A core network fault positioning method, device, equipment, storage medium and product

Abstract

This application discloses a core network fault location method, apparatus, device, storage medium, and product, belonging to the field of communications, for accurately locating core network faults. The method includes: acquiring a fault cause code and first extended call detail record (XDR) data; determining first network element information and first embedded feature vectors for multiple first network elements based on the first XDR data; grouping the multiple first network elements based on the first network element information to obtain multiple first network element groups, and determining a first group feature vector and a first group feature vector within each first network element based on the first embedded feature vectors; determining anomaly group feature vectors and anomaly group feature vectors within the multiple first group feature vectors based on anomaly vector determination strategies, so as to locate the fault based on the fault cause code, the anomaly group feature vectors, and the anomaly group feature vectors within the multiple first group feature vectors.

Description

Technical Field

[0001] This application belongs to the field of communications, and specifically relates to a core network fault location method, apparatus, equipment, storage medium, and product. Background Technology

[0002] The mobile core network is a crucial component of a mobile communication system. It carries user data and control signaling and provides interconnection capabilities with other networks. The core network consists of multiple functional modules, including user data management, session management, and mobility management. Its main functions include data transmission, session management, mobility management, user data management, signaling processing, and billing. The mobile core network is the central hub of the mobile communication system, coordinating communication and interaction between various subsystems to provide users with seamless connectivity and continuous communication services. Currently, domestic mobile core networks include network elements from 2G, 4G, 5G, and IMS network domains. The network structure is vast and complex. Failure to accurately detect core network element faults during network operation and maintenance can lead to operational accidents, impacting the user experience for a large number of users.

[0003] Therefore, a method is needed to accurately locate faults in the core network. Summary of the Invention

[0004] This application provides a core network fault location method that can accurately locate faults in the core network.

[0005] In a first aspect, embodiments of this application provide a core network fault location method, the method comprising: acquiring a fault cause code and first extended call detail record (XDR) data, wherein the fault cause code is a cause code generated when a fault occurs in the target core network, the target core network is the core network to be fault located, the first XDR data is XDR data generated by the target core network, and the difference between the generation time of the first XDR data and the generation time of the fault cause code is less than a first threshold; determining first network element information and a first embedded feature vector of multiple first network elements based on the first XDR data, wherein the first network element is the network element involved in the first XDR data, the first network element information is the relevant information of the first network element, and the first embedded feature vector is... The embedded feature vector of the first network element; based on the information of the first network element, multiple first network elements are grouped to obtain multiple first network element groups, and the first group feature vector and the first group feature vector of each first network element are determined based on the first embedded feature vector. The first group feature vector is the feature vector of the first network element group, and the first group feature vector is the feature vector of the first network element under the constraint of its corresponding first network element group; based on the anomaly vector determination strategy, the anomaly group feature vector and the anomaly group feature vector in the multiple first group feature vectors are determined, so as to perform fault location according to the fault cause code, the anomaly group feature vector and the anomaly group feature vector.

[0006] Secondly, embodiments of this application provide a core network fault location device, comprising: a first acquisition module, configured to acquire a fault cause code and first extended call detail record (XDR) data, wherein the fault cause code is a cause code generated when a fault occurs in the target core network, the target core network is the core network to be fault located, the first XDR data is XDR data generated by the target core network, and the difference between the generation time of the first XDR data and the generation time of the fault cause code is less than a first threshold; and a first determination module, configured to determine first network element information and a first embedded feature vector of multiple first network elements based on the first XDR data, wherein the first network element is the network element involved in the first XDR data, the first network element information is the relevant information of the first network element, and the first embedded feature vector is... The first network element's embedded feature vector; the second determining module, configured to group multiple first network elements based on the first network element information to obtain multiple first network element groups, and to determine the first group feature vector of each first network element group and the first group feature vector of each first network element based on the first embedded feature vector, wherein the first group feature vector is the feature vector of the first network element group, and the first group feature vector is the feature vector of the first network element under the constraint of its corresponding first network element group; the first positioning module, configured to determine the abnormal group feature vector in multiple first group feature vectors and the abnormal group feature vector in multiple first group feature vectors based on the abnormal vector determination strategy, and to perform fault positioning based on the fault cause code, the abnormal group feature vector, and the abnormal group feature vector.

[0007] Thirdly, embodiments of this application provide an electronic device including a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the steps of the method described in the first aspect.

[0008] Fourthly, embodiments of this application provide a readable storage medium on which a program or instructions are stored, which, when executed by a processor, implement the steps of the method described in the first aspect.

[0009] Fifthly, embodiments of this application provide a computer program product that, when executed by a processor, implements the steps of the method described in the first aspect.

[0010] In a sixth aspect, embodiments of this application provide a chip, the chip including a processor and a communication interface, the communication interface being coupled to the processor, the processor being used to run programs or instructions to implement the method as described in the first aspect.

[0011] In this embodiment, by acquiring the fault cause code and the first extended call detail record (XDR) data; determining the first network element information and the first embedded feature vector of multiple first network elements based on the first XDR data; grouping the multiple first network elements based on the first network element information to obtain multiple first network element groups; determining the first group feature vector and the first group feature vector of each first network element based on the first embedded feature vector; and determining the abnormal group feature vector and the abnormal group feature vector in the multiple first group feature vectors based on the abnormal vector determination strategy, so as to accurately locate the fault in the core network by performing fault location based on the fault cause code, abnormal group feature vector, and abnormal group feature vector. Attached Figure Description

[0012] Figure 1 This is a flowchart illustrating a core network fault location method provided in an embodiment of this application; Figure 2 This is a schematic diagram of the embedded feature vector representation of a first network element provided in an embodiment of this application; Figure 3 This is a flowchart illustrating the second core network fault location method provided in the embodiments of this application; Figure 4 This is a flowchart illustrating the third core network fault location method provided in the embodiments of this application; Figure 5 This is a schematic diagram of the structure of a core network fault location system provided in an embodiment of this application; Figure 6 This is a schematic diagram of the structure of a core network fault location device provided in an embodiment of this application; Figure 7 This is a schematic diagram of the structure of a core network fault location device provided in an embodiment of this application. Detailed Implementation

[0013] 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, 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.

[0014] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0015] The core network fault location method provided in this application will be described in detail below with reference to the accompanying drawings, through specific embodiments and application scenarios.

[0016] Figure 1 This illustration shows a core network fault location method according to an embodiment of the present invention. The method can be executed by an electronic device, which may include a server and / or a terminal device, wherein the terminal device may be, for example, a vehicle-mounted terminal or a mobile phone terminal. In other words, the method can be executed by software or hardware installed in a core network fault location device, and the method includes the following steps: Step 102: Obtain the fault cause code and the first extended call detail record (XDR) data.

[0017] Wherein, the fault cause code is the cause code generated when a fault occurs in the target core network, the target core network is the core network to be located for fault, the first XDR data is the XDR data generated by the target core network, and the difference between the generation time of the first XDR data and the generation time of the fault generation code is less than a first threshold.

[0018] The execution entity of the core network fault location method described in this application can be a core network fault location system, core network fault location software, or other execution entities. This application embodiment will use a core network fault location system (hereinafter referred to as fault location system) as an example for illustration.

[0019] The fault location system obtains the fault cause code of the target core network. The target core network is the core network that has experienced a fault and needs to be located. The fault cause code is the cause code generated when the target core network experiences a fault. For example, the fault cause code can be "101: N1_SM_ERROR".

[0020] Furthermore, the fault location system interfaces with external anomaly monitoring interfaces (error cause codes / performance degradation) to detect anomalies in the target core network in real time. The monitoring target can be fault cause codes or performance degradation alarm data. Taking error cause code anomaly monitoring as an example, the fault cause code information corresponding to the acquired XDR data is generally as follows: For example, the fault cause codes for the N11 interface include: 102: SNSSAI_DENIED; 103: DNN_DENIED; 104: PDUTYPE_DENIED; 105: SSC_DENIED, etc.

[0021] The fault location system can also use performance degradation indicators as monitoring targets, such as the number of times response latency exceeds a threshold. The fault location system counts the above error cause codes or the number of times performance degradation occurs, such as the frequency of user equipment unreachable fault codes, the frequency of missing information element fault codes, the frequency of service response time exceeding a threshold, etc., and sets trigger thresholds for various indicators to trigger degradation alarms, thereby determining the cause of the fault.

[0022] The fault location system not only obtains the fault cause code, but also the first XDR data. The first XDR data is the XDR data generated during the operation of the target core network. The obtained XDR data includes not only normal XDR data generated during the normal operation of the target core network, but also abnormal XDR data generated during the abnormal operation of the target core network. This XDR data is the signaling call detail record collected between the interfaces of various network elements in the core network. It can be extended to various call detail record records such as internal equipment documents and billing call detail records. Among them, the signaling call detail record and other document records collected between the interfaces of various network elements in the core network can be used to identify the service flow and network elements.

[0023] Specifically, the first XDR data consists of XDR data generated by multiple network elements (e.g., AMF, SMF, UPF, DRA, AS, AUSF, UDM, PCF, MME, SAEGW, etc.) included in the target core network. Therefore, the fault location system can obtain XDR data from various interfaces between network elements (e.g., N1 / N2, N11, N3, N6, S1-MME, S6a, S10&S11-C, S5&S8-C, SGs, Gn-c, Volte-SIP, etc.). The format of the obtained XDR data can be as follows (taking the XDR data of the N11 interface as an example): the core fields include MSISDN (user identifier, used to distinguish users), Procedure Type (procedure type code, used to extract the procedure name), Interface (interface type), Procedure Status (procedure status), AMF Address (AMF network element address), SMF Address (SMF network element address), and Failure Cause (rejection reason).

[0024] The difference between the generation time of the first XDR data acquired by the fault location system and the generation time of the fault cause code is less than a first threshold. In other words, the difference between the generation time of the first XDR data and the generation time of the fault cause code is the first threshold, which is a preset value that can be 1 minute or 5 minutes. Therefore, the first XDR data acquired by the fault location system is XDR data within a preset time range of the fault occurrence time (the generation time of the fault cause code). For example, when a monitoring alarm occurs, the fault location system automatically acquires XDR data from all interfaces within a 5-minute range before and after the alarm occurrence time, or integrates the XDR set as the first XDR data. Furthermore, when the generation time of the fault cause code is not a specific point in time, but a time period, the first threshold can be 0, meaning the generation time period of the first XDR data is the same as the generation time period of the fault cause code. For example, a sudden increase in fault cause codes within a statistical time period, such as the statistical period 15:00-15:05. For the N1N2 interface registration process, error code 22 has increased. Therefore, the signaling XDR data related to the registration process within the time range of 15:00-15:05 is extracted for location. In other words, when the fault cause code generation time is a specific point in time, the fault location system can identify XDR data whose difference from the fault cause code generation time is less than a first threshold (e.g., five minutes) as the first XDR data; when the fault cause code generation time is a time period, the fault location system can identify XDR data whose difference from the fault cause code generation time is 0 (the first threshold is 0) as the first XDR data.

[0025] Furthermore, the fault location system acquires XDR data in real time or synthesized XDR data in real time through other means. After acquiring the abnormal cause code, the XDR data whose generation time difference with the abnormal cause code is within a first threshold is determined as the first XDR data.

[0026] Step 104: Determine the first network element information and the first embedded feature vector of multiple first network elements based on the first XDR data.

[0027] Wherein, the first network element is the network element involved in the first XDR data, the first network element information is the relevant information of the first network element, and the first embedded feature vector is the embedded feature vector of the first network element.

[0028] After acquiring the first XDR data, the fault location system determines multiple first network elements included in the target core network based on the first XDR data, and determines the first network element information and the first embedded feature vector of each first network element based on the first XDR data. The first network element information is the relevant information of the first network element, such as the IP address, type information, name information, etc. of the first network element, and the first embedded feature vector is the embedded feature vector of the first network element.

[0029] In determining the first network element information of the first network element in the first XDR data, the fault location system first obtains the network element information in the first XDR data, which includes the IP address and network element type information of the first network element; then, based on the network element information, it obtains the name information corresponding to each network element from the preset network element name management database; and finally, based on the network element information and its corresponding name information, it determines the first network element information of the first network element.

[0030] Specifically, when determining network element information in XDR data, the fault location system obtains network element information (interface / protocol / process name / source and destination IPs) and removes duplicates. XDR data typically contains the following network element-related fields. By parsing these fields, the network element name can be obtained indirectly or directly. Information is extracted and deduplicated: network element IP address is extracted (from the sourceIP and destinationIP fields of the call detail record); network element type is extracted (from the interface identifier: interface (N2 / N4 / N6, etc.) + process type (for example, the Initial UE Msg of the N2 interface can determine that the network element corresponding to the target IP address is AMF).

[0031] When determining the name information of the first network element, the fault location system queries the obtained interface, protocol, process type, and IP information in the network element name management database, network element management system (EMS / NMS), or other supporting systems to obtain the corresponding network element name. Generally, the network element management systems of network elements from different manufacturers are not the same. If the core network consists of multiple manufacturers, a round-robin approach can be used to query multiple network element management systems (EMS / NMS) or network element name management databases. Finally, the first network element information is determined based on the queried network element type, network element name, and IP address.

[0032] When determining the first embedded feature vector of the first network element, the fault localization system inputs the information of each first network element into the pre-trained embedded feature vector determination model to obtain the first embedded feature vector of each first network element output by the model. The embedded feature vector determination model can be Word2Vec-SkipGram, Word2Vec-CBOW, GloVe, etc.

[0033] Specifically, the fault location system can obtain a vectorized representation of each word through a word embedding algorithm. The word embedding algorithm can be one of Word2Vec, GloVe, FastText, ELMo, or BERT. For example, the vector representation of an AMF network element "AMF:10.23.4.3" is [0.21,0.24,0.92,0.43,...,0.10,021]. Generally, a 64-dimensional or 128-dimensional vector representation is used to accommodate sufficient topological information.

[0034] Step 106: Group the multiple first network elements based on the first network element information to obtain multiple first network element groups, and determine the first group feature vector of each first network element group and the first group feature vector of each first network element based on the first embedded feature vector.

[0035] Wherein, the first set of feature vectors are the feature vectors of the first network element group, and the feature vectors within the first set are the feature vectors of the first network element under the constraints of its corresponding first network element group.

[0036] After determining the first network element information of each first network element, the fault location system classifies multiple first network elements based on this information, resulting in multiple first network element groups and multiple first network elements included in each group. During classification, the fault location system can classify each first network element according to a preset grouping strategy, thereby obtaining multiple first network element groups. The preset grouping strategy can be a strategy that classifies based on a single piece of information in the first network element information (e.g., grouping multiple first network elements based on their type information), or it can be a strategy that classifies based on multiple pieces of information in the first network element information (e.g., grouping multiple first network elements based on both their type information and IP information).

[0037] Specifically, when grouping multiple first network elements, the fault location system first obtains the type information of each first network element and determines its type, such as AMF network element, SMF network element, etc. Then, it groups the first network elements according to their type, identifying network elements of the same type as the same group. For example, multiple AMF network elements are identified as the same group, and multiple SMF network elements are identified as the same group, thus identifying multiple first network element groups. Furthermore, the fault location system can also classify the first network elements based on their attribute information (such as manufacturer) and type information, that is, identifying network elements of the same type produced by the same manufacturer as the same group. For example, identifying AMF network elements produced by manufacturer A as the same group, and AMF network elements produced by manufacturer B as the same group; it can also classify the first network elements based on their attribute information (such as manufacturer and location information) and type information, that is, identifying network elements of the same type produced by the same manufacturer in the same area as the same group.

[0038] After obtaining multiple first network element groups, the fault location system determines the first set of feature vectors for each first network element group and the first set of feature vectors within each first network element group based on the first embedded feature vectors corresponding to each first network element. The first set of feature vectors represents the feature vectors of the first group, and the first set of feature vectors within each first group represents the feature vectors of the first network element under the constraints of its corresponding first network element group. In other words, the first set of feature vectors represents the feature vectors of each first group, while the first set of feature vectors within each first group is determined based on the first set of feature vectors of the first network element group to which the first network element belongs and the relative feature vectors to those first feature vectors. Therefore, the first set of feature vectors within each first group represents the feature vectors of the first network element based on the first set of feature vectors and the relative feature vectors.

[0039] Specifically, when determining the first set of feature vectors for the first network element group, the fault location system can determine the average of the first embedded feature vectors of the multiple first network elements included in the first network element group as the first set of feature vectors. The formula for calculating the first set of feature vectors can be: V = ∑V i / m, where V represents the first set of feature vectors of the first network element group, Vi represents the first embedded feature vector of the i-th first network element in the first network element group, and m is the number of first network elements in the first network element group. Therefore, the first set of feature vectors of each first network element in the first network element group can be expressed as: V b =V+v b , where V b V represents the first feature vector within the first group of the b-th first network element in the first network element group, and V represents the first feature vector of the first network element group. b This represents the relative eigenvector of the first network element relative to the first set of eigenvectors. Therefore, it can be seen that the eigenvector within the first set of eigenvectors of any first network element in the target core network can be represented as V. ab =Va +v ab , where V ab V represents the feature vector within the first group of the b-th first element in the a-th first element group. a v represents the first feature vector of the a-th first network element. ab This represents the relative feature vector of the b-th first network element in the a-th first network element group.

[0040] More specifically, after determining the first set of feature vectors for each first network element group and the feature vectors within the first set for each first network element, the fault location system performs a core network element topology construction operation based on the feature vectors within the first set for each first network element to obtain a core network topology diagram. When constructing the topology diagram, the fault location system can determine multiple service flow network element sequences based on XDR data, and then generate the core network topology diagram based on the service flow network element sequences and the feature vectors within the first set for each first network element. The service flow network element sequences represent the order in which data flows between service network elements when users perform multiple services in the core network. This core network topology diagram includes not only the interaction relationships between various service network elements, but also relevant information between two service network elements of the same type, i.e., whether two service network elements of the same type are of the same type (determined to be of the same type when located in the same first network element group).

[0041] Step 108: Based on the anomaly vector determination strategy, determine the anomaly group feature vector in the multiple first group feature vectors and the anomaly group feature vector in the multiple first group feature vectors, so as to locate the fault according to the fault cause code, the anomaly group feature vector and the anomaly group feature vector.

[0042] After determining the first set of feature vectors for each first network element group and the feature vectors within the first set for each first network element, the fault location system determines the abnormal vectors in multiple first set feature vectors based on the abnormal vector determination strategy, and uses the determined abnormal vectors as the feature vectors of the abnormal group, and determines the abnormal vectors in multiple first set feature vectors, and uses the determined abnormal vectors as the feature vectors within the abnormal group.

[0043] Anomaly vector determination strategies can be based on a pre-trained anomaly vector determination model to determine anomaly group feature vectors and anomaly group feature vectors, or other anomaly vector determination methods. When determining anomaly group feature vectors and anomaly group feature vectors based on the anomaly vector determination model, the fault location system first acquires historical group feature vectors and historical group feature vectors. These historical group feature vectors and historical group feature vectors are feature vectors of historical network elements and historical network elements determined based on historical XDR data of the target core network. Each determined historical group feature vector and historical group feature vector has corresponding label information, which is used to characterize whether the historical group feature vector or historical group feature vector is an anomaly vector. Then, the fault location system trains the model based on the historical group feature vectors and historical group feature vectors respectively, so that the pre-trained model learns the normal and anomaly feature vectors in the historical group feature vectors, and the normal and anomaly feature vectors in the historical group feature vectors. Finally, the pre-trained model determines the anomaly group feature vectors in multiple first group feature vectors and the anomaly group feature vectors in multiple first group feature vectors. Furthermore, when determining abnormal feature vectors, the fault location system can also perform vectorization and matrix construction on the abnormal sample XDR data to obtain the sample abnormal feature vectors through a dimensionality reduction strategy. Then, the dimensionality reduction strategy is used to reduce the dimensionality of the first group of feature vectors and the feature vectors within the first group to determine the abnormal group feature vectors and the feature vectors within the abnormal group.

[0044] After determining the anomaly group feature vector and the feature vector within the anomaly group, the fault location system performs fault location based on the anomaly group feature vector, the feature vector within the anomaly group, and the anomaly cause code. Further, the fault location system determines the corresponding first network element group based on the anomaly group feature vector and the corresponding first network element based on the anomaly group feature vector. That is, the determined first network element group and first network element are the network element group and network element with anomalies, and then the fault location is performed based on the determined network element group with anomalies, the network element with anomalies, and the fault cause code.

[0045] More specifically, after determining the anomaly group feature vector, the fault location system can identify the first network element group with the anomaly. That is, the fault location system identifies the first network element group corresponding to the anomaly group feature vector as the faulty network element group. In other words, when performing fault location, the fault location system can first determine the anomaly group feature vector, then determine the first network element group with the anomaly based on the anomaly group feature vector. For example, it can determine which type of network element group is faulty, which manufacturer's network element group is faulty, and which type of network element group in which region (or location) is faulty. Finally, it performs specific fault location, for example, determining which specific network element is faulty. Therefore, when performing fault location, the fault location system can not only locate the specific faulty network element, but also the problematic network element group, thus identifying not only the faulty network element, but also the network element group that is prone to failure (i.e., determining which type of network element group is more likely to fail).

[0046] The core network fault location method provided in this invention obtains fault cause codes and first extended call detail records (XDR) data; determines first network element information and first embedded feature vectors for multiple first network elements based on the first XDR data; groups the multiple first network elements based on the first network element information to obtain multiple first network element groups; determines the first group feature vector and the first group feature vector within each first network element based on the first embedded feature vector; and determines the abnormal group feature vector and the abnormal group feature vector within each first group feature vector based on an anomaly vector determination strategy. This allows for fault location based on the fault cause code, abnormal group feature vector, and abnormal group feature vector, enabling accurate fault location in the core network.

[0047] In one implementation, the step of determining the first network element information and the first embedded feature vector of multiple first network elements based on the first XDR data (step 104) can be performed via steps A1-A2: Step A1: Determine the first service flow sequence based on the first XDR data.

[0048] Wherein, the first service flow sequence is the order in which data flows among the first network elements when the target core network carries the service.

[0049] After acquiring the first XDR data, the fault location system determines the first service flow sequence based on the first XDR data. The first service flow sequence is the order in which data flows between first network elements when the target core network carries services. For example, part of the first service flow sequence can be AMF network element-SMF network element, which means that data flows from the AMF network element to the SMF network element. That is, when the target core network carries services, data processing is first performed in the AMF network element and then in the SMF network element.

[0050] Each service process (attachment process, PDU session establishment process, call process) in the target core network is associated with a large number of first service flow timing sequences. Specifically, when determining the first service flow timing sequence, the fault location system first acquires the 3GPP protocol and first XDR data related to 4G and 5G, and then reconstructs the corresponding service flow timing sequence between network elements based on the 3GPP protocol (3rd Generation Partnership Project) and the first XDR data. Furthermore, if the first XDR data acquired by the fault location system is composite XDR data, then the first service flow timing sequence is directly reconstructed; if the fault location system acquires the first XDR data of a sub-interface, then the multi-interface XDRs are concatenated and composited using information such as MSISND and XDR time, and then the service flow timing sequence is reconstructed (based on the 3GPP protocol and based on the Procedure Type encoding). The first service flow sequence generated by the fault location system can be: UE->gNB->AMF->SMF->PCF->SMF->AMF->gNB->AMF->SMF->AMF->UE; UE->gNB->AMF->SMF->UDM->SMF->AMF->gNB->UE->gNB->AMF->SMF->AMF.

[0051] Step A2: Determine the service flow feature vector corresponding to the first service flow sequence, and determine the first embedded feature vector of each of the first network elements included in the first service flow sequence based on the service flow feature vector.

[0052] The business flow feature vector is the embedded feature vector corresponding to the first business flow sequence.

[0053] After determining the sequence of multiple first service flows, the fault location system determines the service flow feature vector corresponding to each first service flow sequence. This service flow feature vector is the embedded feature vector corresponding to the first service flow sequence. In other words, the fault location system determines the embedded feature vector for each first service flow sequence.

[0054] After determining the embedded feature vectors of each first service flow sequence, the fault location system determines the embedded feature vectors of each first network element included in the first service flow sequence based on the service flow feature vectors, and sets the determined embedded feature vectors as the first embedded feature vectors. In other words, the fault location system treats the multiple first network elements included in the first service flow sequence as a whole, determines the overall embedded feature vector, and then determines the embedded feature vectors of each first network element based on the overall embedded feature vector.

[0055] Specifically, in the first service flow sequence, each first network element is sorted in order. Therefore, in the determined service flow feature vector, the embedded features corresponding to each first network element are also arranged in order. For example, when the first service flow sequence is AMF-SMF, the determined service flow feature vector is also the embedded feature vector corresponding to AMF - the embedded feature vector corresponding to SMF.

[0056] The embedded feature vector determined by the fault location system is not only a unique identifier for network elements, but also contains the topological relationships between network elements. Generally, the vector representations of network elements in the same resource pool, the same group of network elements, and directly connected network elements all exhibit correlation and proximity in specific dimensions. For example... Figure 2 The diagram shows the embedded feature vector representation of the first network element (a schematic diagram of a 2D vector representation of a network element). It can be seen that there are significant differences between AMF, PCF, and SMF type network elements, while the differences within the same type of network element are relatively small. Furthermore, when determining the first network element group, the fault location system can generate a topology diagram based on the service network element flow sequence and the embedded feature vector representation diagram. This generated topology diagram not only includes the network element flow order of different services but also highlights the correlation between different network elements (e.g., determining...). Figure 2 AMF1, AMF2, and AMF3 network elements belong to the same network element group, while AMF3 and AMF5 network elements do not belong to the same network element group.

[0057] In this embodiment, by determining the first service flow sequence and determining the embedded feature vector of each first network element based on the first service flow sequence, the first embedded feature vector of each first network element generated in this way can not only characterize the features of each first network element, but also show the feature relationship between each first network element, so as to determine the first group feature vector and the feature vector within the first group based on grouping multiple first network elements.

[0058] In one implementation, determining the first service flow sequence based on the first XDR data (step A1) can be performed via steps B1-B3: Step B1: Obtain user attribute information.

[0059] The user attribute information refers to the attribute information of a first user, who is a user using services in the target core network.

[0060] When determining the sequence of the first service flow, the fault location system first obtains user attribute information, which is the attribute information of the first user, who is the user using the service in the target core network. In other words, the fault location system obtains the attribute information, such as ID information, of each user using the service in the target core network.

[0061] Step B2: Determine the user service flow sequence based on the first XDR data and the user attribute information.

[0062] The user service flow sequence refers to the order in which data flows between the first network elements when the target core network carries the service of the first user.

[0063] After determining the user attribute information of each user, the fault location system determines the user service flow sequence based on the first XDR data and the user attribute information. The user service flow sequence is the order in which data flows between the first network elements when the target core network carries the first user's service.

[0064] The fault location system can add the address information of the first network element carrying the first user service to each first network element in the first service flow sequence based on the first service flow sequence, so as to determine the user service flow sequence. For example, when the first service flow sequence is AMF-SMF-PCF, the user service flow sequence can be obtained as AMF:10.23.4.3-SMF:10.21.4.5-PCF:10.25.24.83.

[0065] Specifically, the fault location system can use the MSISDN (user's mobile phone number) and the network element information (network element type, IP address) in the first XDR data to substitute the network element information into the first service flow sequence, thereby obtaining the user service flow sequence.

[0066] Step B3: Determine the user service flow sequence as the first service flow sequence.

[0067] After determining the user service flow sequence, the fault location system identifies the user service flow sequence as the first service flow sequence.

[0068] Figure 3 This is a flowchart illustrating a second core network fault location method provided in one embodiment of this specification, as shown below. Figure 3 As shown, the schematic diagram includes: Step 302: Obtain the fault cause code and the first extended call detail record (XDR) data.

[0069] Among them, the fault cause code is the cause code generated when the target core network fails, the target core network is the core network to be located, the first XDR data is the XDR data generated by the target core network, and the difference between the generation time of the first XDR data and the generation time of the fault generation code is less than the first threshold.

[0070] Step 304: Determine the first network element information of multiple first network elements based on the first XDR data.

[0071] Among them, the first network element is the network element involved in the first XDR data, and the first network element information is the relevant information of the first network element.

[0072] Step 306: Obtain user attribute information.

[0073] Among them, the user attribute information is the attribute information of the first user, who is the user who uses the service in the target core network.

[0074] Step 308: Determine the user service flow sequence based on the first XDR data and user attribute information.

[0075] Among them, the user service flow sequence is the order in which data flows between the first network elements when the target core network carries the service of the first user.

[0076] Step 310: Determine the user service flow sequence as the first service flow sequence.

[0077] Among them, the first service flow sequence is the data flow order among the first network elements when the target core network carries the service.

[0078] Step 312: Determine the service flow feature vector corresponding to the first service flow sequence, so as to determine the first embedded feature vector of each first network element included in the first service flow sequence based on the service flow feature vector.

[0079] Among them, the service flow feature vector is the embedded feature vector corresponding to the first service flow sequence, and the first embedded feature vector is the embedded feature vector of the first network element.

[0080] Step 314: Group multiple first network elements based on the first network element information to obtain multiple first network element groups, and determine the first group feature vector of each first network element group and the first group intra-feature vector of each first network element based on the first embedded feature vector.

[0081] Among them, the first set of feature vectors are the feature vectors of the first network element group, and the feature vectors within the first set are the feature vectors of the first network element under the constraints of its corresponding first network element group.

[0082] Step 316: Based on the anomaly vector determination strategy, determine the anomaly group feature vector in multiple first group feature vectors and the anomaly group feature vector in multiple first group feature vectors, so as to locate the fault according to the fault cause code, the anomaly group feature vector and the anomaly group feature vector.

[0083] In the embodiments described in the specification, by determining the embedded feature vector of each first network element according to the user service flow sequence, the feature vector of the first network element group and the intra-group feature vector of the first network element can be accurately represented, thereby improving the accuracy of fault location of the first network element group and the first network element.

[0084] In one implementation, the step of grouping multiple first network elements based on the first network element information to obtain multiple first network element groups (step 104) can be executed via steps C1-C4: Step C1: Based on the type information included in each of the first network element information, determine the second network element and multiple third network elements.

[0085] Wherein, the second network element is any of the first network elements of the first type, the third network element is a non-first network element of the first type, and the first type is any of the multiple types corresponding to the first network element.

[0086] When grouping multiple first network elements, the fault location system can not only classify them directly according to the type of the first network element, but also according to the embedded feature vector of each first network element.

[0087] When classifying multiple first network elements based on the embedded feature vector of the first network element, the fault location system first determines the second network element and multiple third network elements based on the type information of each first network element included in the information of each first network element. The second network element is any one of the multiple first network elements of the first type, and the third network element is a first network element that is not the second network element among the multiple first network elements of the first type. The first type is any one of the multiple types corresponding to the multiple first network elements.

[0088] In other words, the fault location system first determines the type of each first network element, then selects multiple first network elements corresponding to a certain type, then selects one first network element from the multiple first network elements of that type as the second network element, and uses the remaining first network elements of that type as the third network element.

[0089] Step C2: Determine the first distance between the first embedded feature vector of the second network element and the first embedded feature vector of each of the third network elements.

[0090] The first distance is used to characterize the distance between the first embedded feature vector of the second network element and the first embedded feature vector of the third network element.

[0091] After identifying the second network element and multiple third network elements, the fault location system determines a first distance between the first embedded feature vector of the second network element and the first embedded feature vectors of each third network element. This first distance is the distance between the first embedded feature vectors of the second network element and the first embedded feature vectors of the third network elements. In other words, the fault location system determines the distance between the embedded feature vectors of the second network element and each of the third network elements.

[0092] Step C3: The third network element whose first distance is less than the second threshold is identified as the fourth network element.

[0093] The second threshold is determined based on the average distance between multiple first network elements of the first type.

[0094] After determining the first distance between the second network element and each third network element, the fault location system identifies the third network element whose first distance is less than the second threshold as the fourth network element. In other words, the fault location system identifies the third network element that is relatively close to the second network element (less than the second threshold) as the third network element.

[0095] The second threshold can be a pre-set threshold or a dynamically adjusted threshold. For example, it can be determined based on the average distance between multiple first-type network elements, i.e., based on the average distance between network elements of the same type.

[0096] Specifically, the fault location system calculates all Euclidean distances between the feature vector representations of the same type of first network element pairwise, and takes 1 / 4 of the mean of the Euclidean distances as the radius R (i.e., the second threshold). The formula for determining R is R=∑√(∑(xi-yi)²) / n 1 / 2, where i = 1, 2, ..., n.

[0097] Step C4: Determine the first network element group based on the second network element and the fourth network element.

[0098] After identifying multiple fourth network elements, the fault location system identifies the second network element and the multiple fourth network elements as the same group, in order to identify the first network element group and the multiple first network elements included in the first network element group.

[0099] Specifically, after determining the fourth network element, the fault location system, after determining the second network element, can first add the second network element to set A1, set A1={a1}. Then, using the second threshold as the search radius, it calculates the Euclidean distance between all network elements in set A and other first network elements of the same type. If the distance is less than the second threshold, it is added to set A1. The steps are repeated until no more elements are added to set A1, and A1 becomes the first network element group. Finally, the steps are repeated for each remaining first network element until all first network elements are assigned to the corresponding first network element groups, resulting in multiple first network element groups A1, A2, A3, ..., Ak.

[0100] Specifically, such as Figure 2As shown in the embedded feature vector representation of the first network element, it can be seen that within the same type of network element, AMF01, AMF02, AMF03, and AMF04 are more similar, and AMF05, AMF06, and AMF07 are also relatively similar. However, the relative distance between these two groups is relatively large. This is because network elements exist in group POOLs; AMF01~AMF04 belong to one POOL, and AMF05~AMF07 belong to another POOL. Therefore, when grouping, the fault location system assigns the more closely spaced AMF01~AMF04 to one first network element group, and the more closely spaced AMF05~AMF07 to another first network element group. The network element vectorization representation processed by natural language algorithms effectively preserves network element information and core network topology information. Figure 4 This is a flowchart illustrating the third core network fault location method provided in one embodiment of this specification, as shown below. Figure 4 As shown, the schematic diagram includes: Step 402: Obtain the fault cause code and the first extended call detail record (XDR) data.

[0101] Among them, the fault cause code is the cause code generated when the target core network fails, the target core network is the core network to be located, the first XDR data is the XDR data generated by the target core network, and the difference between the generation time of the first XDR data and the generation time of the fault generation code is less than the first threshold.

[0102] Step 404: Determine the first network element information of multiple first network elements based on the first XDR data.

[0103] Among them, the first network element is the network element involved in the first XDR data, and the first network element information is the relevant information of the first network element.

[0104] Step 406: Determine the first service flow sequence based on the first XDR data.

[0105] Among them, the first service flow sequence is the data flow order among the first network elements when the target core network carries the service.

[0106] Step 408: Determine the service flow feature vector corresponding to the first service flow sequence, so as to determine the first embedded feature vector of each first network element included in the first service flow sequence based on the service flow feature vector.

[0107] Among them, the service flow feature vector is the embedded feature vector corresponding to the first service flow sequence, and the first embedded feature vector is the embedded feature vector of the first network element.

[0108] Step 410: Based on the type information included in each first network element information, determine the second network element and multiple third network elements.

[0109] Among them, the second network element is any first network element of the first type, the third network element is a non-first network element of the first type, and the first type is any one of the multiple types corresponding to the first network element.

[0110] Step 412: Determine the first distance between the first embedded feature vector of the second network element and the first embedded feature vector of each third network element.

[0111] The first distance is used to characterize the distance between the first embedded feature vector of the second network element and the first embedded feature vector of the third network element.

[0112] Step 414: The third network element whose first distance is less than the second threshold is identified as the fourth network element.

[0113] The second threshold is determined based on the average distance between multiple first-type first network elements.

[0114] Step 416: Determine the first network element group based on the second network element and the fourth network element, and determine the first set of feature vectors of each first network element group and the first set of internal feature vectors of each first network element based on the first embedded feature vector.

[0115] Among them, the first set of feature vectors are the feature vectors of the first network element group, and the feature vectors within the first set are the feature vectors of the first network element under the constraints of its corresponding first network element group.

[0116] Step 418: Based on the anomaly vector determination strategy, determine the anomaly group feature vector in multiple first group feature vectors and the anomaly group feature vector in multiple first group feature vectors, so as to locate the fault according to the fault cause code, the anomaly group feature vector and the anomaly group feature vector.

[0117] In the embodiments described in the specification, multiple first network elements are grouped by a clustering algorithm. This can divide first network elements of the same type into different groups according to the distribution of their embedded feature vectors, thereby improving the grouping quality and enabling accurate fault location based on the feature vectors of each group and the feature vectors of each first network element.

[0118] In one implementation, before determining the outlier group feature vectors among the multiple first group feature vectors and the outlier intra-group feature vectors among the multiple first group intra-feature vectors based on the outlier vector determination strategy (step 108), steps D1-D4 may also be performed: Step D1: Obtain sample XDR data.

[0119] The sample XDR data refers to the historical XDR data generated by the target core network.

[0120] When determining the abnormal group feature vectors in the first set of feature vectors and the abnormal group feature vectors within the first set of feature vectors based on the anomaly vector determination model, the fault location system can not only train the model based on the XDR data generated during normal and abnormal operation of the target core network, but also train the model based only on the XDR data generated during normal operation of the target core network.

[0121] Specifically, when training the model based on the XDR data generated during the normal operation of the target core network, the fault location system first acquires sample XDR data, which is the historical XDR data generated by the target core network during normal operation.

[0122] Step D2: Determine the sample network element information and sample embedding feature vectors of multiple sample network elements based on the sample XDR data.

[0123] Wherein, the sample network element is the network element involved in the sample XDR data, the sample network element information is the relevant information of the sample network element, and the sample embedding feature vector is the embedding feature vector of the sample network element.

[0124] After acquiring the sample XDR data, the fault location system determines the sample network element information and sample embedding feature vectors of multiple sample network elements based on the sample XDR data. Specifically, the method by which the fault location system determines the sample network element information and sample embedding feature vectors of multiple sample network elements based on the sample XDR data is the same as the method described above for determining the first network element information and first embedding feature vectors of multiple first network elements based on the first XDR data, and will not be elaborated here.

[0125] Step D3: Group the sample network elements based on the sample network element information to obtain multiple sample network element groups, and determine the sample group feature vector and the sample group feature vector of each sample network element group based on the sample embedding feature vector.

[0126] Wherein, the sample group feature vector is the feature vector of the sample network element group, and the sample group feature vector is the feature vector of the sample network element under the constraint of the corresponding sample network element group.

[0127] After determining the sample network element information, the fault location system groups multiple sample network elements based on this information, thereby identifying multiple sample network element groups and the multiple sample network elements included in each sample network element group. Specifically, the method of grouping multiple sample network elements according to the sample network element information to determine multiple sample network element groups is the same as the method described above of grouping multiple first network elements according to the first sample network element information to determine multiple first network element groups, and will not be elaborated here.

[0128] After determining multiple sample network element groups, the fault location system determines the sample group feature vector and the intra-sample group feature vector of each sample network element based on the sample embedding feature vector. The sample group feature vector is the feature vector of the sample network element group, and the intra-sample group feature vector is the feature vector of the sample network element under the constraint of its expected corresponding sample network element group. Specifically, the method for determining the sample group feature vector and the intra-sample group feature vector based on the sample embedding feature vector is the same as the method described above for determining the first set of feature vectors and the first set of intra-sample group feature vectors based on the first embedded feature vector, and will not be elaborated here.

[0129] Step D4: Train the model based on the feature vectors of the sample group and the feature vectors within the sample group to obtain a pre-trained fault location model. Use the pre-trained fault location model to determine the abnormal group feature vectors in the multiple first group feature vectors and the abnormal group feature vectors within the multiple first group feature vectors.

[0130] After determining the feature vectors of the sample group and the feature vectors within the sample group, the fault location system trains the anomaly vector determination model based on these feature vectors. This enables the pre-trained anomaly vector determination model to identify normal sample group feature vectors and normal within-sample group feature vectors. The fault location system inputs the first set of feature vectors into the anomaly vector determination model to determine whether the feature vector is a normal vector; if not, it identifies it as an anomaly group feature vector. Similarly, it inputs the feature vectors within the first set of feature vectors into the anomaly vector determination model to determine whether the feature vector is a normal vector; if not, it identifies it as an anomaly group feature vector.

[0131] In one implementation, after determining the sample group feature vector and the sample group intra-sample feature vector of each sample network element based on the sample embedding feature vector (step D3), steps E1-E7 can also be performed: Step E1: Input the feature vector of the sample group into the first vector dimensionality reduction model to be trained for dimensionality reduction processing to obtain the dimensionality reduction vector of the sample group.

[0132] After determining the feature vectors of the sample group, the fault location system inputs the feature vectors of the sample group into the first vector dimensionality reduction model to be trained for dimensionality reduction processing, and obtains the dimensionality reduction result (i.e., the dimensionality reduction vector of the sample group).

[0133] Before inputting the sample group feature vectors into the first vector dimensionality reduction model to be trained, the fault localization system first preprocesses the sample group feature vectors by normalizing their representation. Next, the fault localization system constructs an encoder that uses two hidden layers to map the sample group feature vectors into a low-dimensional encoding space, employing the Sigmoid activation function. Finally, the fault localization system reverse-maps the encoding result to convert it into an output similar to the original (i.e., the sample group dimensionality reduction vector), again using the Sigmoid activation function.

[0134] Step E2: Determine the first error between the feature vector of the sample group and the dimensionality reduction vector of the sample group.

[0135] After determining the dimensionality reduction vector of the sample group, the fault location system determines the first error between the sample group feature vector and the sample group dimensionality reduction vector based on a preset error determination function. This first error characterizes the error between the sample group feature vector and the sample group dimensionality reduction vector. For example, the fault location system uses the mean squared error loss function to measure the difference between the reconstructed output and the original input, and determines the first error.

[0136] Step E3: Iteratively update the first vector dimensionality reduction model to be trained based on the first error to obtain the pre-trained first vector dimensionality reduction model.

[0137] After determining the first error, the fault location system iteratively updates the first vector dimensionality reduction model to be trained based on the first error, and obtains the pre-trained first vector dimensionality reduction model.

[0138] Specifically, when the fault location system uses the mean squared error loss function to measure the difference between the reconstructed output and the original input, the fault location system uses the gradient descent optimization algorithm to adjust the parameters of the encoder and decoder (i.e. the first vector dimensionality reduction model to be trained) to minimize the reconstruction error, thereby obtaining the pre-trained first vector dimensionality reduction model.

[0139] Step E4: Input the feature vector within the sample group into the second vector dimensionality reduction model to be trained for dimensionality reduction processing to obtain the dimensionality-reduced vector within the sample group.

[0140] After determining the feature vectors within the sample group, the fault location system inputs the feature vectors within the sample group into the second vector dimensionality reduction model to be trained for dimensionality reduction processing, and obtains the dimensionality-reduced vectors within the sample group.

[0141] Specifically, the method of inputting the feature vectors within the sample group into the second vector dimensionality reduction model for dimensionality reduction processing to obtain the dimensionality-reduced vectors within the sample group is the same as the method described above of inputting the feature vectors of the sample group into the first vector dimensionality reduction model to be trained for processing to obtain the dimensionality-reduced vectors of the sample group, and will not be elaborated here.

[0142] Step E5: Determine the second error between the feature vector within the sample group and the dimension reduction vector within the sample group.

[0143] After obtaining the dimension reduction vector within the sample group, the fault location system determines the second error between the feature vector within the sample group and the dimension reduction vector within the sample group. The second error is used to characterize the error between the feature vector within the sample group and the dimension reduction vector within the sample group.

[0144] Specifically, the method for determining the second error between the feature vector within the sample group and the dimension-reduced vector within the sample group is the same as the method for determining the first error between the feature vector within the sample group and the dimension-reduced vector of the sample group, and will not be elaborated here.

[0145] Step E6: Iteratively update the second vector dimensionality reduction model to be trained based on the second error to obtain the pre-trained second vector dimensionality reduction model.

[0146] After determining the second error, the fault location system iteratively updates the second vector dimensionality reduction model to be trained based on the second error, thereby obtaining the pre-trained second vector dimensionality reduction model.

[0147] Specifically, the method for obtaining the pre-trained second vector dimensionality reduction model by iteratively updating based on the second error is the same as the method for obtaining the pre-trained first vector dimensionality reduction model by iteratively updating based on the first error, and will not be elaborated here.

[0148] Step E7: Based on the pre-trained first vector dimensionality reduction model and the pre-trained second vector dimensionality reduction model, perform dimensionality reduction processing on the first group of feature vectors and the feature vectors within the first group, respectively, so as to perform fault location based on the first group of dimensionality reduction vectors and the dimensionality reduction vectors within the first group.

[0149] After obtaining the pre-trained first vector dimensionality reduction model and the second vector dimensionality reduction model, the fault location system inputs the first set of feature vectors into the first vector dimensionality reduction model for dimensionality reduction processing, thereby obtaining the first set of dimensionality-reduced vectors. Based on the first set of dimensionality-reduced vectors, it determines whether they are normal feature vectors. If not, the first set of feature vectors is determined to be an abnormal group feature vector. The feature vectors within the first set are input into the second vector dimensionality reduction model for dimensionality reduction processing, thereby obtaining the dimensionality-reduced vectors within the first set. Based on the dimensionality-reduced vectors within the first set, it determines whether they are normal feature vectors. If not, the dimensionality-reduced vectors within the first set are determined to be an abnormal group feature vector.

[0150] Specifically, the first and second vector dimensionality reduction models reduce the dimensionality of the first set of feature vectors and the feature vectors within the first set, respectively, to highlight the dimensionality of feature vectors exhibiting anomalies. Both models are trained using feature vectors from XDR data generated during normal operation. Therefore, when reducing the dimensionality of the first set of feature vectors and the feature vectors within the first set using the pre-trained model, it can highlight anomalous feature vectors when anomalies exist in either the first set or the feature vectors within the first set. Furthermore, the fault location system can integrate the first and second vector dimensionality reduction models to obtain a comprehensive vector dimensionality reduction model, and then use this comprehensive model to reduce the dimensionality of both the first and second sets of feature vectors for anomaly identification.

[0151] Figure 5 This is a schematic diagram of the structure of a core network fault location system provided in one embodiment of this specification, as shown below. Figure 5 As shown, the core network fault location system includes an XDR processing engine module, a network element information management module, a network element service flow serialization module, a natural language vectorization processing module, an anomaly perception module, a vector sequence data generation module, and a delimitation and location module.

[0152] The XDR processing engine module is used for: 1. XDR association processing: Reading composite XDR data or multi-interface XDR data from the XDR database. If it is multi-interface XDR data, it combines the XDRs according to the business process to construct composite XDR data. 2. XDR network element information extraction: Extracting the interface, protocol, and source / destination network element IP address from the XDR. Based on the above information, it queries the network element name in the network management system and stores the network element type, network element name, and network element IP address in the network element information management module.

[0153] The network element information management module is used to maintain network element names, network element IP addresses, network element vector information, network element group vector information, and maintain the core network topology.

[0154] The network element service flow serialization module is used to: combine the network elements that the service flows through according to the service flow logic in a temporal sequence to form a corpus (sample), which serves as the input for network element vectorization training in the modeling stage and autoencoder training in the delimitation and localization stage.

[0155] The Natural Language Vectorization Processing Module is used for: 1. Generating a unique vector for each network element based on the correlation between the preceding and following network element sequences, based on natural language processing methods. The vector contains the topology and service relationships of the network element, and the associated network element IP information and network element name are stored in the network element information management module; 2. Extracting network element grouping information based on the DBSCAN algorithm and storing it in the network element information management module.

[0156] The anomaly detection module is used to interface with external anomaly monitoring systems, such as error cause code anomaly fluctuation statistics and performance indicator degradation alarm monitoring. When an external alarm is received, this module collects anomaly XDR information within a relevant time range based on the alarm information and triggers the delimitation and location logic.

[0157] The vector sequence data generation module is used to: receive the serialized information of network elements, and obtain the network element vector and network element grouping vector information corresponding to the serialized network elements from the network element information management module, and construct vectorized sequence data. The vectorized sequence data will simplify the implementation of the autoencoder-based dimensionality reduction clustering algorithm in the subsequent delimitation and localization.

[0158] The delimitation and localization module is used to: take the vectorized sequence set obtained in the vector sequence data generation module ⑥ as a sample, perform dimensionality reduction and clustering through an autoencoder to obtain the delimitation and localization results, and output the delimitation and localization results by associating them with the network element information in the network element information management module.

[0159] It should be noted that the core network fault location method provided in this application embodiment can be executed by a core network fault location device, or a control module in the core network fault location device for executing the core network fault location method. This application embodiment uses the execution of the core network fault location method by a core network fault location device as an example to illustrate the core network fault location device provided in this application embodiment.

[0160] Figure 6 This is a schematic diagram of the core network fault location device according to an embodiment of the present invention. Figure 6 As shown, the core network fault location device includes: a first acquisition module 602, a first determination module 604, a second determination module 606, and a first location module 608.

[0161] The first acquisition module 602 is used to acquire fault cause code and first extended call detail record (XDR) data. The fault cause code is a cause code generated when a fault occurs in the target core network. The target core network is the core network to be located. The first XDR data is the XDR data generated by the target core network. The difference between the generation time of the first XDR data and the generation time of the fault generation code is less than a first threshold. The first determining module 604 is used to determine the first network element information and the first embedded feature vector of a plurality of first network elements based on the first XDR data. The first network element is the network element involved in the first XDR data, the first network element information is the relevant information of the first network element, and the first embedded feature vector is the embedded feature vector of the first network element. The second determining module 606 is used to group multiple first network elements based on the first network element information to obtain multiple first network element groups, and to determine the first group feature vector of each first network element group and the first group feature vector of each first network element based on the first embedded feature vector. The first group feature vector is the feature vector of the first network element group, and the first group feature vector is the feature vector of the first network element under the constraint of its corresponding first network element group. The first positioning module 608 is used to determine the abnormal group feature vector and the abnormal group feature vector in the first group feature vector based on the abnormal vector determination strategy, so as to perform fault positioning based on the fault cause code, the abnormal group feature vector and the abnormal group feature vector.

[0162] The core network fault location device in this application embodiment can be a device, or it can be a component, integrated circuit, or chip in a terminal. The device can be a mobile electronic device or a non-mobile electronic device. For example, mobile electronic devices can be mobile phones, tablets, laptops, PDAs, in-vehicle electronic devices, wearable devices, ultra-mobile personal computers (UMPCs), netbooks, or personal digital assistants (PDAs), etc., while non-mobile electronic devices can be servers, network attached storage (NAS), personal computers (PCs), televisions (TVs), ATMs, or self-service machines, etc. This application embodiment does not make specific limitations.

[0163] The core network fault location device in this application embodiment can be a device with an operating system. This operating system can be Android, iOS, or other possible operating systems; this application embodiment does not specifically limit it.

[0164] The core network fault location device provided in this application embodiment can achieve... Figures 1 to 5 The various processes implemented in the method embodiments are not described in detail here to avoid repetition.

[0165] Based on the same technical concept, embodiments of this application also provide an electronic device for performing the core network fault location method described above. Figure 7This is a schematic diagram of the structure of an electronic device to implement various embodiments of this application. The electronic device can vary significantly due to differences in configuration or performance, and may include a processor 702, a communications interface 704, a memory 706, and a communication bus 708. The processor 702, communications interface 704, and memory 706 communicate with each other via the communication bus 708. The processor 702 can call a computer program stored in the memory 706 and executable on the processor 702 to perform the following steps: Obtain the fault cause code and the first extended call detail record (XDR) data. The fault cause code is the cause code generated when a fault occurs in the target core network. The target core network is the core network to be located. The first XDR data is the XDR data generated by the target core network. The difference between the generation time of the first XDR data and the generation time of the fault cause code is less than a first threshold. Based on the first XDR data, the first network element information and the first embedded feature vector of multiple first network elements are determined. The first network element is the network element involved in the first XDR data, the first network element information is the relevant information of the first network element, and the first embedded feature vector is the embedded feature vector of the first network element. Based on the first network element information, multiple first network elements are grouped to obtain multiple first network element groups. Based on the first embedded feature vector, the first group feature vector of each first network element group and the first group feature vector of each first network element are determined. The first group feature vector is the feature vector of the first network element group, and the first group feature vector is the feature vector of the first network element under the constraint of its corresponding first network element group. Based on the anomaly vector determination strategy, anomaly group feature vectors and anomaly group feature vectors in the first group feature vectors are determined, so as to locate the fault according to the fault cause code, the anomaly group feature vectors and the anomaly group feature vectors.

[0166] In one implementation, determining the first network element information and the first embedded feature vector of multiple first network elements based on the first XDR data includes: The first service flow sequence is determined based on the first XDR data. The first service flow sequence is the order in which data flows among the first network elements when the target core network carries the service. Determine the service flow feature vector corresponding to the first service flow sequence, and determine the first embedded feature vector of each first network element included in the first service flow sequence based on the service flow feature vector, wherein the service flow feature vector is the embedded feature vector corresponding to the first service flow sequence.

[0167] In one implementation, determining the first service flow sequence based on the first XDR data includes: Obtain user attribute information, wherein the user attribute information is the attribute information of a first user, and the first user is a user who uses services in the target core network; The user service flow sequence is determined based on the first XDR data and the user attribute information. The user service flow sequence is the order in which data flows between the first network elements when the target core network carries the service of the first user. The user service flow sequence is determined as the first service flow sequence.

[0168] In one implementation, the step of grouping multiple first network elements based on the first network element information to obtain multiple first network element groups includes: Based on the type information included in each of the first network element information, a second network element and multiple third network elements are determined. The second network element is any of the first network elements of the first type, and the third network element is a non-first network element of the first type. The first type is any of the multiple types corresponding to the first network element. A first distance is determined between the first embedded feature vector of the second network element and the first embedded feature vector of each of the third network elements, wherein the first distance is used to characterize the distance between the first embedded feature vector of the second network element and the first embedded feature vector of the third network element; The third network element whose first distance is less than the second threshold is determined as the fourth network element, and the second threshold is determined based on the average distance between multiple first network elements of the first type. The first network element group is determined based on the second network element and the fourth network element.

[0169] In one implementation, before determining the outlier group feature vectors among the plurality of first group feature vectors and the outlier intra-group feature vectors among the plurality of first group intra-feature vectors based on the outlier vector determination strategy, the method further includes: Acquire sample XDR data, which is historical XDR data generated by the target core network; Based on the sample XDR data, sample network element information and sample embedding feature vectors of multiple sample network elements are determined. The sample network element is the network element involved in the sample XDR data, the sample network element information is the relevant information of the sample network element, and the sample embedding feature vector is the embedding feature vector of the sample network element. Based on the sample network element information, the sample network elements are grouped to obtain multiple sample network element groups. Based on the sample embedding feature vector, the sample group feature vector and the sample group feature vector of each sample network element are determined. The sample group feature vector is the feature vector of the sample network element group, and the sample group feature vector is the feature vector of the sample network element under the constraint of its corresponding sample network element group. Based on the feature vectors of the sample group and the feature vectors within the sample group, a model is trained to obtain a pre-trained fault location model. Based on the pre-trained fault location model, abnormal group feature vectors in multiple first group feature vectors and abnormal group feature vectors within multiple first group feature vectors are determined.

[0170] In one implementation, after determining the sample group feature vector and the sample group feature vector of each sample network element based on the sample embedding feature vector, the method further includes: The feature vector of the sample group is input into the first vector dimensionality reduction model to be trained for dimensionality reduction processing to obtain the dimensionality reduction vector of the sample group. Determine the first error between the feature vector of the sample group and the dimension-reduced vector of the sample group; Based on the first error, the first vector dimensionality reduction model to be trained is iteratively updated to obtain the pre-trained first vector dimensionality reduction model. The feature vectors within the sample group are input into the second vector dimensionality reduction model to be trained for dimensionality reduction processing to obtain the dimensionality-reduced vectors within the sample group. Determine the second error between the feature vector within the sample group and the dimension-reduced vector within the sample group; Based on the second error, the second vector dimensionality reduction model to be trained is iteratively updated to obtain the pre-trained second vector dimensionality reduction model; The first vector dimensionality reduction model and the second vector dimensionality reduction model, which are pre-trained, are used to reduce the dimensionality of the first set of feature vectors and the feature vectors within the first set, respectively, so as to perform fault location based on the first set of dimensionality reduction vectors and the dimensionality reduction vectors within the first set.

[0171] The specific execution steps can be found in the various steps of the above-described core network fault location method embodiment, and can achieve the same technical effect. To avoid repetition, they will not be repeated here.

[0172] It should be noted that the electronic devices in the embodiments of this application include: servers, terminals, or other devices besides terminals.

[0173] The above electronic device structure does not constitute a limitation on the electronic device. An electronic device may include more or fewer components than illustrated, or combine certain components, or arrange them differently. For example, an input unit may include a Graphics Processing Unit (GPU) and a microphone, and a display unit may use a liquid crystal display (LCD), organic light-emitting diode (OLED), or other similar display panels. User input units include at least one of a touch panel and other input devices. A touch panel is also called a touchscreen. Other input devices may include, but are not limited to, physical keyboards, function keys (such as volume control buttons, power buttons, etc.), trackballs, mice, and joysticks, which will not be elaborated further here.

[0174] Memory can be used to store software programs and various data. Memory can primarily include a first storage area for storing programs or instructions and a second storage area for storing data. The first storage area can store the operating system, application programs or instructions required for at least one function (such as sound playback, image playback, etc.). Furthermore, memory can include volatile memory or non-volatile memory, or both. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM), static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous linked dynamic random access memory (Synchlink DRAM, SLDRAM), and direct memory bus RAM (DRRAM).

[0175] The processor may include one or more processing units; optionally, the processor integrates an application processor and a modem processor, wherein the application processor mainly handles operations related to the operating system, user interface, and applications, while the modem processor mainly handles wireless communication signals, such as a baseband processor. It is understood that the aforementioned modem processor may also not be integrated into the processor.

[0176] This application also provides a readable storage medium storing a program or instructions. When the program or instructions are executed by a processor, they implement the various processes of the above-described core network fault location method embodiments and achieve the same technical effect. To avoid repetition, they will not be described again here.

[0177] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0178] This application also provides a computer program product. When the computer program product is executed by a processor, it implements the various processes of the above-described core network fault location method embodiments and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0179] This application embodiment also provides a chip, which includes a processor and a communication interface. The communication interface and the processor are coupled. The processor is used to run programs or instructions to implement the various processes of the above-described core network fault location method embodiments and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0180] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0181] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, 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 limitations, 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 that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0182] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0183] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

Claims

1. A core network fault localization method, characterized in that, include: Obtain the fault cause code and the first extended call detail record (XDR) data. The fault cause code is the cause code generated when a fault occurs in the target core network. The target core network is the core network to be located. The first XDR data is the XDR data generated by the target core network. The difference between the generation time of the first XDR data and the generation time of the fault cause code is less than a first threshold. Based on the first XDR data, the first network element information and the first embedded feature vector of multiple first network elements are determined. The first network element is the network element involved in the first XDR data, the first network element information is the relevant information of the first network element, and the first embedded feature vector is the embedded feature vector of the first network element. Based on the first network element information, multiple first network elements are grouped to obtain multiple first network element groups. Based on the first embedded feature vector, the first group feature vector of each first network element group and the first group feature vector of each first network element are determined. The first group feature vector is the feature vector of the first network element group, and the first group feature vector is the feature vector of the first network element under the constraint of its corresponding first network element group. Based on the anomaly vector determination strategy, anomaly group feature vectors and anomaly group feature vectors in the first group feature vectors are determined, so as to locate the fault according to the fault cause code, the anomaly group feature vectors and the anomaly group feature vectors.

2. The method of claim 1, wherein, The step of determining the first network element information and the first embedded feature vector of multiple first network elements based on the first XDR data includes: The first service flow sequence is determined based on the first XDR data. The first service flow sequence is the order in which data flows among the first network elements when the target core network carries the service. Determine the service flow feature vector corresponding to the first service flow sequence, and determine the first embedded feature vector of each first network element included in the first service flow sequence based on the service flow feature vector, wherein the service flow feature vector is the embedded feature vector corresponding to the first service flow sequence.

3. The method of claim 2, wherein, The step of determining the first service flow sequence based on the first XDR data includes: Obtain user attribute information, wherein the user attribute information is the attribute information of a first user, and the first user is a user who uses services in the target core network; The user service flow sequence is determined based on the first XDR data and the user attribute information. The user service flow sequence is the order in which data flows between the first network elements when the target core network carries the service of the first user. The user service flow sequence is determined as the first service flow sequence.

4. The method of claim 2, wherein, The step of grouping multiple first network elements based on the first network element information to obtain multiple first network element groups includes: Based on the type information included in each of the first network element information, a second network element and multiple third network elements are determined. The second network element is any of the first network elements of the first type, and the third network element is a non-first network element of the first type. The first type is any of the multiple types corresponding to the first network element. A first distance is determined between the first embedded feature vector of the second network element and the first embedded feature vector of each of the third network elements, wherein the first distance is used to characterize the distance between the first embedded feature vector of the second network element and the first embedded feature vector of the third network element; The third network element whose first distance is less than the second threshold is determined as the fourth network element, and the second threshold is determined based on the average distance between multiple first network elements of the first type. The first network element group is determined based on the second network element and the fourth network element.

5. The method of claim 1, wherein, Before determining the outlier group feature vectors and the outlier intra-group feature vectors among the multiple first group feature vectors based on the outlier vector determination strategy, the method further includes: Acquire sample XDR data, which is historical XDR data generated by the target core network; Based on the sample XDR data, sample network element information and sample embedding feature vectors of multiple sample network elements are determined. The sample network element is the network element involved in the sample XDR data, the sample network element information is the relevant information of the sample network element, and the sample embedding feature vector is the embedding feature vector of the sample network element. Based on the sample network element information, the sample network elements are grouped to obtain multiple sample network element groups. Based on the sample embedding feature vector, the sample group feature vector and the sample group feature vector of each sample network element are determined. The sample group feature vector is the feature vector of the sample network element group, and the sample group feature vector is the feature vector of the sample network element under the constraint of its corresponding sample network element group. Based on the feature vectors of the sample group and the feature vectors within the sample group, a model is trained to obtain a pre-trained fault location model. Based on the pre-trained fault location model, abnormal group feature vectors in multiple first group feature vectors and abnormal group feature vectors within multiple first group feature vectors are determined.

6. The method of claim 5, wherein, After determining the sample group feature vector and the sample group feature vector of each sample network element based on the sample embedding feature vector, the method further includes: The feature vector of the sample group is input into the first vector dimensionality reduction model to be trained for dimensionality reduction processing to obtain the dimensionality reduction vector of the sample group. Determine the first error between the feature vector of the sample group and the dimension-reduced vector of the sample group; Based on the first error, the first vector dimensionality reduction model to be trained is iteratively updated to obtain the pre-trained first vector dimensionality reduction model. The feature vectors within the sample group are input into the second vector dimensionality reduction model to be trained for dimensionality reduction processing to obtain the dimensionality-reduced vectors within the sample group. Determine the second error between the feature vector within the sample group and the dimension-reduced vector within the sample group; Based on the second error, the second vector dimensionality reduction model to be trained is iteratively updated to obtain the pre-trained second vector dimensionality reduction model; The first vector dimensionality reduction model and the second vector dimensionality reduction model, which are pre-trained, are used to reduce the dimensionality of the first set of feature vectors and the feature vectors within the first set, respectively, so as to perform fault location based on the first set of dimensionality reduction vectors and the dimensionality reduction vectors within the first set.

7. A core network fault localization apparatus, characterized by, include: The first acquisition module is used to acquire fault cause code and first extended call detail record (XDR) data. The fault cause code is a cause code generated when a fault occurs in the target core network. The target core network is the core network to be located. The first XDR data is the XDR data generated by the target core network. The difference between the generation time of the first XDR data and the generation time of the fault generation code is less than a first threshold. The first determining module is used to determine the first network element information and the first embedded feature vector of a plurality of first network elements based on the first XDR data. The first network element is the network element involved in the first XDR data, the first network element information is the relevant information of the first network element, and the first embedded feature vector is the embedded feature vector of the first network element. The second determining module is used to group multiple first network elements based on the first network element information to obtain multiple first network element groups, and to determine the first group feature vector of each first network element group and the first group feature vector of each first network element based on the first embedded feature vector. The first group feature vector is the feature vector of the first network element group, and the first group feature vector is the feature vector of the first network element under the constraint of its corresponding first network element group. The first positioning module is used to determine the abnormal group feature vector and the abnormal group feature vector in the first group feature vector based on the abnormal vector determination strategy, so as to perform fault positioning based on the fault cause code, the abnormal group feature vector and the abnormal group feature vector.

8. An electronic device, comprising: The device includes: Processor; and A memory is configured to store computer-executable instructions configured to be executed by the processor, the executable instructions including steps for performing the core network fault location method as described in any one of claims 1 to 6.

9. A storage medium, characterized in that, The storage medium is used to store computer-executable instructions that cause the computer to perform the core network fault location method as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, It includes a computer program that, when executed by a processor, implements the core network fault location method according to any one of claims 1 to 6.

Images