behavior log backtracking related functions to achieve real-time tracking or historical backtracking of the terminal's behavior log, so as to discover abnormal behavior of the terminal and pay attention and deal with it in a timely manner. At present, the industry's 5G private network behavior log backtracking function is based on the UPF traffic probe, which can filter all behavior records within the user-selected tracking time and identify abnormal records. However, the problem with the existing technology is that only pre-defined anomalies can be identified, that is, the current abnormal cause is located and confirmed by looking up the table of artificially defined anomalies, and unknown faults can neither be identified nor located. Summary of the invention

[0005] The present invention aims to solve one of the technical problems in the related art at least to a certain extent.

[0006] To this end, the present invention proposes a method for locating the cause of anomalies based on 5G private network behavior log backtracing. By identifying the related behaviors in the logs through big data analysis, and then tracking the abnormal logs associated with the logs where the abnormal behaviors occurred, the identification of unknown abnormal faults can be achieved, thereby solving the problem that the prior art can neither identify nor locate unknown faults.

[0007] Another object of the present invention is to propose a device for locating abnormal causes based on 5G private network behavior log backtracking.

[0008] To achieve the above object, the present invention proposes, on one hand, a method for locating abnormal causes based on 5G private network behavior log backtracking, comprising:

[0009] Obtain the logs to be analyzed based on the 5G behavior log stream;

[0010] Pairing the marked abnormal logs in the log to be analyzed with the log to be analyzed, so as to perform log history backtracking and output a business flow log sequence according to the configuration information of the corresponding log;

[0011] Based on the business flow log sequence, a deep learning LSTM model is trained to identify and output abnormal logs;

[0012] Obtain all the fault data of the business flow with the newly identified exception log in the corresponding time period whose cause has not been located, and use the newly identified exception log to train the neural network prediction model to identify the association between the newly identified exception log and the fault data.

[0013] The abnormal cause location method based on 5G private network behavior log backtracing in the embodiment of the present invention may also have the following additional technical features:

[0014] In one embodiment of the present invention, obtaining a log to be analyzed based on a 5G behavior log stream includes:

[0015] Get 5G behavior log stream;

[0016] The whitelist logs and logs marked as abnormal in the 5G behavior log stream are filtered, and the unfiltered logs in the 5G behavior log stream are marked as logs to be analyzed.

[0017] In one embodiment of the present invention, the marked abnormal logs existing in the log to be analyzed are paired with the log to be analyzed, so as to perform log history backtracking and output a business flow log sequence according to the configuration information of the corresponding log, including:

[0018] Extract the log slices to be analyzed;

[0019] Obtaining the marked abnormal logs in the log slice to be analyzed, and pairing the log to be analyzed with each abnormal log respectively, so as to perform log history backtracking according to the configuration information of the two paired logs;

[0020] Based on the log history backtracking result, all log data within the first preset time period is obtained, and the log sequences under the same business process are aggregated according to the business identifiers marked by the log system to output the business flow log sequence.

[0021] In one embodiment of the present invention, training a deep learning LSTM model based on the service flow log sequence to identify and output abnormal logs includes:

[0022] Extracting training sample data of the business flow log sequence using a sliding window;

[0023] Using the training sample data to train the deep learning LSTM model enables the model to have the ability to identify abnormal logs in the log sequence to obtain a trained deep learning LSTM model;

[0024] Identify and output abnormal logs based on the trained deep learning LSTM model.

[0025] In one embodiment of the present invention, all the fault data of the business flow with the newly identified abnormal log in the corresponding time period that have not been located, and the neural network prediction model is trained by using the newly identified abnormal log to identify the association relationship between the newly identified abnormal log and the fault data, including:

[0026] Call the exception log of the new identifier;

[0027] Acquire the business flow of the abnormal log with the new identifier, and extract all the fault data of the business flow whose cause has not been located within the second preset time period;

[0028] Inputting the distribution information, log sequence information and log variable information of the newly identified abnormal log in the business flow into the neural network prediction model to obtain a trained neural network prediction model based on the model output result;

[0029] The trained neural network prediction model is used to identify the association between the newly marked abnormal log and the fault data.

[0030] To achieve the above-mentioned purpose, the present invention proposes, on the other hand, a device for locating abnormal causes based on 5G private network behavior log backtracking, comprising:

[0031] The module for obtaining logs to be analyzed is used to obtain logs to be analyzed based on 5G behavior log streams;

[0032] A log history backtracking module, used to pair the marked abnormal logs existing in the log to be analyzed with the log to be analyzed, so as to perform log history backtracking and output a business flow log sequence according to the configuration information of the corresponding log;

[0033] An abnormal log identification module is used to train a deep learning LSTM model based on the business flow log sequence to identify and output abnormal logs;

[0034] The association relationship positioning module is used to obtain all the fault data of the business flow with the newly identified abnormal log in the corresponding time period whose cause has not been located, and use the newly identified abnormal log to train the neural network prediction model to identify the association relationship between the newly identified abnormal log and the fault data.

[0035] The method and device for locating the cause of anomalies based on 5G private network behavior log backtracing in the embodiments of the present invention, by associating logs of unknown cause to logs of known anomalies, tracking the contextual relationship between logs based on the full-link business process of the logs, identify the log anomalies when the two occur at the same time, and then train the identified abnormal logs through a neural network model to determine whether they are associated with faults, so that the model has the ability to predict faults, thereby realizing a solution for mining undefined abnormal logs and associated faults.

[0036] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0038] Figure 1 It is a flowchart of a method for locating an abnormality cause based on 5G private network behavior log backtracing according to an embodiment of the present invention;

[0039] Figure 2 It is a logic diagram of a method for locating an abnormality cause based on 5G private network behavior log backtracing according to an embodiment of the present invention;

[0040] Figure 3 is a flowchart of pairing the log to be analyzed with the abnormal log and tracing back the history according to an embodiment of the present invention;

[0041] Figure 4 is a schematic diagram of deep learning LSTM model training according to an embodiment of the present invention;

[0042] Figure 5 is a flow chart of abnormal log identification according to an embodiment of the present invention;

[0043] Figure 6 is a schematic diagram of training a neural network with training sample data according to an embodiment of the present invention;

[0044] Figure 7 is a flowchart of locating the cause of a fault according to an embodiment of the present invention;

[0045] Figure 8 It is a structural diagram of a device for locating abnormal causes of 5G private network behavior log backtracing according to an embodiment of the present invention. DETAILED DESCRIPTION

[0046] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0047] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0048] The following describes, with reference to the accompanying drawings, a method and device for locating abnormal causes based on 5G private network behavior log backtracing proposed in accordance with an embodiment of the present invention.

[0049] Figure 1 is a flow chart of a method for locating abnormal causes based on 5G private network behavior log backtracing according to an embodiment of the present invention. Figure 1 As shown, the method includes but is not limited to the following steps:

[0050] S1, obtaining the log to be analyzed based on the 5G behavior log stream;

[0051] S2, pairing the marked abnormal logs in the log to be analyzed with the log to be analyzed, so as to perform log history backtracking and output a business flow log sequence according to the configuration information of the corresponding log;

[0052] S3, trains a deep learning LSTM model based on the business flow log sequence to identify and output abnormal logs;

[0053] S4, obtaining all the fault data of the business flow with the newly identified abnormal log in the corresponding time period whose cause has not been located, and using the newly identified abnormal log to train the neural network prediction model to identify the association between the newly identified abnormal log and the fault data.

[0054] It is understandable that the present invention predefines a log-event template, which can associate logs and events based on the UPF soft probe monitoring information, that is, to filter out whitelist events and identify log abnormal events in the collected logs, and locate the abnormal cause of the log abnormal event; in terms of log event detection, the located fault is provided to the real-time fault location for output. At the same time, in terms of real-time log processing, the log is sliced ​​in the real-time anomaly detection, and the log slices are filtered out whitelist events and identifiable log abnormal events through log anomaly detection. The remaining are logs to be analyzed, and the logs to be analyzed are analyzed for association with abnormal logs, and further associated with fault events occurring at corresponding times, thereby forming abnormal event location that associates non-abnormal logs with fault events. Figure 2 shown.

[0055] Specifically, Figure 3 The flowchart of pairing the log to be analyzed with the abnormal log and tracing back the history implemented by the present invention is as follows: Figure 3 As shown, the process includes:

[0056] S301, obtaining a real-time log stream;

[0057] S302, filtering whitelist logs and logs marked as abnormal through a log anomaly detection template;

[0058] S303, marking the unfiltered logs as logs to be analyzed, and extracting each log slice to be analyzed;

[0059] S304: If there are other marked abnormal logs in the log slice to be analyzed, the log to be analyzed is paired with each abnormal log respectively;

[0060] S305, performing log history backtracking according to the configuration information of the two paired logs; configuration information: status and parameter information of the terminal device associated with both logs; log history backtracking: retrieving all log data within a preset time period (i.e., the first preset time period in the embodiment of the present invention) when the two logs occurred on the associated terminal device from the historical log database, such as within one hour;

[0061] S306: The slice has a paired log with no backtracked log data;

[0062] S307, aggregating the log sequences under the same business process according to the business identifiers marked by the log system for all the extracted log data;

[0063] S308, there are unpaired log slices to be analyzed;

[0064] S309, output the business flow log sequence.

[0065] In one embodiment of the present invention, the real-time log history is traced back according to the configuration information of the log to be analyzed and the abnormal log in the same slice input by the user, and the configuration information is passed to the UPF soft probe module of the UPF soft probe interface. The UPF soft probe module filters out the relevant UPF signaling according to the text and time parameters of the configuration information, that is, the associated historical log information and then transmits it back. UPF (User Plane Function) User plane function.

[0066] It is understandable that UPF is a key component in the 5G core network (5GC), which is mainly responsible for the processing and forwarding of user plane data. In the 5G network architecture, the control plane (CP) is separated from the user plane (UP) (CUPS, Control and User Plane Separation). This design improves flexibility and efficiency, enabling the network to better support diverse service requirements, especially for scenarios where low latency and large bandwidth applications are critical. The core responsibilities of UPF include: Packet routing and forwarding: UPF is responsible for determining the routing of user data packets based on policies and efficiently forwarding data from one network node to another, including terminal devices, base stations or other network functional entities. Session and data management: Working in conjunction with the SMF (Session Management Function) of the control plane, UPF participates in the establishment, modification and release of user sessions and manages the context of user data flows. Policy execution: According to the policies and rules provided by the PCF (Policy Control Function), execute data packet processing decisions, such as QoS (Quality of Service) guarantee, billing, packet inspection, etc. Edge computing and network slicing: As a key interface for 5G network slicing and edge computing, UPF can be deployed closer to users, such as edge data centers, to reduce latency and improve user experience. Security and billing: UPF also participates in data security processing, such as encryption and decryption, as well as reporting of user data traffic billing information to ensure that service providers can accurately measure and bill.

[0067] It is understandable that the log data of the operation and maintenance system records the various status and parameter information of the terminal device, and the terminal device has stronger universality and identifiability than other components of the 5G private network. Therefore, the present invention uses the information of the terminal device as the association node, and pairs the log data of unknown abnormality to the log data of known abnormalities, so as to find the log that can be associated with the abnormal log in the complex log data of unknown abnormality and mine out the undefined abnormal log.

[0068] The present invention extracts all abnormal logs in the slice where the log is located and the two connected slices for the log to be analyzed, takes the terminal device associated with the log (all abnormal logs) as the association standard, and identifies the log associated with the business flow of the log to be analyzed. The log to be analyzed and its associated logs are backtracked in a two-by-two matching manner. Specifically, the logs under the same business process are aggregated according to the business identifier, and then the business flows that are simultaneously associated with the two logs within a set time period are screened, thereby filtering out the historical log information associated with both, and outputting the business flow log sequence related to the log to be analyzed, so as to identify the abnormal log through the log sequence.

[0069] Further, Figure 4 The abnormal log identification flow chart implemented by the present invention is as follows: Figure 4 As shown, the process includes:

[0070] S401, extracting a business flow log sequence;

[0071] S402, extracting training sample data of the log sequence through a sliding window;

[0072] S403, using the training sample data to train the neural network using algorithms such as gradient descent, so that the neural network has the ability to identify abnormal logs in the log sequence, so as to obtain a trained deep learning LSTM model;

[0073] S404, identifying abnormal logs based on the trained deep learning LSTM model;

[0074] S405, determining whether there is a business flow log sequence with unidentified abnormal logs;

[0075] S406, outputting an exception log.

[0076] Specifically, the embodiment of the present invention needs to extract the associated log sequence of the two logs, namely the log abnormal event that has been identified and the log to be analyzed, so as to train the deep learning LSTM model to identify abnormal logs. LSTM (Long Short-term Memory Networks), long short-term memory neural network.

[0077] It is understandable that LSTM introduces a memory cell and three gates: input gate, forget gate, and output gate. The functions of these gates are as follows: Forget gate: determines which information is discarded from the memory cell, and a value between 0 and 1 is calculated through a sigmoid function, where 0 means complete forgetting and 1 means complete retention. Input gate: determines which new information should be added to the memory cell, and is also calculated through a sigmoid function. There is also a tanH function to process the new information (called the candidate value of the input gate). Cell state: combines the outputs of the forget gate and the input gate to update the state of the memory cell, which includes both forgetting old information and adding new information. Output gate: controls which information in the memory cell will be used for the output at the current moment, and also determines what to output through a sigmoid function, and then processes the memory cell state through a tanh function to obtain the final output. Its application scenarios: LSTM is widely used in various sequence data processing tasks due to its advantages in processing long-term dependencies, including: Natural language processing: machine translation, sentiment analysis, text generation, and speech recognition. Speech recognition: recognition and transcription of long-sequence speech signals. Video analysis: action recognition, scene understanding. Image recognition: image classification and generation combined with sequence information. Time series prediction: stock market prediction, weather forecast, etc.

[0078] LSTM achieves learning of long-term dependencies through its unique gating mechanism, overcoming the limitations of traditional RNN in processing long sequence data, and is an important progress in the field of sequence learning. With the advancement of deep learning research, LSTM has also been continuously optimized and expanded, combined with other technologies, such as attention mechanism, bidirectional LSTM, etc., to further improve the performance and application scope of the model.

[0079] like Figure 5 As shown, the present invention takes the virtual machine mapping service flow as an example, and the associated log sequence Sn={E10, E12, E26, E32, E20}:

[0080] E10, Auditing locally available compute resources for node xxxx;

[0081] E12. Compute_service record updated for xxx:xxx;

[0082] E26. Final resource view: name=xxx phys_ram=385384MB used_ram=90112MB phys_disk=339GB used_disk=0GB total_vcpus=40used_vcpus=20pci_stats=[];

[0083] E32, Total usable vcpus:40, total allocated vcpus:20;

[0084] E20. compute.metrics.update is not a versioned notification and notwhitelisted.See. / doc / source / notification.rst.

[0085] Exemplarily, the present invention sets the deep learning LSTM model window length h (h=3 as an example), the step length s=1, extracts training sample data from each sequence in turn by sliding the window, and uses the training sample data to train the neural network with algorithms such as the gradient descent method to identify abnormal logs in the log sequence. Figure 6 shown.

[0086] Specifically, during training, when Sn-1 is passed, the Poisson probability P(Tn|Sn-1) distribution is obtained. The probability distribution of each log Tn obtained through training is recorded, and the probability distribution TopN of each log Tn is recorded. For example, given a log sequence to be detected: E10, E12, E26, E32, E35, the Sn-1 of log Tn in this log sequence is E12, E26, E32, E35, and E35 is not in the TopN interval of log Tn, then log E35 is identified as an abnormal log.

[0087] Therefore, the present invention screens the logs to be analyzed that are close in time to the logs where the exception occurs and cannot be analyzed whether they are related to the fault through real-time log slicing, takes the terminal device that is associated with the log to be analyzed and the exception log as the associated node, screens the business flows that are simultaneously associated with the two logs within a set time period, identifies the abnormal logs in the screened business flows by training a deep learning LSTM model, and realizes the identification and positioning of abnormal logs that are not pre-defined.

[0088] Further, Figure 7 This is a flow chart of fault cause location implemented by the present invention, such as Figure 7 As shown, the process includes:

[0089] S701, calling the exception log of the new identifier;

[0090] S702, for all business flows of the abnormal log with the new identifier, extract all the fault data of the business flows within a second preset time period (the time period can be any time period) whose causes have not been located, and the embodiment of the present invention gives priority to the device terminal fault data;

[0091] S703, using the distribution information of the log in the business flow, the log sequence information, and the log variable information as input, and whether each fault occurs as output, to train a neural network prediction model that can identify whether a certain type of fault is associated with the log;

[0092] S704, using the trained neural network prediction model to identify the association between each newly identified abnormal log and the device terminal failure data, so as to associate the abnormal log with the terminal device failure.

[0093] Specifically, an embodiment of the present invention matches all faults whose causes have not been located within the corresponding time period of relevant historical log information. For each fault, it is analyzed whether it is associated with the log marked as abnormal, and a neural network prediction model is established to perform prediction. Specifically, the distribution information, sequence information, and variable status information of the log are used as input, and whether the fault occurs is used as output, thereby realizing the association identification of abnormal behaviors in the behavior log that may cause the fault.

[0094] Therefore, the present invention establishes a neural network prediction model for the identified abnormal log, takes its multiple configuration information as input, and outputs whether a terminal device failure of unknown cause occurs when it occurs, thereby obtaining an identification capability of determining whether the abnormal log is associated with a fault, thereby realizing fault association of undefined abnormal logs, that is, locating the log abnormality corresponding to the terminal device fault.

[0095] According to the abnormal cause positioning method based on 5G private network behavior log backtracing according to the embodiment of the present invention, the related behaviors in the log are identified through big data analysis, and then the abnormal logs associated with the logs where the abnormal behavior occurs are tracked to realize the identification of unknown abnormal faults, thereby solving the problem that the prior art can neither identify nor locate unknown faults.

[0096] In order to implement the above embodiment, Figure 8 As shown, this embodiment also provides an abnormality cause positioning device 10 based on 5G private network behavior log backtracing, and the device 10 includes:

[0097] The log acquisition module 100 to be analyzed is used to acquire the log to be analyzed based on the 5G behavior log stream;

[0098] The log history backtracking module 200 is used to pair the marked abnormal logs in the log to be analyzed with the log to be analyzed, so as to perform log history backtracking and output a business flow log sequence according to the configuration information of the corresponding log;

[0099] The abnormal log identification module 300 is used to train a deep learning LSTM model based on a business flow log sequence to identify and output abnormal logs;

[0100] The association location module 400 is used to obtain all the fault data of the business flow with the newly identified exception log in the corresponding time period whose cause has not been located, and use the newly identified exception log to train the neural network prediction model to identify the association relationship between the newly identified exception log and the fault data.

[0101] Furthermore, the module 100 for obtaining the logs to be analyzed is also used for:

[0102] Get 5G behavior log stream;

[0103] The whitelist logs and the logs marked as abnormal in the 5G behavior log stream are filtered, and the unfiltered logs in the 5G behavior log stream are marked as logs to be analyzed.

[0104] Furthermore, the log history backtracking module 200 is also used for:

[0105] Extract the log slices to be analyzed;

[0106] Obtaining the marked abnormal logs in the log slice to be analyzed, and pairing the log to be analyzed with each abnormal log respectively, so as to perform log history backtracking according to the configuration information of the two paired logs;

[0107] Based on the log history backtracking result, all log data within the first preset time period is obtained, and the log sequences under the same business process are aggregated according to the business identifiers marked by the log system to output the business flow log sequence.

[0108] Furthermore, the above-mentioned abnormal log identification module 300 is also used for:

[0109] Extracting training sample data of the business flow log sequence using a sliding window;

[0110] Using the training sample data to train the deep learning LSTM model enables the model to have the ability to identify abnormal logs in the log sequence to obtain a trained deep learning LSTM model;

[0111] Identify and output abnormal logs based on the trained deep learning LSTM model.

[0112] Furthermore, the above association relationship positioning module 400 is also used for:

[0113] Call the exception log of the new identifier;

[0114] Obtaining the business flow of the abnormal log with the new identifier, and extracting all the fault data of the business flow whose cause has not been located within the second preset time period;

[0115] Inputting the distribution information, log sequence information and log variable information of the newly identified abnormal log in the business flow into the neural network prediction model to obtain a trained neural network prediction model based on the model output result;

[0116] The trained neural network prediction model is used to identify the correlation between the newly identified abnormal logs and fault data.

[0117] According to the embodiment of the present invention, the abnormal cause positioning device based on 5G private network behavior log backtracing identifies the related behaviors in the log through big data analysis, and then tracks the abnormal logs associated with the logs where the abnormal behaviors occur, thereby realizing the identification of unknown abnormal faults, thereby solving the problem that the prior art can neither identify nor locate unknown faults.

[0118] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0119] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

Claims

1. A method for locating abnormal causes based on 5G private network behavior log backtracking, characterized in that: include: Obtain the logs to be analyzed based on the 5G behavior log stream; Pairing the marked abnormal logs in the log to be analyzed with the log to be analyzed, so as to perform log history backtracking and output a business flow log sequence according to the configuration information of the corresponding log; Based on the business flow log sequence, a deep learning LSTM model is trained to identify and output abnormal logs; Obtain all the fault data of the business flow with the newly identified exception log in the corresponding time period whose cause has not been located, and use the newly identified exception log to train the neural network prediction model to identify the association between the newly identified exception log and the fault data.

2. The method according to claim 1, characterized in that: Obtain the logs to be analyzed based on the 5G behavior log stream, including: Get 5G behavior log stream; The whitelist logs and logs marked as abnormal in the 5G behavior log stream are filtered, and the unfiltered logs in the 5G behavior log stream are marked as logs to be analyzed.

3. The method according to claim 2, characterized in that Pairing the marked abnormal logs in the log to be analyzed with the log to be analyzed, so as to perform log history backtracking and output a business flow log sequence according to the configuration information of the corresponding log, including: Extract the log slices to be analyzed; Obtaining the marked abnormal logs in the log slice to be analyzed, and pairing the log to be analyzed with each abnormal log respectively, so as to perform log history backtracking according to the configuration information of the two paired logs; Based on the log history backtracking result, all log data within the first preset time period is obtained, and the log sequences under the same business process are aggregated according to the business identifiers marked by the log system to output the business flow log sequence.

4. The method according to claim 1, characterized in that: Based on the business flow log sequence, a deep learning LSTM model is trained to identify and output abnormal logs, including: Extracting training sample data of the business flow log sequence using a sliding window; Using the training sample data to train the deep learning LSTM model enables the model to have the ability to identify abnormal logs in the log sequence to obtain a trained deep learning LSTM model; Identify and output abnormal logs based on the trained deep learning LSTM model.

5. The method according to claim 1, characterized in that Obtaining all the fault data of the business flow with the newly identified abnormal log in the corresponding time period whose cause has not been located, and using the newly identified abnormal log to train the neural network prediction model to identify the association between the newly identified abnormal log and the fault data, including: Call the exception log of the new identifier; Acquire the business flow of the abnormal log with the new identifier, and extract all the fault data of the business flow whose cause has not been located within the second preset time period; Inputting the distribution information, log sequence information and log variable information of the newly identified abnormal log in the business flow into the neural network prediction model to obtain a trained neural network prediction model based on the model output result; The trained neural network prediction model is used to identify the association between the newly identified abnormal log and the fault data.

6. A device for locating abnormal causes based on 5G private network behavior log backtracking, characterized in that: include: The module for obtaining logs to be analyzed is used to obtain logs to be analyzed based on 5G behavior log streams; A log history backtracking module, used to pair the marked abnormal logs existing in the log to be analyzed with the log to be analyzed, so as to perform log history backtracking and output a business flow log sequence according to the configuration information of the corresponding log; An abnormal log identification module is used to train a deep learning LSTM model based on the business flow log sequence to identify and output abnormal logs; The association relationship positioning module is used to obtain all the fault data of the business flow with the newly identified abnormal log in the corresponding time period whose cause has not been located, and use the newly identified abnormal log to train the neural network prediction model to identify the association relationship between the newly identified abnormal log and the fault data.

7. The device according to claim 6, characterized in that The module for obtaining the logs to be analyzed is further used for: Get 5G behavior log stream; The whitelist logs and logs marked as abnormal in the 5G behavior log stream are filtered, and the unfiltered logs in the 5G behavior log stream are marked as logs to be analyzed.

8. The device according to claim 7, characterized in that The log history backtracking module is also used to: Extract the log slices to be analyzed; Obtaining the marked abnormal logs in the log slice to be analyzed, and pairing the log to be analyzed with each abnormal log respectively, so as to perform log history backtracking according to the configuration information of the two paired logs; Based on the log history backtracking result, all log data within the first preset time period is obtained, and the log sequences under the same business process are aggregated according to the business identifiers marked by the log system to output the business flow log sequence.

9. The device according to claim 6, characterized in that The abnormal log identification module is also used for: Extracting training sample data of the business flow log sequence using a sliding window; Using the training sample data to train the deep learning LSTM model enables the model to have the ability to identify abnormal logs in the log sequence to obtain a trained deep learning LSTM model; Identify and output abnormal logs based on the trained deep learning LSTM model.

10. The device according to claim 6, characterized in that The association relationship positioning module is also used for: Call the exception log of the new identifier; Acquire the business flow of the abnormal log with the new identifier, and extract all the fault data of the business flow whose cause has not been located within the second preset time period; Inputting the distribution information, log sequence information and log variable information of the newly identified abnormal log in the business flow into the neural network prediction model to obtain a trained neural network prediction model based on the model output result; The trained neural network prediction model is used to identify the association between the newly identified abnormal log and the fault data.

Citation Information

Patent Citations

  • Method and apparatus for rapid location of anomalies in IP traffic logs

    US20050207413A1

  • Log analysis system, log analysis method and program recording medium

    US20180046529A1

  • Method and device for detecting abnormal log, and storage medium

    WO2024001656A1

  • Root cause positioning method, and communication device and computer-readable storage medium

    WO2024012186A1

Cited By

  • Problem log processing method and device and product

    CN119892613A