CN114079957A

CN114079957A - Method and equipment for detecting abnormal state of cell

Info

Publication number: CN114079957A
Application number: CN202010794979.XA
Authority: CN
Inventors: 杨骄龙; 袁雁南
Original assignee: China Mobile Communications Group Co Ltd; China Mobile Communications Ltd Research Institute
Current assignee: China Mobile Communications Group Co Ltd; China Mobile Communications Ltd Research Institute
Priority date: 2020-08-10
Filing date: 2020-08-10
Publication date: 2022-02-22

Abstract

The invention provides a method and a device for detecting abnormal states of a cell, wherein the method comprises the following steps: acquiring at least one Key Performance Indicator (KPI) parameter set of a first cell acquired within a preset time period, wherein the KPI parameter set comprises values of a plurality of KPI parameters acquired at corresponding sampling time points; performing dimension reduction processing on each KPI parameter group of the first cell respectively to generate at least one first dimension reduction vector of the first cell; and inputting the at least one first dimension reduction vector to an evaluator, and generating evaluation information of the abnormal state of the first cell. The invention can simplify the detection of the abnormal load state of the network and improve the accuracy of the detection result.

Description

Method and equipment for detecting abnormal state of cell

Technical Field

The invention relates to the technical field of mobile communication, in particular to a method and equipment for detecting abnormal states of a cell.

Background

In network operation and maintenance, a mobile communication network operator generally defines a plurality of high-load related statistical data of self-busy cells which need to be counted under the condition of high network load, manually divides a threshold value, and judges a time period meeting the threshold value condition as a high-load time period. For example, when a cell in the existing network is under high load at busy hours for more than three days a week, a work order is issued and submitted to a network optimization engineer of the existing network for processing. The judgment processing of the cell load based on the threshold is rough, a network optimization engineer needs to continuously observe for a period of time, and meanwhile, various parameters such as Key Performance Indicator (KPI) parameters and adjacent cell information are combined to perform manual judgment, and cells which are fully loaded on some parameters but are not judged to be high load may be missed.

On the other hand, when optimizing a high-load cell, a current network engineer usually judges the type of the high load according to own experience and finds out the reason causing the high load, and the process is still mainly completed by analyzing the KPI parameters. However, since the number of cell KPI parameters is very large, even though the secondary parameters are reduced, the joint state space of the mapping is huge, which is difficult for engineers to handle. Thus, existing web engineers typically track only a few commonly used parameters while optimizing in conjunction with other empirical methods. The workload of the above network optimization schemes is usually very large, and in addition, the practical effect is also limited by expert experience.

Disclosure of Invention

At least one embodiment of the present invention provides a method, a terminal, and a network device for detecting a cell abnormal state, which are used to simplify the detection of a network abnormal load state and improve the accuracy of a detection result.

According to an aspect of the present invention, at least one embodiment provides a method for detecting a cell abnormal state, including:

acquiring at least one Key Performance Indicator (KPI) parameter set of a first cell acquired within a preset time period, wherein the KPI parameter set comprises values of a plurality of KPI parameters acquired at corresponding sampling time points;

performing dimension reduction processing on each KPI parameter group of the first cell respectively to generate at least one first dimension reduction vector of the first cell;

and inputting the at least one first dimension reduction vector to an evaluator, and generating evaluation information of the abnormal state of the first cell.

Optionally, the performing, separately, a dimension reduction process on each KPI parameter set of the first cell includes:

and respectively inputting each KPI parameter group of the first cell into a pre-trained self-encoder to obtain a first dimension reduction vector output by the self-encoder.

Optionally, the evaluator is configured to perform clustering on the at least one first dimension reduction vector, determine and output a first category to which the at least one first dimension reduction vector belongs and corresponding label information, where the first category is one of a plurality of pre-generated categories, and the label information is used to indicate a cell abnormal state of the corresponding category.

Optionally, before obtaining at least one KPI parameter set of the first cell, the method further includes:

acquiring a plurality of KPI parameter sets of a plurality of cells acquired at a plurality of sampling time points, wherein each KPI parameter set comprises the plurality of KPI parameters acquired by the same cell at one sampling time point;

training an autoencoder by using the plurality of KPI parameter groups until a preset training end condition is reached; the self-encoder comprises an encoder and a decoder, and the encoder and the decoder comprise multilayer neural networks.

Optionally, the method further includes:

obtaining a plurality of dimensionality reduction vectors obtained after the plurality of KPI parameter sets are input to the trained self-encoder;

and clustering the dimension reduction vectors to obtain a plurality of categories.

Optionally, after obtaining the plurality of categories, the method further includes:

for each category, the following processing is performed:

selecting a parameter group belonging to the category from the plurality of KPI parameter groups, training a classifier by using the selected parameter group, and calculating the weight of each KPI parameter in the parameter group according to the training result;

repeatedly performing the following steps until no KPI parameters are present in the set of parameters: selecting a target KPI parameter with the maximum weight from the rest KPI parameters in the parameter group, adding the target KPI parameter into a parameter set, and deleting the target KPI parameter and the KPI parameter with the correlation with the target KPI parameter larger than a preset threshold from the parameter group;

selecting a preset number of KPI parameters from the parameter set according to the sequence of the weights from large to small as the KPI parameters of the category, and calculating the deviation degree of the statistical characteristics of the selected KPI parameters in the category relative to the statistical characteristics in all samples;

outputting the KPI parameters, the weights and the deviation degrees of the category as the characteristics of the category;

receiving the label information input aiming at the category, and establishing the corresponding relation between the label information and the category.

Optionally, the annotation information includes at least one of the following information: the evaluation value of the abnormal degree of the cell, the evaluation value of the load degree of the cell, the evaluation value of the degree to be optimized of the cell and remark information.

Optionally, the number of the plurality of categories is

Wherein N is the number of groups of the plurality of KPI parameter groups.

Optionally, the determining the first category to which the at least one first dimension reduction vector belongs and the corresponding labeling information thereof includes:

clustering the at least one first dimension reduction vector, determining a first central point of the at least one first dimension reduction vector, calculating Euclidean distances between the first central point and the central points of all categories, and selecting a category corresponding to the shortest Euclidean distance as a first category to which the at least one first dimension reduction vector belongs;

and determining the labeling information corresponding to the first category according to the pre-established correspondence between the labeling information and the categories.

Optionally, before the step of clustering the at least one first dimension-reduced vector, the method further includes:

and carrying out abnormal value detection on the at least one first dimension reduction vector, and removing outliers in the at least one first dimension reduction vector to obtain an updated first dimension reduction vector.

According to another aspect of the present invention, at least one embodiment provides a device for detecting a cell abnormal state, including:

a parameter obtaining module, configured to obtain at least one key performance indicator KPI parameter set of a first cell, where the KPI parameter set is acquired within a preset time period, and the KPI parameter set includes values of multiple KPI parameters acquired at corresponding sampling time points;

a dimension reduction processing module, configured to perform dimension reduction processing on each KPI parameter set of the first cell, respectively, and generate at least one first dimension reduction vector of the first cell;

and the label processing module is used for inputting the at least one first dimension reduction vector to an evaluator and generating evaluation information of the abnormal state of the first cell.

Optionally, the dimension reduction processing module is further configured to input each KPI parameter set of the first cell to a pre-trained self-encoder respectively, so as to obtain a first dimension reduction vector output by the self-encoder.

Optionally, the labeling processing module is further configured to perform clustering processing on the at least one first dimension reduction vector through the evaluator, determine and output a first category to which the at least one first dimension reduction vector belongs and corresponding labeling information, where the first category is one of a plurality of pre-generated categories, and the labeling information is used to indicate a cell abnormal state of the corresponding category.

Optionally, the detection device further includes:

a self-encoder training module, configured to, before obtaining at least one KPI parameter set of the first cell, obtain multiple KPI parameter sets of multiple cells acquired at multiple sampling time points, where each KPI parameter set includes the multiple KPI parameters acquired by the same cell at one sampling time point; training an autoencoder by using the plurality of KPI parameter groups until a preset training end condition is reached; the self-encoder comprises an encoder and a decoder, and the encoder and the decoder comprise multilayer neural networks.

Optionally, the detection device further includes:

the clustering module is used for acquiring a plurality of dimensionality reduction vectors obtained after the plurality of KPI parameter sets are input to the trained self-encoder; and clustering the dimension reduction vectors to obtain a plurality of categories.

Optionally, the detection device further includes:

the marking information establishing module is used for respectively executing the following processing aiming at each category after obtaining a plurality of categories:

Optionally, the labeling processing module is further configured to perform clustering processing on the at least one first dimension reduction vector, determine a first center point of the at least one first dimension reduction vector, calculate euclidean distances between the first center point and center points of each category, and select a category corresponding to a shortest euclidean distance as a first category to which the at least one first dimension reduction vector belongs; and determining the labeling information corresponding to the first category according to the pre-established correspondence between the labeling information and the categories.

Optionally, the detection device further includes:

and the outlier removing module is used for detecting an outlier of the at least one first dimension-reduced vector before clustering the at least one first dimension-reduced vector, and removing outliers in the at least one first dimension-reduced vector to obtain an updated first dimension-reduced vector.

According to another aspect of the present invention, at least one embodiment provides a device for detecting a cell abnormal state, comprising a transceiver and a processor, wherein,

the transceiver is configured to acquire at least one key performance indicator KPI parameter set of a first cell acquired within a preset time period, where the KPI parameter set includes values of multiple KPI parameters acquired at corresponding sampling time points;

the processor is configured to perform dimension reduction processing on each KPI parameter set of the first cell, and generate at least one first dimension reduction vector of the first cell; and inputting the at least one first dimension reduction vector to an evaluator, and generating evaluation information of the abnormal state of the first cell.

According to another aspect of the invention, at least one embodiment provides a computer readable storage medium having a program stored thereon, which when executed by a processor, performs the steps of the method as described above.

Compared with the prior art, the method and the device for detecting the abnormal state of the cell, provided by the embodiment of the invention, can simplify the calculation amount of subsequent classification processing, simplify the detection of the abnormal load state of the network and improve the detection efficiency.

Drawings

Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to refer to like parts throughout the drawings. In the drawings:

fig. 1 is a schematic view of an application scenario according to an embodiment of the present invention;

fig. 2 is a flowchart of a method for detecting abnormal conditions of a cell according to an embodiment of the present invention;

fig. 3 is a schematic general flowchart of a method for detecting a cell abnormal state according to an embodiment of the present invention;

FIG. 4 is a schematic structural diagram of an auto-encoder according to an embodiment of the present invention;

fig. 5 is a schematic structural diagram of a device for detecting an abnormal state of a cell according to an embodiment of the present invention;

fig. 6 is another schematic structural diagram of a device for detecting a cell abnormal state according to an embodiment of the present invention.

Detailed Description

Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

The terms first, second and the like in the description and in the claims of the present application are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the application described herein are, for example, capable of operation in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus. In the description and in the claims "and/or" means at least one of the connected objects.

The techniques described herein are not limited to NR systems and Long Time Evolution (LTE)/LTE Evolution (LTE-a) systems, and may also be used for various wireless communication systems, such as Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), Orthogonal Frequency Division Multiple Access (OFDMA), Single carrier Frequency Division Multiple Access (SC-FDMA), and other systems. The terms "system" and "network" are often used interchangeably. CDMA systems may implement Radio technologies such as CDMA2000, Universal Terrestrial Radio Access (UTRA), and so on. UTRA includes Wideband CDMA (Wideband Code Division Multiple Access, WCDMA) and other CDMA variants. TDMA systems may implement radio technologies such as Global System for Mobile communications (GSM). The OFDMA system may implement radio technologies such as Ultra Mobile Broadband (UMB), evolved-UTRA (E-UTRA), IEEE 802.21(Wi-Fi), IEEE 802.16(WiMAX), IEEE 802.20, Flash-OFDM, etc. UTRA and E-UTRA are parts of the Universal Mobile Telecommunications System (UMTS). LTE and higher LTE (e.g., LTE-A) are new UMTS releases that use E-UTRA. UTRA, E-UTRA, UMTS, LTE-A, and GSM are described in documents from an organization named "third Generation Partnership Project" (3 GPP). CDMA2000 and UMB are described in documents from an organization named "third generation partnership project 2" (3GPP 2). The techniques described herein may be used for both the above-mentioned systems and radio technologies, as well as for other systems and radio technologies. However, the following description describes the NR system for purposes of example, and NR terminology is used in much of the description below, although the techniques may also be applied to applications other than NR system applications.

The following description provides examples and does not limit the scope, applicability, or configuration set forth in the claims. Changes may be made in the function and arrangement of elements discussed without departing from the spirit and scope of the disclosure. Various examples may omit, substitute, or add various procedures or components as appropriate. For example, the described methods may be performed in an order different than described, and various steps may be added, omitted, or combined. In addition, features described with reference to certain examples may be combined in other examples.

Referring to fig. 1, fig. 1 is a block diagram of a wireless communication system to which an embodiment of the present invention is applicable. The wireless communication system includes a terminal 11 and a network device 12. The terminal 11 may also be referred to as a User terminal or a User Equipment (UE), where the terminal 11 may be a Mobile phone, a Tablet Personal Computer (Tablet Personal Computer), a Laptop Computer (Laptop Computer), a Personal Digital Assistant (PDA), a Mobile Internet Device (MID), a Wearable Device (Wearable Device), or a vehicle-mounted Device, and the specific type of the terminal 11 is not limited in the embodiment of the present invention. The network device 12 may be a Base Station and/or a core network element, wherein the Base Station may be a 5G or later-version Base Station (e.g., a gNB, a 5G NR NB, etc.), or a Base Station in other communication systems (e.g., an eNB, a WLAN access point, or other access points, etc.), wherein the Base Station may be referred to as a node B, an evolved node B, an access point, a Base Transceiver Station (BTS), a radio Base Station, a radio Transceiver, a Basic Service Set (BSS), an Extended Service Set (ESS), a node B, an evolved node B (eNB), a home node B, a home evolved node B, a WLAN access point, a WiFi node, or some other suitable terminology in the field, as long as the same technical effect is achieved, the Base Station is not limited to a specific technical vocabulary, it should be noted that, in the embodiment of the present invention only takes the Base Station in the NR system as an example, but does not limit the specific type of base station.

The base stations may communicate with the terminals 11 under the control of a base station controller, which may be part of the core network or some of the base stations in various examples. Some base stations may communicate control information or user data with the core network through a backhaul. In some examples, some of the base stations may communicate with each other, directly or indirectly, over backhaul links, which may be wired or wireless communication links. A wireless communication system may support operation on multiple carriers (waveform signals of different frequencies). A multi-carrier transmitter can transmit modulated signals on the multiple carriers simultaneously. For example, each communication link may be a multi-carrier signal modulated according to various radio technologies. Each modulated signal may be transmitted on a different carrier and may carry control information (e.g., reference signals, control channels, etc.), overhead information, data, and so on.

The base station may communicate wirelessly with the terminal 11 via one or more access point antennas. Each base station may provide communication coverage for a respective coverage area. The coverage area of an access point may be divided into sectors that form only a portion of the coverage area. A wireless communication system may include different types of base stations (e.g., macro, micro, or pico base stations). The base stations may also utilize different radio technologies, such as cellular or WLAN radio access technologies. The base stations may be associated with the same or different access networks or operator deployments. The coverage areas of different base stations (including coverage areas of base stations of the same or different types, coverage areas utilizing the same or different radio technologies, or coverage areas belonging to the same or different access networks) may overlap.

The communication links in a wireless communication system may comprise an Uplink for carrying Uplink (UL) transmissions (e.g., from terminal 11 to network device 12) or a Downlink for carrying Downlink (DL) transmissions (e.g., from network device 12 to terminal 11). The UL transmission may also be referred to as reverse link transmission, while the DL transmission may also be referred to as forward link transmission. Downlink transmissions may be made using licensed frequency bands, unlicensed frequency bands, or both. Similarly, uplink transmissions may be made using licensed frequency bands, unlicensed frequency bands, or both.

As described in the background art, in the prior art, the judgment of the network load state in the operation and maintenance of the mobile network generally depends on the expert experience of the network optimization engineer, and the accuracy of the judgment result is not high. In order to solve at least one of the above problems, embodiments of the present invention provide a method for detecting a cell abnormal state, which can simplify the detection of a network abnormal load state and improve the accuracy of a detection result.

Referring to fig. 2, a method for detecting a cell abnormal state according to an embodiment of the present invention may be applied to a network side device, such as the base station shown in fig. 1, and may also be applied to a core network element, where the method includes:

step 21, obtaining at least one key performance indicator KPI parameter set of a first cell collected in a preset time period, where the KPI parameter set includes values of multiple KPI parameters collected at corresponding sampling time points.

Specifically, each KPI parameter group corresponds to a sampling time point, the KPI parameter group is values of the KPI parameters of the first cell acquired at the corresponding sampling time point, and different KPI parameter groups are values of the KPI parameters of the first cell acquired at different time points.

Here, when the load of the first cell needs to be detected, the embodiment of the present invention obtains at least one KPI parameter set of the first cell, which is acquired within a preset time period, where each KPI parameter set includes a plurality of KPI parameters selected in advance. At the end of this document, specific examples of the plurality of KPI parameters are given.

And step 22, performing dimension reduction processing on each KPI parameter set of the first cell respectively, and generating at least one first dimension reduction vector of the first cell.

Here, in order to simplify subsequent computation, in the embodiment of the present invention, an auto-encoder may be used to perform dimension reduction on each KPI parameter set, and each KPI parameter set of the first cell is respectively input to a pre-trained auto-encoder to obtain a first dimension reduction vector output by the auto-encoder, so that each KPI parameter set may obtain a corresponding first dimension reduction vector, and a corresponding number of first dimension reduction vectors may be obtained for the at least one KPI parameter set. Through dimension reduction processing, the embodiment of the invention can simplify the operation amount of subsequent classification processing, simplify the detection of the abnormal load state of the network and improve the detection efficiency.

And step 23, inputting the at least one first dimension reduction vector into an evaluator, and generating evaluation information of the abnormal state of the first cell.

Here, the embodiment of the present invention may generate the evaluation information of the cell abnormal state by using an evaluator generated in advance. The evaluator is configured to perform clustering processing on the at least one first dimension reduction vector, determine and output a first category to which the at least one first dimension reduction vector belongs and corresponding label information, where the first category is one of a plurality of pre-generated categories, and the label information is used to indicate a cell load abnormal state of the corresponding category.

The embodiment of the invention can obtain a plurality of classifications in advance by utilizing various clustering algorithms. Then, in step 23, the same clustering algorithm is used to perform clustering processing on the at least one first dimension reduction vector obtained in step 22, determine a central point (referred to as a first central point for convenience of description) of the at least one first dimension reduction vector, calculate euclidean distances between the first central point and the central points of the pre-generated categories, and select a category corresponding to the shortest euclidean distance as the first category to which the at least one first dimension reduction vector belongs. In addition, the labeling information corresponding to the first category can be determined according to the pre-established correspondence between the labeling information and the categories. Here, the label information corresponding to each category may be provided by the network optimization engineer in advance and stored locally in the device. The labeling information may specifically include at least one of the following information: the evaluation value of the abnormal degree of the cell, the evaluation value of the load degree of the cell, the evaluation value of the degree to be optimized of the cell and remark information.

Optionally, before the processing in step 23 is executed, abnormal value detection may be performed on the at least one first dimension reduction vector, and outliers in the at least one first dimension reduction vector are removed, so as to obtain an updated first dimension reduction vector.

Through the steps, the embodiment of the invention can utilize a plurality of KPI parameters to detect the abnormal load of the cell, and can utilize more KPI parameters to detect due to the adoption of dimension reduction processing and clustering processing, thereby improving the accuracy of the detection result.

Prior to step 21, the self-encoder may be trained in advance and the plurality of classes may be generated according to the embodiment of the present invention. Training of the self-encoder and generation of the plurality of classes will be described below.

The embodiment of the invention can acquire a plurality of KPI parameter groups of a plurality of cells acquired at a plurality of sampling time points, wherein each KPI parameter group comprises a plurality of KPI parameters acquired by the same cell at one sampling time point.

For example, KPI parameters of a plurality of cells collected from the network manager are assumed to have been formatted in a data table form, and the ith row and the jth column represent the jth KPI parameter of the ith sampling point (i.e., cell). Wherein the unique identifier of each row is in the form of a unique name of the cell + a data acquisition timestamp. The method of the embodiment of the invention finally outputs the proposal information related to the certain cell and the load, which is deduced according to the KPI parameters in a certain period of time, and also can output the load proposal information of the state of any KPI acquisition time of the certain cell.

The overall flow of the scheme of the embodiment of the invention is shown in fig. 3, and can be divided into two stages: a training phase and an inference phase (the inference phase is sometimes referred to as an actual application phase herein), wherein the dimensionality reduction step needs to train a neural network of an auto-encoder, and the state mapping step needs manual intervention to complete labeling of category information. Wherein the dashed lines indicate the additional steps involved in the training phase.

As shown in FIG. 3, before the dimension reduction process is performed, the embodiment of the present invention may filter some noisy or low quality data through a data cleaning step. For example, removing KPI parameters that are too single in value according to the following calculation: and deleting the parameter when the frequency of occurrence of the value with the maximum occurrence frequency of a certain KPI parameter is greater than a preset threshold. The KPI parameters after data cleaning can then also be normalized to a distribution form of zero mean and unit variance.

Then, an autoencoder is trained by using the plurality of KPI parameter sets until a preset training end condition is reached (e.g., the training duration reaches a preset duration threshold, or the training data amount reaches a preset data amount threshold). As shown in fig. 3, the self-encoder generally includes an encoder and a decoder, each of which includes a multi-layer neural network, for example, the encoder and the decoder of fig. 3 include 3-layer neural networks, respectively. The first layer and the second layer respectively comprise a full connection layer, a PRelu activation layer and a batch normalization layer, and the third layer is a full connection network. Three layers (Layer 1-Layer 3) of the encoder are respectively provided with N₁、N₂、N₃A number of neurons, and the number of neurons of the decoder in the three layers (Layer 1-Layer 3) is N₂、N₁、N₀A, wherein N₀The data dimension of the input coder, namely the number of KPI parameters in each KPI data group. N is a radical of₁、N₂、N₃Typical values are 64, 16, 2.

The dimension reduction process of the auto-encoder differs in the actual application phase of step 22 and in the training phase before application. In the training stage, an encoder and a decoder of the self-encoder are considered as a whole, the decoder finally outputs an estimated value of original data, and a loss function is calculated according to the estimated value and the labeled value by taking the original data as a labeled value, wherein the loss function is a band l₁Smooth l of regularization term₂loss (as shown in the following equation 1), the loss function is compiled using Adam's algorithmThe decoder and decoder train. The training time batch normalization uses the mean and variance calculated by batch.

The formula for the calculation of the loss function is as follows:

wherein x is a label value,

is an estimate, and M is the dimension of x.

Is a₁A regularization term.

In the actual application stage, the encoder only works, the encoder processes input data, and the output of the encoder is a data dimension reduction result. The data dimension reduction result can be further processed by an abnormal point detection component to screen out a small number of outliers and sent to the next step. In the actual application stage, the mean and variance of the data overall calculation are used for batch normalization. Here, outliers, which are sometimes also called flying spots, outliers, or outliers, refer to individual values in a sample whose data significantly deviates from the rest of the observations. Outlier detection algorithms, such as boxed graphs, simple statistics, 3 sigma principles, etc., can be used to detect outliers.

For more details of the self-encoder, reference may be made to the description of the related art, which is not described in detail herein.

After the trained self-encoder is obtained, the embodiment of the present invention may obtain a plurality of dimension reduction vectors obtained after the plurality of KPI parameter sets are input to the trained self-encoder, and perform clustering processing on the plurality of dimension reduction vectors to obtain a plurality of categories. The clustering algorithm may employ a K-Means algorithm, for example, the number of classes divided may be

Wherein N is the number of groups of the plurality of KPI parameter sets. For example, inIn the data table, each row of data represents a plurality of KPI parameters of a certain cell at a certain data acquisition time, and at this time, N is the row number of the data table.

After obtaining a plurality of categories, the following processing may be further performed for each category:

1) selecting the parameter group belonging to the category from the plurality of KPI parameter groups, training a classifier by using the selected KPI parameter group, and calculating the weight of each KPI parameter in the KPI parameter groups according to the training result. The classifier here may be a base classifier, and specifically may be a CART decision tree.

Here, it is assumed that K categories are obtained after clustering, and category labels obtained by clustering can be mapped to various KPI parameter groups, so that KPI parameter groups under each category can be screened out through the category labels to obtain all KPI parameter groups under the category. Fewer samples under a certain category (i.e., KPI parameter set) may be processed to generate a sufficient number of samples by performing equalization on the samples.

In addition, here, a weight for each KPI parameter is calculated based on the training results. The weight of a KPI parameter represents the importance of the role a certain KPI parameter plays in distinguishing that category from other categories. The weight of a KPI parameter is obtained by calculating the sum of all values of purity degradation caused by using the node containing the KPI parameter.

2) Repeatedly performing the following steps until no KPI parameters exist in the KPI parameter set: and selecting a target KPI parameter with the maximum weight from the rest KPI parameters in the parameter group, adding the target KPI parameter into a parameter set, and deleting the target KPI parameter and the KPI parameters with the correlation with the target KPI parameter larger than a preset threshold from the parameter group.

Here, the weights may be sorted, and pearson correlation coefficients of all KPI parameters in the category may be calculated, then, starting from the KPI parameter with the highest weight, the KPI parameter with the highest weight is added to a parameter set, and then, other KPI parameters whose correlation coefficients with the KPI parameter with the highest weight are greater than a certain threshold (e.g., 0.8) are sequentially removed. By repeating the above processes, a parameter set including one or more KPI parameters can be finally obtained.

3) And selecting a preset number of KPI parameters from the parameter set according to the sequence of the weights from large to small as the KPI parameters of the category, and calculating the deviation degree of the statistical characteristics of the selected KPI parameters in the category relative to the statistical characteristics in all samples. Here, the all samples refer to the plurality of KPI parameter sets used in training the self-encoder.

Here, considering that there may be more KPI parameters in the parameter set, a preset number (for example, 10) may be set, and then all or part of the parameters are selected from the parameter set as the KPI parameters of the category, and the deviation degree of the statistical features of the selected KPI parameters in the category from the statistical features in all samples is calculated.

4) And outputting the KPI parameters, the weights and the deviation degrees of the category as the characteristics of the category.

Here, the KPI parameter, the weight, and the degree of deviation of each category are output as features of the category, and can be provided to a net optimization engineer for scoring.

5) Receiving the label information input aiming at the category, and establishing the corresponding relation between the label information and the category.

Here, the scoring result (i.e. label information) of the network optimization engineer for each category may be received, for example, the scoring item includes three items, which are the load abnormality degree, the load degree and the degree to be optimized. Each item has five options (extremely low, medium, high and extremely high), which correspond to 1-5 points respectively, and one possible sample form is shown in Table 1.

Degree of load abnormality	Degree of load	Degree of optimization	Remarks on the causes

TABLE 1

And mapping the scoring result of the current network engineer into the category. Therefore, under the condition that the work order information exists, the work order information can be extracted for comprehensive mapping.

Specific examples of several analyses are provided below:

cell analysis example:

the method of the embodiment of the invention can output the category of a certain cell and the corresponding marking information thereof, and the information can be provided for a network optimization engineer to make a decision.

For example, for a certain cell, the cell analysis comprises the following operation steps:

1) the statistical abnormality degree > is the ratio of the state of 3 on the cell time axis, and when the ratio is greater than 10%, the indication information of the abnormal cell is output, and the labeling reason corresponding to the abnormal state with the largest ratio is output.

2) And counting the occupation ratio of the abnormal degree > 3 and the high load degree > 3, and outputting the indication information of the high load cell when the occupation ratio is more than 10%, and outputting the labeling reason corresponding to the high load abnormal state with the largest occupation ratio.

3) And counting the abnormal degree > -3, and the degree to be optimized > -3, when the percentage is more than 10%, outputting the indication information of the cell to be optimized, and outputting the labeling reason corresponding to the abnormal state to be optimized with the largest percentage.

4) Intermediate calculation results such as KPI parameter weight of the category, state time sequence and the like are also provided for engineers to consult as selectable items.

Cell timing analysis example:

since the calculated states have timing characteristics, their timing information can be used for further exception classification.

Example 1: before clustering, the continuous P time sequence states are regarded as joint states for clustering. P is a predetermined positive integer, and may be 2, for example.

Example 2: when determining the cell is abnormal, the cell is determined to be abnormal when an abnormal state continuously appears Q times, where Q is a preset positive integer, and may be 4, for example.

From the above, it can be seen that the embodiments of the present invention have at least the following advantages:

1) compared with the existing threshold-based scheme, the unsupervised cell abnormal state detection method can utilize any number of parameters, self-adapt to network change and detect more accurately without manpower.

2) Compared with the output of the cell anomaly detection based on the existing threshold value, the embodiment of the invention can position a plurality of problems at one time, can output a plurality of candidate anomalies, and can adjust the precision ratio or recall ratio through the distance threshold value by weight sorting.

3) Compared with the existing method based on manual experience, the embodiment of the invention automatically maps the human experience in the actual application stage without manual processing.

4) Compared with the conventional method for screening key KPIs based on statistical mean deviation, the method and the device can detect abnormal scenes irrelevant to the statistical deviation and the category. For example, some parameters have large fluctuations, and it is possible that a certain type of abnormality is attributed to the parameter but not to the main cause of the abnormality.

5) Compared with the conventional statistical method, the problem positioning accuracy can be improved, and a plurality of potential candidate reasons can be output.

6) Compared with the prior method, the method has the advantages that,number of cluster categories to

And determining to match the actual abnormal category number of the KPI parameters of the current network cell level.

7) Compared with the existing manual labeling scheme, the embodiment of the invention has more structuralization compared with the existing operation and maintenance labeling data, and is beneficial to data training.

Various methods of embodiments of the present invention have been described above. An apparatus for carrying out the above method is further provided below.

Referring to fig. 5, an embodiment of the present invention provides a device 50 for detecting a cell abnormal state, including:

a parameter obtaining module 51, configured to obtain at least one key performance indicator KPI parameter set of a first cell, where the KPI parameter set is acquired within a preset time period, and the KPI parameter set includes values of multiple KPI parameters acquired at corresponding sampling time points;

a dimension reduction processing module 52, configured to perform dimension reduction processing on each KPI parameter set of the first cell, respectively, to generate at least one first dimension reduction vector of the first cell;

and the label processing module 53 is configured to input the at least one first dimension reduction vector to an evaluator, and generate evaluation information of the abnormal state of the first cell.

Optionally, the parameter obtaining module 51 is further configured to input each KPI parameter set of the first cell to a pre-trained self-encoder respectively, so as to obtain a first dimension reduction vector output by the self-encoder.

Optionally, the labeling processing module 53 is further configured to perform clustering processing on the at least one first dimension reduction vector through the evaluator, determine and output a first category to which the at least one first dimension reduction vector belongs and corresponding labeling information, where the first category is one of a plurality of pre-generated categories, and the labeling information is used to indicate a cell abnormal state of the corresponding category.

Optionally, the detection device further includes:

Optionally, the number of the plurality of categories is

Wherein N is the number of the multiple groups of KPI parameters.

Optionally, the detection device further includes:

It should be noted that the apparatus in this embodiment is a device corresponding to the method shown in fig. 2, and the implementation manners in the above embodiments are all applicable to the embodiment of this device, and the same technical effects can be achieved. The device provided by the embodiment of the present invention can implement all the method steps implemented by the method embodiment, and can achieve the same technical effects, and detailed descriptions of the same parts and beneficial effects as the method embodiment in this embodiment are not repeated herein.

Referring to fig. 6, an embodiment of the present invention provides a structural schematic diagram of a device 600 for detecting a cell abnormal state, including: a processor 601, a transceiver 602, a memory 603, and a bus interface, wherein:

in this embodiment of the present invention, the apparatus 600 for detecting a cell abnormal state further includes: a program stored in the memory 603 and executable on the processor 601, the program when executed by the processor 601 implementing the steps of:

acquiring at least one Key Performance Indicator (KPI) parameter set of a first cell acquired within a preset time period, wherein the KPI parameter set comprises values of a plurality of KPI parameters acquired at corresponding sampling time points; performing dimension reduction processing on each KPI parameter group of the first cell respectively to generate at least one first dimension reduction vector of the first cell; and inputting the at least one first dimension reduction vector to an evaluator, and generating evaluation information of the abnormal state of the first cell.

Optionally, the processor further implements the following steps when executing the program:

and clustering the at least one first dimension reduction vector through the evaluator, and determining and outputting a first category to which the at least one first dimension reduction vector belongs and corresponding labeling information, wherein the first category is one of a plurality of pre-generated categories, and the labeling information is used for indicating the abnormal state of the cell of the corresponding category.

Optionally, the processor further implements the following steps when executing the program: obtaining a plurality of dimensionality reduction vectors obtained after the plurality of KPI parameter sets are input to the trained self-encoder; and clustering the dimension reduction vectors to obtain a plurality of categories.

Optionally, the processor further implements the following steps when executing the program: after obtaining a plurality of categories, the following processing is respectively executed for each category:

Optionally, the number of the plurality of categories is

Wherein N is the number of groups of the plurality of KPI parameter groups.

Optionally, the processor further implements the following steps when executing the program: clustering the at least one first dimension reduction vector, determining a first central point of the at least one first dimension reduction vector, calculating Euclidean distances between the first central point and the central points of all categories, and selecting a category corresponding to the shortest Euclidean distance as a first category to which the at least one first dimension reduction vector belongs; and determining the labeling information corresponding to the first category according to the pre-established correspondence between the labeling information and the categories.

Optionally, the processor further implements the following steps when executing the program: and carrying out abnormal value detection on the at least one first dimension reduction vector, and removing outliers in the at least one first dimension reduction vector to obtain an updated first dimension reduction vector.

In fig. 6, the bus architecture may include any number of interconnected buses and bridges, with one or more processors represented by processor 601 and various circuits of memory represented by memory 603 being linked together. The bus architecture may also link together various other circuits such as peripherals, voltage regulators, power management circuits, and the like, which are well known in the art, and therefore, will not be described any further herein. The bus interface provides an interface. The transceiver 602 may be a number of elements including a transmitter and a receiver that provide a means for communicating with various other apparatus over a transmission medium.

The processor 601 is responsible for managing the bus architecture and general processing, and the memory 603 may store data used by the processor 601 in performing operations.

It should be noted that the apparatus in this embodiment is an apparatus corresponding to the method shown in fig. 2, and the implementation manners in the above embodiments are all applicable to the embodiment of this apparatus, and the same technical effects can be achieved. In the device, the transceiver 602 and the memory 603, and the transceiver 602 and the processor 601 may be communicatively connected through a bus interface, and the functions of the processor 601 may also be implemented by the transceiver 602, and the functions of the transceiver 602 may also be implemented by the processor 601. It should be noted that, the apparatus provided in the embodiment of the present invention can implement all the method steps implemented by the method embodiment and achieve the same technical effect, and detailed descriptions of the same parts and beneficial effects as the method embodiment in this embodiment are omitted here.

In some embodiments of the invention, there is also provided a computer readable storage medium having a program stored thereon, which when executed by a processor, performs the steps of:

When executed by the processor, the program can implement all the implementation manners in the method for detecting a cell abnormal state of a device, and can achieve the same technical effect, and is not described herein again to avoid repetition.

The following provides a list of cell-level KPI parameters that may be used in the embodiments of the present invention, and the present invention may use some or all of the following parameters:

an uplink average MCS;

RSSI_PUCCH_AVG；

a downlink double-current ratio;

a downlink average MCS;

SINR_PUSCH_AVG；

downlink transmission efficiency (bytes/prb);

AVERAGE_CQI；

downlink retransmission rate LTE _5208 a;

RSSI_PUSCH_AVG；

downstream traffic (MB);

PHR mean value;

a handover execution attempt;

the number of downlink packets in a cell;

the uplink packet loss rate of the cell user plane;

VoLTE uplink packet loss rate;

RRC/RACH ratio;

downlink maximum rate (Mbit/s);

downlink average rate (Mbit/s);

PDCCH _ CCE average utilization;

MAC layer uplink block error rate;

the number of PRACH requests;

an uplink retransmission rate LTE _5207 b;

the number of uplink packets of the cell;

SINR_PUCCH_AVG；

RRC average connection number;

a CA capability terminal;

AGG8 ratio (%);

AGG1_USED_PDCCH；

AGG4_USED_PDCCH；

DRB establishment success number;

UE_PER_DL_TTI_AVG；

an effective RRC connection average;

RRC maximum connection number;

the PRACH establishment success rate;

DRB establishment success rate;

wireless disconnection rate;

upstream traffic (MB);

AGG2_USED_PDCCH；

a handover-in success rate;

average utilization rate of uplink PRB;

uplink transmission efficiency (bytes/prb);

number of handover failures;

the uplink packet loss number (QCI ═ 1) of the cell;

a PRACH load;

the PRACH success times;

an uplink interference value;

AGG8_USED_PDCCH；

the uplink packet loss number of the cell;

UE_PER_UL_TTI_AVG；

the number of successful switching-in times;

UE_PER_UL_TTI_MAX；

the number of RRC establishment requests;

the downlink packet loss rate of the cell user plane;

the downlink block error rate of the MAC layer;

DRB establishes request number;

HO_PREP_IN_SUCC；

UE_PER_DL_TTI_MAX；

the preparation switching success rate is included;

the ERAB establishment request times;

average utilization rate of downlink PRB;

maximum number of active RRC connections;

an uplink average rate (Mbit/s);

including preparing for a handover attempt;

the preparation switching success rate is included;

a number of handover attempts;

e _ RAB establishment success times QCI _ 1;

cell downlink packet number (QCI ═ 1);

the switching is successfully executed;

wireless disconnection rate _ denominator;

VoLTE downlink packet loss rate;

the handover execution fails;

the number of successful RRC establishment times;

the downlink packet loss number of the cell;

a handover execution success rate;

VoLTE voice traffic (Erl);

containing a prepare handover failure;

an uplink maximum rate (Mbit/s);

ENB_INIT_TO_IDLE_RNL；

e _ RAB establishment success rate QCI _ 2;

FAIL_ENB_HO_PREP_AC；

SIGN_CONN_ESTAB_REJ_NEMG；

the number of ERAB build failures;

the number of successful ERAB establishments;

ERAB_ADD_SETUP_FAIL_RNL_MOB；

eVRCC switching duty ratio;

ERAB_INI_SETUP_FAIL_TNL_TRU；

ERAB_ADD_SETUP_FAIL_TNL_TRU；

ENB_INIT_TO_IDLE_OTHER；

eVRCC switching success times;

eVRCC switching request times;

eVRCC switching success rate;

FAIL_ENB_HO_PREP_QCI；

e _ RAB abnormal release times QCI _ 1;

e _ RAB establishment request times QCI _ 1;

e _ RAB disconnection rate QCI _ 1;

e _ RAB establishment request times QCI _ 2;

VoLTE video traffic (Erl);

e _ RAB establishment success times QCI _ 2;

RRC establishment success rate;

the number of congestions;

wireless drop rate _ numerator;

load reason handover request times;

load reason handover success times;

PDCCH congestion rate (%);

number of connection failures;

FAIL_ENB_HO_PREP_TIME；

the CA configures the number of users;

the number of downlink packet losses (QCI ═ 1) in the cell;

a cell uplink packet number (QCI ═ 1);

the number of CA activated users;

e _ RAB establishment success rate QCI _ 1;

including the number of attempts to prepare for handoff;

wireless call completing rate;

the E _ RAB disconnection rate denominator QCI _ 1;

the number of times of RRC establishment failure;

SIGN_EST_F_RRCCOMPL_MISSING；

FAIL_ENB_HO_PREP_OTHER；

the ERAB establishment success rate;

cell availability.

Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.

It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.

In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment of the present invention.

In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit.

The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: various media capable of storing program codes, such as a U disk, a removable hard disk, a ROM, a RAM, a magnetic disk, or an optical disk.

The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention, and all the changes or substitutions should be covered within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A method for detecting abnormal conditions of a cell, comprising:

2. The method of claim 1, wherein performing the respective dimension reduction on each KPI parameter set of the first cell comprises:

3. The method of claim 1,

the evaluator is configured to perform clustering processing on the at least one first dimension reduction vector, determine and output a first category to which the at least one first dimension reduction vector belongs and corresponding label information, where the first category is one of a plurality of pre-generated categories, and the label information is used to indicate a cell abnormal state of the corresponding category.

4. The method according to any of claims 1 to 3, wherein prior to obtaining at least one KPI parameter set for the first cell, the method further comprises:

5. The method of claim 4, further comprising:

6. The method of claim 5, wherein after obtaining a plurality of categories, the method further comprises:

for each category, the following processing is performed:

7. The method of claim 6, wherein the annotation information comprises at least one of: the evaluation value of the abnormal degree of the cell, the evaluation value of the load degree of the cell, the evaluation value of the degree to be optimized of the cell and remark information.

8. The method of claim 5, wherein the number of the plurality of categories is

Wherein N is the number of groups of the plurality of KPI parameter groups.

9. The method of claim 3, wherein determining the first class to which the at least one first dimension-reducing vector belongs and its corresponding annotation information comprises:

10. The method of claim 3, wherein prior to the step of clustering the at least one first reduced-dimension vector, the method further comprises:

11. A device for detecting an abnormal state of a cell, comprising:

12. The detection apparatus of claim 11,

the dimension reduction processing module is further configured to input each KPI parameter set of the first cell to a pre-trained self-encoder respectively, and obtain a first dimension reduction vector output by the self-encoder.

13. The detection apparatus of claim 11,

the label processing module is further configured to perform clustering processing on the at least one first dimension reduction vector through the evaluator, determine and output a first category to which the at least one first dimension reduction vector belongs and label information corresponding to the first category, where the first category is one of a plurality of pre-generated categories, and the label information is used to indicate a cell abnormal state of the corresponding category.

14. The detection apparatus according to any one of claims 11 to 13, further comprising:

15. The detection device of claim 14, further comprising:

16. The detection device of claim 15, further comprising:

17. The detection apparatus of claim 13,

the labeling processing module is further configured to perform clustering processing on the at least one first dimension reduction vector, determine a first center point of the at least one first dimension reduction vector, calculate euclidean distances between the first center point and center points of the categories, and select a category corresponding to the shortest euclidean distance as a first category to which the at least one first dimension reduction vector belongs; and determining the labeling information corresponding to the first category according to the pre-established correspondence between the labeling information and the categories.

18. The detection device of claim 13, further comprising:

19. A device for detecting an abnormal state of a cell, comprising a transceiver and a processor, wherein,

20. A computer-readable storage medium, having stored thereon a computer program which, when being executed by a processor, carries out the steps of the method of detecting an abnormal state of a cell according to any one of claims 1 to 10.