A knowledge graph-based wireless network event root cause analysis method

Abstract

The application discloses a wireless network event root cause analysis method based on a knowledge graph. The method comprises the following steps: acquiring historical operation data of a wireless network; determining multiple event relationship information of each cell based on various index data of each cell and in combination with a predefined index threshold table; determining neighbor cell related event relationship information corresponding to each event relationship information of each cell based on neighbor cell handover data of each cell; generating an event knowledge graph based on the event relationship information of each cell and the corresponding neighbor cell related event relationship information; and querying other events related to a target event based on the event knowledge graph, and taking the other events as a root cause analysis result of the target event. The method utilizes the event knowledge graph to simplify the root cause analysis process of the wireless network fault, and effectively improves the accuracy and efficiency of the analysis.

Description

Technical Field

[0001] This application relates to the field of communication technology, and in particular to a knowledge graph-based method for root cause analysis of wireless network events. Background Technology

[0002] With the rapid development of information technology, mobile communication networks have penetrated into all aspects of personal life and business operations, becoming an indispensable infrastructure of modern society. However, frequent network failures not only affect users' daily experience but can also lead to reduced operational efficiency and economic losses for businesses. Therefore, quickly and effectively identifying and resolving wireless network failures is crucial.

[0003] In wireless networks, the causes of failures are often complex and varied, including hardware problems, software defects, network configuration errors, and external environmental interference. Traditional troubleshooting methods typically rely on manual analysis by experienced technicians. This approach is not only inefficient but also susceptible to the influence of subjective judgment, leading to prolonged troubleshooting time and an increased problem recurrence rate.

[0004] In recent years, Root Cause Analysis (RCA) has gained increasing importance in wireless network fault management as a systematic problem-solving method. RCA aims to identify the root causes of problems through in-depth analysis of fault phenomena, thereby developing targeted solutions. This method helps improve network stability and reliability, and reduces the impact of faults on users.

[0005] However, current root cause analysis methods largely rely on pre-defined mappings between problems and solution processes, requiring analysts to possess extensive professional knowledge and experience. This not only complicates the troubleshooting process but also increases the difficulty of system management and maintenance. Therefore, a novel method is urgently needed to simplify the root cause analysis process for wireless network faults and improve the accuracy and efficiency of the analysis. Summary of the Invention

[0006] This application provides a knowledge graph-based root cause analysis method for wireless network events, aiming to simplify the root cause analysis process for wireless network anomalies.

[0007] To achieve the above objectives, this application provides the following technical solution:

[0008] A knowledge graph-based method for root cause analysis of wireless network events includes:

[0009] Acquire historical operational data of the wireless network; the historical operational data includes various indicator data and neighbor cell handover data of multiple cells over a period of time; the various indicator data includes at least the change data of performance indicators and related indicators;

[0010] Based on various indicator data of each cell, and in conjunction with a predefined indicator threshold table, multiple event relationship information for each cell is determined; the indicator threshold table includes thresholds for various indicators; the event relationship information includes a central event and corresponding related events; the central event indicates that the daily changes of the performance indicator are abnormal; the related events indicate that the daily changes of the related indicators are abnormal.

[0011] Based on the neighbor cell handover data of each cell, determine the neighbor cell related event relationship information corresponding to each event relationship information of each cell;

[0012] An event knowledge graph is generated based on the event relationship information of each cell and the corresponding event relationship information of neighboring cells.

[0013] Based on the event knowledge graph, other events related to the target event are queried as the root cause analysis results of the target event; the target event includes abnormal events that occur at a specified time.

[0014] A knowledge graph-based wireless network event root cause analysis device, comprising:

[0015] The network data acquisition unit is used to acquire historical operational data of the wireless network; the historical operational data includes various indicator data and neighbor cell handover data of multiple cells in the past time period; the various indicator data includes at least the change data of performance indicators and related indicators.

[0016] The event relationship determination unit is used to determine multiple event relationship information for each cell based on various indicator data of each cell and in conjunction with a predefined indicator threshold table; the indicator threshold table includes thresholds for various indicators; the event relationship information includes a central event and corresponding associated events; the central event indicates that the daily change of the performance indicator is abnormal; the associated event indicates that the daily change of the associated indicator is abnormal;

[0017] The neighbor cell event determination unit is used to determine the neighbor cell related event relationship information corresponding to each event relationship information of each cell based on the neighbor cell handover data of each cell;

[0018] The knowledge graph determination unit is used to generate an event knowledge graph based on the event relationship information of each cell and the corresponding neighboring cell related event relationship information;

[0019] The root cause analysis query unit is used to query other events related to the target event based on the event knowledge graph, as the root cause analysis result of the target event; the target event includes abnormal events that occur at a specified time.

[0020] A storage medium comprising a stored program, wherein the program is executed by a processor to perform the knowledge graph-based wireless network event root cause analysis method.

[0021] An electronic device includes: a processor, a memory, and a bus; the processor and the memory are connected via the bus.

[0022] The memory is used to store the program, and the processor is used to run the program, wherein the program is executed by the processor to perform the knowledge graph-based wireless network event root cause analysis method.

[0023] The technical solution provided in this application acquires historical operational data of a wireless network. Based on various indicator data of each cell and combined with a predefined indicator threshold table, multiple event relationship information for each cell is determined. Based on the neighbor cell handover data of each cell, the related neighbor cell event relationship information corresponding to each event relationship information of each cell is determined. Based on the event relationship information of each cell and the corresponding related neighbor cell event relationship information, an event knowledge graph is generated. Based on the event knowledge graph, other events related to the target event are queried as the root cause analysis results of the target event. This method utilizes an event knowledge graph to simplify the root cause analysis process of wireless network faults and effectively improve the accuracy and efficiency of the analysis. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 A flowchart illustrating a knowledge graph-based wireless network event root cause analysis method provided in this application embodiment;

[0026] Figure 2 A flowchart illustrating another knowledge graph-based wireless network event root cause analysis method provided in this application embodiment;

[0027] Figure 3 A flowchart illustrating another knowledge graph-based wireless network event root cause analysis method provided in this application embodiment;

[0028] Figure 4 A schematic diagram of the architecture of a knowledge graph-based wireless network event root cause analysis device provided in an embodiment of this application;

[0029] Figure 5A data flow diagram of an event knowledge graph provided in an embodiment of this application;

[0030] Figure 6 This is a schematic diagram of the root cause analysis process of an event knowledge graph provided in an embodiment of this application. Detailed Implementation

[0031] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0032] In this application, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. The terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0033] like Figure 1 The diagram shown is a flowchart of a knowledge graph-based wireless network event root cause analysis method provided in an embodiment of this application, including the following steps.

[0034] S101: Obtain historical operating data of the wireless network.

[0035] The historical operational data includes various indicator data and neighbor cell handover data that occurred in multiple cells over a period of time; the various indicator data include at least the change data of performance indicators and related indicators.

[0036] In some examples, the neighbor cell handover data includes at least multiple neighbor cell relationships, the number of handover requests corresponding to each neighbor cell relationship for which the cell is applying to handover, and the total number of handover requests for which the cell is applying to handover.

[0037] In some examples, there are multiple performance metrics, including but not limited to wireless connection success rate, wireless drop rate, handover success rate, and daily uplink interference average.

[0038] In some examples, related metrics include metrics that are correlated with performance metrics, such as other performance metrics, parameter metrics, and alarm metrics. Correspondingly, the change data for related metrics includes data from other performance metrics, parameter data, and alarm data.

[0039] In some examples, historical operational data of the wireless network can be retrieved from the system database.

[0040] S102: Based on various indicator data of each community and combined with a predefined indicator threshold table, determine the relationship information of multiple events in each community.

[0041] The indicator threshold table includes thresholds for various indicators; the event relationship information includes the central event and its corresponding related events; the central event indicates that the daily changes in performance indicators are abnormal; and the related events indicate that the daily changes in related indicators are abnormal.

[0042] In some examples, the event relationship information also includes preset event weights for related events, which can be determined based on the degree of influence of the related events on the performance indicators corresponding to the central event.

[0043] Optionally, the threshold includes absolute threshold and relative threshold; wherein, if only one of the current day's index value and the previous day's index value meets the absolute threshold of the performance index, or if the difference between the current day's index value and the previous day's index value exceeds the relative threshold of the performance index, the change in the performance index is determined to be a central event.

[0044] Optionally, the central event also includes a label, with the label type including excellent and poor. Specifically, if the indicator value of the day meets the absolute threshold of the performance indicator, and the indicator value of the day does not meet the absolute threshold of the performance indicator, a label of type excellent is added to the central event; if the indicator value of the day does not meet the absolute threshold of the performance indicator, and the indicator value of the day does meet the absolute threshold of the performance indicator, a label of type poor is added to the central event.

[0045] In some examples, defining absolute and relative thresholds for all performance metrics can be used to identify anomalous events.

[0046] In a possible implementation, the indicator threshold table can be found in Table 1.

[0047] Table 1

[0048]

[0049] It is understood that the content shown in Table 1 is only for illustrative purposes. The decrease and increase shown in the table can be understood as the change in the comparison result of the current day's indicator value and the previous day's indicator value with the absolute threshold. For example, if the wireless connection rate indicator value meets the absolute threshold yesterday but does not meet the absolute threshold today, the relative threshold can be recorded as a decrease. If the wireless connection rate indicator value does not meet the absolute threshold yesterday but meets the absolute threshold today, the relative threshold can be recorded as an increase.

[0050] In some examples, associated events include one or more of performance-relative events, parameter-relative events, and alarm-relative events.

[0051] In a possible implementation, for related events, if the change in the value of other performance metrics on the current day compared to the previous day exceeds a predefined threshold, a performance-related event is determined to have occurred. If the value of a parameter metric on the current day is inconsistent with the previous day's value, a parameter-related event is determined to have occurred. If an alarm metric is active on the current day, an alarm-related event is determined to have occurred.

[0052] In some examples, a central event labeled "Excellent" can be considered an excellent event, and its associated events and subsequent information about neighboring events can be considered solutions for the excellent event. Conversely, a central event labeled "Inferior" can be considered an inferior event, and its associated events and subsequent information about neighboring events can be considered the causes of the inferior event.

[0053] S103: Based on the neighbor cell handover data of each cell, determine the neighbor cell related event relationship information corresponding to each of the event relationship information of each cell.

[0054] Furthermore, based on the neighbor cell handover data of each cell, the weight of the neighbor cell related event corresponding to each of the event relationship information of each cell can be determined, so that users can intuitively determine the degree of influence of different neighbor cell related event relationship information on the central event.

[0055] Optionally, the process of determining the neighbor-related event relationship information corresponding to each event relationship information of each cell based on the neighbor cell handover data of each cell can be found in [reference needed]. Figure 2 The steps shown are accompanied by corresponding explanations.

[0056] It should be noted that the neighboring cell related event relationship information includes the event relationship information of the neighboring cells, which may specifically include the central event of the neighboring cell and its corresponding related events.

[0057] S104: Generate an event knowledge graph based on the event relationship information of each cell and the corresponding neighboring cell event relationship information.

[0058] Optionally, the event knowledge graph includes two types of nodes: entities and events. Entities include cells, and events include either central events or related events. The edges between entities and events are determined based on the attribution relationship between events and cells. The edges between entities are determined based on the neighboring cell relationships of cells and the scenario. The edges between events are determined based on event relationship information and neighboring cell related event relationship information.

[0059] In some examples, the event knowledge graph serves as a structured semantic knowledge base. Its basic building block is the "entity-relationship-entity" triple. Entities are interconnected through relations, forming a network-like knowledge structure. The event relationship information generated above, along with specific cells and neighboring cells, are considered entities. Relevance is used as the relationship between events, and attribution is used as the relationship between events and cells or neighboring cells. These interconnected events (or entities) form a wireless network of event knowledge graphs.

[0060] S105: Based on the event knowledge graph, query other events related to the target event as the root cause analysis results of the target event.

[0061] The target event includes abnormal events that occur at a specified time.

[0062] Optionally, based on the event knowledge graph, other events related to the target event can be queried as the root cause analysis results of the target event. For details on this process, please refer to [link to relevant documentation]. Figure 3 The steps shown are accompanied by corresponding explanations.

[0063] Combination Figures 2-3 The knowledge graph-based wireless network event root cause analysis method shown in this application embodiment has the following advantages: (1) Model flexibility: The event knowledge graph-based root cause analysis method can adapt to different scenarios and needs by adding or deleting elements such as entities, attributes, and relationships. This flexibility can make the model more adaptable and scalable; (2) Data richness: Event knowledge graphs can integrate data from different sources, including historical operation data, real-time data, log data, etc., which can make the model richer and more accurate; (3) Visualization: Event knowledge graphs can visualize relationships and structures, making the root cause analysis results more intuitive and easier to understand.

[0064] For wireless network event root cause analysis, the event knowledge graph-based method leverages the structured characteristics and rich semantic information of knowledge graphs to provide a novel approach to network fault identification and analysis. Event knowledge graphs can automate wireless network root cause analysis. A knowledge graph is a graph-based data structure composed of nodes and edges. Each node represents an entity, and each edge represents the relationship between entities; essentially, a knowledge graph is a semantic network. Event knowledge graphs are event-centric, including two types of nodes: events and entities, and three types of directed edges: event-to-event, event-to-entity, and entity-to-entity. Applying event knowledge graphs to wireless network fault root cause analysis greatly simplifies the process, increases accuracy, and enables timely response to fault occurrences.

[0065] In a possible implementation, the relative and absolute thresholds are defined using live wireless network data. Changes in performance metrics are defined as relative events, while alarms are defined as absolute events. Based on wireless network performance data, thresholds for abnormal events are set, and changes in performance metrics from good to deterioration are defined as relative events, thereby enabling dynamic monitoring and anomaly detection of network performance.

[0066] In a possible implementation, the construction of event relationship information involves establishing a one-to-one association between performance anomaly events within the cell and between neighboring cells and all parameter-relative events and alarm-absolute events, forming relationship pairs of performance events, parameter events, and alarm events between the cell and neighboring cells. Simultaneously, the proportion of neighboring cell handovers is used as a weight to further enhance the accuracy and effectiveness of event association.

[0067] In a possible implementation, an event knowledge graph for wireless network events is constructed based on root cause analysis using an event knowledge graph. By statistically analyzing event pairs with the same scenario and their cumulative counts, the quantity of each event pair is obtained, and then the proportion of each root cause corresponding to the abnormal event is calculated. This knowledge graph enables in-depth querying and root cause analysis of abnormal events, providing intelligent support for network fault management.

[0068] The processes shown in S101-S105 above utilize event knowledge graphs to simplify the root cause analysis process for wireless network faults, effectively improving the accuracy and efficiency of the analysis.

[0069] like Figure 2 The diagram shown is a flowchart of a knowledge graph-based wireless network event root cause analysis method provided in an embodiment of this application, including the following steps.

[0070] S201: For each cell, based on the cell's neighbor cell handover data, determine the neighbor cell weights of multiple neighbor cells that have a neighbor cell relationship with the cell.

[0071] The calculation process of the neighbor cell weight is shown in formula (1).

[0072] (1)

[0073] In formula (1), The neighbor weights represent the weights of a particular neighboring cell. This represents the number of handover requests made by the current cell to a neighboring cell. This represents the total number of handover requests for the cell.

[0074] S202: Identify multiple target neighboring cells from multiple neighboring cells that belong to the same scenario as the cell.

[0075] The scenarios for a community can be predefined by the wireless network operator, and the types of scenarios include, but are not limited to, subway scenarios, residential community scenarios, and transportation hub scenarios.

[0076] S203: Based on the event relationship information of each target neighbor cell and the cell, determine the same event relationship information of each target neighbor cell and the corresponding occurrence frequency.

[0077] The calculation process for the number of occurrences of the same event relationship information can be shown in formula (2).

[0078] (2)

[0079] In formula (2), The number of times information about the relationship between the same events occurs. Information representing the relationship between a certain event in the community The number of times it appears.

[0080] S204: Based on the same event relationship information of each target neighbor cell and the corresponding occurrence frequency, determine the neighbor cell related event relationship information and the relevance of the neighbor cell related event relationship information corresponding to each event relationship information of the cell.

[0081] The calculation process for the relevance of related event information in neighboring areas can be shown in formula (3).

[0082] (3)

[0083] In formula (3), The frequency of occurrence of information related to events in neighboring areas. This represents the relationship information of a certain event in the community. The number of occurrences in the event relationship information of each neighboring region The first representing the community The neighbor weights of each neighboring cell.

[0084] S205: Based on the relevance of the neighboring event relationship information of each neighboring cell, and combined with the neighboring cell weight of the neighboring cell to which each neighboring event relationship information belongs, determine the inter-neighboring event weight of each neighboring cell's related event relationship information.

[0085] The calculation process of the weight of related events in the neighboring interval can be seen in formulas (4), (5) and (6).

[0086] (4)

[0087] In formula (4), This represents the relevance of information regarding events in neighboring areas within the same residential community. This represents the number of times information related to a specific event occurs within the community. This represents the total number of times information related to a specific event appears within the community.

[0088] (5)

[0089] In formula (5), Represents the relevance of information related to events in neighboring areas. This represents the frequency with which information about related events in neighboring areas appears within the community. This represents the total number of occurrences of information related to events in neighboring areas.

[0090] (6)

[0091] In formula (6), Represents the weight of related events between neighboring intervals. Represents the relevance of information related to events in neighboring areas. Represents the weight of neighboring cells.

[0092] It is understandable that the appearance of related event information in neighboring areas within a residential community indicates that the same related event information has occurred in that community.

[0093] The processes shown in S201-S205 above can determine the neighbor cell related event relationship information corresponding to each event relationship information of each cell based on the neighbor cell handover data of each cell and the event relationship information of each cell.

[0094] like Figure 3 The diagram shown is a flowchart of a knowledge graph-based wireless network event root cause analysis method provided in an embodiment of this application, including the following steps.

[0095] S301: Based on the central event generated on the day from the event knowledge graph, it is identified as an abnormal event.

[0096] The number of abnormal events can be multiple.

[0097] In some examples, the data flow of the event knowledge graph can be seen in... Figure 5 As shown.

[0098] S302: Based on the abnormal event with the label "inferior", identify it as the target event and determine the target scenario of the target cell to which the target event belongs.

[0099] The event knowledge graph also includes the scenarios and tags for each event, which can be recorded in the corresponding node attributes.

[0100] S303: Based on the event knowledge graph, query other events related to the target event as the root cause analysis results of the target event.

[0101] Other events include related events and neighboring event relationship information corresponding to the target event, as well as related events corresponding to events belonging to the target scenario and labeled as inferior.

[0102] In a possible implementation, the process of using event knowledge graphs to perform root cause analysis of anomalous events can be summarized as follows: Figure 6 As shown.

[0103] In some examples, the analysis focuses on the performance "poor" events in the central cell. It combines abnormal events within and between the cell and neighboring cells with an event knowledge graph to analyze the root causes and probabilities of these performance anomalies. Root cause analysis can be summarized as follows: Based on the daily generated cell anomalies, query the event pairs within and between the cell and neighboring cells corresponding to the anomalies, and simultaneously confirm the target scenario to which the target cell belongs; query the corresponding poor events and their associated events in the event knowledge graph; if the associated events of the poor events found in the event knowledge graph also exist in the daily abnormal events, then the associated events are considered the root cause of the poor events queried on that day, and can be regarded as other events related to the anomalies (referred to as related events). The priority of related events is then calculated to distinguish the priority of different root causes.

[0104] In a possible implementation, the calculation process for the priority of relevant events can be found in formula (7).

[0105] (7)

[0106] In formula (7), Representing the The weight of each related event (e.g., the preset weight of associated events, the weight of related events between neighboring regions based on the relationship information of related events between neighboring regions). Representing the Custom weights for each abnormal event.

[0107] The processes shown in S301-S303 above can complete the root cause analysis of "bad" events in the community. Furthermore, root cause analysis based on knowledge graphs can effectively avoid the shortcomings of relying on reasoning models and establishing a correlation mapping between problems and root cause localization processes through advantages such as automation, flexibility, data richness, and visualization, thereby improving the efficiency and accuracy of root cause analysis.

[0108] like Figure 4 The diagram shown is an architectural schematic of a knowledge graph-based wireless network event root cause analysis device provided in an embodiment of this application, including the following units.

[0109] The network data acquisition unit 100 is used to acquire historical operating data of the wireless network. The historical operating data includes various indicator data of multiple cells in the past time period and neighbor cell handover data. The various indicator data includes at least the change data of performance indicators and related indicators.

[0110] The event relationship determination unit 200 is used to determine multiple event relationship information for each cell based on various indicator data of each cell and in combination with a predefined indicator threshold table. The indicator threshold table includes thresholds for various indicators. The event relationship information includes the central event and its corresponding related events. The central event indicates that the daily changes of performance indicators are abnormal. The related events indicate that the daily changes of related indicators are abnormal.

[0111] Optionally, the threshold includes absolute threshold and relative threshold; wherein, if only one of the current day's index value and the previous day's index value meets the absolute threshold of the performance index, or if the difference between the current day's index value and the previous day's index value exceeds the relative threshold of the performance index, the change in the performance index is determined to be a central event.

[0112] Optionally, the indicator threshold table also includes the superiority / inferiority judgment information corresponding to the relative thresholds of various indicators. The superiority / inferiority judgment information includes the judgment conditions for superior events and the judgment conditions for inferior events. Specifically, if the difference between the indicator value of the current day and the indicator value of the previous day meets the judgment conditions for superior events of the performance indicator, a label of type "superior" is added to the central event. If the difference between the indicator value of the current day and the indicator value of the previous day meets the judgment conditions for inferior events of the performance indicator, a label of type "inferior" is added to the central event.

[0113] The neighbor cell event determination unit 300 is used to determine the neighbor cell related event relationship information corresponding to each event relationship information of each cell based on the neighbor cell handover data of each cell.

[0114] Optionally, the neighbor cell event determination unit 300 is specifically used for: for each cell, determining the neighbor cell weights of multiple neighbor cells that have neighbor cell relationships with the cell based on the cell's neighbor cell handover data; determining multiple target neighbor cells belonging to the same scenario as the cell from the multiple neighbor cells; determining the same event relationship information and corresponding occurrence counts of each target neighbor cell based on the event relationship information of each target neighbor cell and the cell; determining the neighbor cell-related event relationship information and the relevance of the neighbor cell-related event relationship information corresponding to each event relationship information of the cell based on the same event relationship information and the corresponding occurrence counts of each target neighbor cell; and determining the inter-neighbor related event weights of each neighbor cell-related event relationship information based on the relevance of the neighbor cell-related event relationship information and the neighbor cell weights of the neighbor cells to which each neighbor cell-related event relationship information belongs.

[0115] The knowledge graph determination unit 400 is used to generate an event knowledge graph based on the event relationship information of each cell and the corresponding neighboring cell related event relationship information.

[0116] Optionally, the event knowledge graph includes two types of nodes: entities and events. Entities include cells, and events include either central events or related events. The edges between entities and events are determined based on the attribution relationship between events and cells. The edges between entities are determined based on the neighboring cell relationships of cells and the scenario. The edges between events are determined based on event relationship information and neighboring cell related event relationship information.

[0117] The root cause analysis query unit 500 is used to query other events related to the target event based on the event knowledge graph, as the root cause analysis result of the target event; the target event includes abnormal events that occur at a specified time.

[0118] Optionally, the root cause analysis query unit 500 is specifically used for: identifying a central event generated on the same day based on the event knowledge graph as an anomalous event; identifying an anomalous event labeled as inferior as a target event, and determining the target scenario of the target cell to which the target event belongs; querying other events related to the target event based on the event knowledge graph, as the root cause analysis result of the target event; other events include related events corresponding to the target event and related events in neighboring cells, as well as related events corresponding to events belonging to the target scenario and labeled as inferior.

[0119] The units described above utilize event knowledge graphs to simplify the root cause analysis process for wireless network faults, effectively improving the accuracy and efficiency of the analysis.

[0120] This application also provides a computer-readable storage medium including a stored program, wherein the program executes the knowledge graph-based wireless network event root cause analysis method provided in this application.

[0121] This application also provides an electronic device, including a processor, a memory, and a bus. The processor and the memory are connected via the bus. The memory is used to store a program, and the processor is used to run the program. During program execution, the knowledge graph-based wireless network event root cause analysis method provided in this application is executed.

[0122] While several specific implementation details are included in the foregoing discussion, these should not be construed as limiting the scope of this application. Certain features described in the context of individual embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented individually or in any suitable sub-combination in multiple embodiments.

[0123] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this application.

Claims

1. A knowledge graph-based method for root cause analysis of wireless network events, characterized in that, include: Acquire historical operational data of the wireless network; the historical operational data includes various indicator data and neighbor cell handover data of multiple cells over a period of time. The various indicator data include at least the change data of performance indicators and related indicators; Based on various indicator data of each cell, and in conjunction with a predefined indicator threshold table, multiple event relationship information for each cell is determined; the indicator threshold table includes thresholds for various indicators; the event relationship information includes a central event and its corresponding associated events; the central event indicates that the daily changes of the performance indicators are abnormal; The associated event indicates that the daily changes of the associated indicator are abnormal; Based on the neighbor cell handover data of each cell, determine the neighbor cell related event relationship information corresponding to each event relationship information of each cell; An event knowledge graph is generated based on the event relationship information of each cell and the corresponding event relationship information of neighboring cells. Based on the event knowledge graph, other events related to the target event are queried as the root cause analysis results of the target event; the target event includes abnormal events that occur at a specified time.

2. The method according to claim 1, characterized in that, The thresholds include absolute thresholds and relative thresholds; Specifically, if only one of the current day's indicator value and the previous day's indicator value meets the absolute threshold of the performance indicator, or if the difference between the current day's indicator value and the previous day's indicator value exceeds the relative threshold of the performance indicator, then the change in the performance indicator data is determined to have occurred as the central event.

3. The method according to claim 2, characterized in that, The central event also includes labels, and the types of labels include good and bad; If the current day's indicator value meets the absolute threshold of the performance indicator, and the previous day's indicator value does not meet the absolute threshold of the performance indicator, a label of type "excellent" is added to the central event. If the current day's indicator value does not meet the absolute threshold of the performance indicator, and the previous day's indicator value does meet the absolute threshold of the performance indicator, add a label of type "inferior" to the central event.

4. The method according to claim 1, characterized in that, Based on the neighbor cell handover data of each cell, determine the neighbor cell related event relationship information corresponding to each event relationship information of each cell, including: For each cell, based on the cell’s neighbor cell handover data, the neighbor cell weights of multiple neighbor cells that have a neighbor cell relationship with the cell are determined; From the multiple neighboring cells, identify multiple target neighboring cells that belong to the same scenario as the cell; Based on the event relationship information of each target neighbor cell and each cell, determine the same event relationship information and the corresponding occurrence count of each target neighbor cell; Based on the same event relationship information of each target neighbor cell and the corresponding occurrence frequency, determine the neighbor cell related event relationship information and the relevance of the neighbor cell related event information corresponding to each event relationship information of the cell; Based on the relevance of the neighboring event relationship information for each neighboring cell, and combined with the neighboring cell weight of the neighboring cell to which each neighboring event relationship information belongs, the inter-neighboring event weight of each neighboring event relationship information is determined.

5. The method according to claim 1, characterized in that, The event knowledge graph includes two types of nodes: entities and events. Entities include cells, and events include either central events or related events. The edges between entities and events are determined based on the attribution relationship between the event and the cell. The edges between entities are also determined based on the neighboring cell relationships and the context of the cell. The edges between events are determined based on the event relationship information and the neighboring cell related event relationship information.

6. The method according to claim 1, characterized in that, Based on the event knowledge graph, other events related to the target event are queried as the root cause analysis results of the target event, including: Based on the central events generated on that day using the event knowledge graph, these events were identified as anomalous events. Based on the abnormal events labeled as "inferior", the target events are identified as target events, and the target scenario of the target cell to which the target event belongs is determined; Based on the event knowledge graph, other events related to the target event are queried as the root cause analysis results of the target event; the other events include the related events and neighboring related events corresponding to the target event, as well as the related events corresponding to the events belonging to the target scenario and labeled as inferior.

7. A knowledge graph-based wireless network event root cause analysis device, characterized in that, include: The network data acquisition unit is used to acquire historical operational data of the wireless network; the historical operational data includes various indicator data and neighbor cell handover data of multiple cells in the past time period; the various indicator data includes at least the change data of performance indicators and related indicators. The event relationship determination unit is used to determine multiple event relationship information for each cell based on various indicator data of each cell and in conjunction with a predefined indicator threshold table; the indicator threshold table includes thresholds for various indicators; the event relationship information includes a central event and corresponding related events; the central event indicates that the daily changes of the performance indicators are abnormal; The associated event indicates that the daily changes of the associated indicator are abnormal; The neighbor cell event determination unit is used to determine the neighbor cell related event relationship information corresponding to each event relationship information of each cell based on the neighbor cell handover data of each cell; The knowledge graph determination unit is used to generate an event knowledge graph based on the event relationship information of each cell and the corresponding neighboring cell related event relationship information; The root cause analysis query unit is used to query other events related to the target event based on the event knowledge graph, as the root cause analysis result of the target event; the target event includes abnormal events that occur at a specified time.

8. The apparatus according to claim 7, characterized in that, The neighboring cell event determination unit is specifically used for: For each cell, based on the cell’s neighbor cell handover data, the neighbor cell weights of multiple neighbor cells that have a neighbor cell relationship with the cell are determined; From the multiple neighboring cells, identify multiple target neighboring cells that belong to the same scenario as the cell; Based on the event relationship information of each target neighbor cell and each cell, determine the same event relationship information and the corresponding occurrence count of each target neighbor cell; Based on the same event relationship information of each target neighbor cell and the corresponding occurrence frequency, determine the neighbor cell related event relationship information and the relevance of the neighbor cell related event information corresponding to each event relationship information of the cell; Based on the relevance of the neighboring event relationship information for each neighboring cell, and combined with the neighboring cell weight of the neighboring cell to which each neighboring event relationship information belongs, the inter-neighboring event weight of each neighboring event relationship information is determined.

9. A storage medium, characterized in that, The storage medium includes a stored program, wherein the program is executed by a processor to perform the knowledge graph-based wireless network event root cause analysis method according to any one of claims 1-6.

10. An electronic device, characterized in that, include: Processor, memory, and bus; The processor and the memory are connected via the bus; The memory is used to store a program, and the processor is used to run the program, wherein the program is executed by the processor to perform the knowledge graph-based wireless network event root cause analysis method according to any one of claims 1-6.

Images