CN112153685B - RRC fault detection method and device - Google Patents
RRC fault detection method and device Download PDFInfo
- Publication number
- CN112153685B CN112153685B CN201910560786.5A CN201910560786A CN112153685B CN 112153685 B CN112153685 B CN 112153685B CN 201910560786 A CN201910560786 A CN 201910560786A CN 112153685 B CN112153685 B CN 112153685B
- Authority
- CN
- China
- Prior art keywords
- rrc
- threshold
- sample number
- candidate
- clustering
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/04—Arrangements for maintaining operational condition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Probability & Statistics with Applications (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The embodiment of the invention provides a method and a device for detecting RRC fault, wherein the method comprises the following steps: acquiring an RRC performance index data set containing a plurality of groups of RRC performance index data; performing DBSCAN clustering on the RRC performance index data set to obtain a clustering result of each RRC performance index data; and if the clustering result of any RRC performance index data is a noise point, confirming that the RRC corresponding to any RRC performance index data has a fault. According to the method and the device provided by the embodiment of the invention, the DBSCAN clustering is applied to the RRC fault detection in the wireless network, the DBSCAN clustering is an unsupervised learning process, manual intervention is not needed, and a threshold value is not needed to be set, so that the addition of subjective factors in the RRC fault detection process is effectively avoided, and the reliability and the accuracy of the RRC fault detection are improved.
Description
Technical Field
The present invention relates to the field of mobile communications technologies, and in particular, to a method and an apparatus for detecting an RRC fault.
Background
While the mobile communication network architecture and the Key technology are rapidly developed, a large number of configuration parameters and KPIs (Key Performance indicators) are brought, so that the difficulty of network optimization and fault detection is greatly increased.
RRC (Radio Resource Control) fault detection is an important component of network fault detection. The existing RRC fault detection method is to set a fixed and unchangeable threshold value aiming at key performance indexes related to the wireless network and the RRC according to past experience, calculate to obtain the key performance indexes, and judge whether the RRC needs to be alarmed according to whether the key performance indexes reach the constant threshold value or not.
The existing RRC fault detection fixed threshold method has the problems that the performance of a detection result cannot be judged and the judgment is completely dependent on subjective experience, so that the reliability of the obtained RRC fault detection result is low.
Disclosure of Invention
The embodiment of the invention provides a method and a device for detecting an RRC fault, which are used for solving the problems of strong subjectivity and low reliability of the conventional method for detecting the fixed threshold of the RRC fault.
In a first aspect, an embodiment of the present invention provides a method for detecting an RRC fault, including:
acquiring an RRC performance index data set containing a plurality of groups of RRC performance index data;
performing DBSCAN clustering on the RRC performance index data set to obtain a clustering result of each RRC performance index data;
and if the clustering result of any RRC performance index data is a noise point, confirming that the RRC corresponding to any RRC performance index data has a fault.
Preferably, the performing DBSCAN clustering on the RRC performance indicator data set to obtain a clustering result of each RRC performance indicator data specifically includes:
acquiring a distance threshold and a neighborhood sample number threshold for DBSCAN clustering based on the RRC performance index data set, the profile coefficient and the estimated alarm rate;
and carrying out DBSCAN clustering on the RRC performance index data set based on the distance threshold and the neighborhood sample number threshold, and obtaining a clustering result of each RRC performance index data.
Preferably, the obtaining a distance threshold and a neighborhood sample number threshold for DBSCAN clustering based on the RRC performance indicator dataset, the profile coefficient and the estimated alarm rate specifically includes:
acquiring an interval of the neighborhood sample number threshold based on the data quantity and the data dimension of the RRC performance index data set; a plurality of candidate neighborhood sample number thresholds are included within the interval;
acquiring the alarm rate corresponding to any candidate neighborhood sample number threshold value and meeting the maximum distance threshold value of the estimated alarm rate;
acquiring a contour coefficient corresponding to the threshold of the number of any candidate neighborhood samples and the maximum distance threshold;
and selecting the candidate neighborhood sample number threshold and the maximum distance threshold corresponding to the maximum contour coefficient as the neighborhood sample number threshold and the distance threshold.
Preferably, the obtaining the interval of the neighborhood sample number threshold based on the data size and the data dimension of the RRC performance indicator data set specifically includes:
taking the product of the data volume and a preset percentage as the maximum value of the neighborhood sample number threshold;
acquiring a first minimum value of the preset neighborhood sample number threshold;
adding 1 to the data dimension as a second minimum of the neighborhood sample number threshold;
obtaining an interval of the neighborhood sample count threshold based on the maximum value, the first minimum value, and the second minimum value of the neighborhood sample count threshold.
Preferably, the obtaining of the alarm rate corresponding to any one of the candidate neighborhood sample number thresholds satisfies the maximum distance threshold of the estimated alarm rate, specifically includes:
obtaining the estimated clustering quantity corresponding to the estimated alarm rate;
acquiring a correlation curve of a candidate distance threshold value and a cluster number corresponding to any one candidate neighborhood sample number threshold value;
selecting a candidate distance threshold corresponding to each estimated clustering quantity from the association curve;
and selecting the maximum candidate distance threshold as the maximum distance threshold corresponding to any one of the candidate neighborhood sample number thresholds.
Preferably, the acquiring the contour coefficient corresponding to any one of the candidate neighborhood sample number threshold and the maximum distance threshold specifically includes:
based on any candidate neighborhood sample number threshold and the maximum distance threshold, carrying out DBSCAN clustering on the RRC performance index data set to obtain a candidate clustering result;
obtaining the contour coefficient based on the candidate clustering result by the following formula:
wherein s (i) is a contour coefficient of the RRC performance indicator data i, a (i) is an average value of distances from i to each of the rest of the data in the cluster to which the data belongs, and b (i) is a minimum value of the distances from i to each of the rest of the clusters.
Preferably, the RRC performance indicator data includes an RRC access success rate and a number of failures.
In a second aspect, an embodiment of the present invention provides an RRC failure detection apparatus, including:
a data set obtaining unit, configured to obtain an RRC performance indicator data set including multiple sets of RRC performance indicator data;
the clustering unit is used for carrying out DBSCAN clustering on the RRC performance index data set to obtain a clustering result of each RRC performance index data;
and the fault judging unit is used for confirming that the RRC corresponding to any RRC performance index data has a fault if the clustering result of any RRC performance index data is a noise point.
In a third aspect, an embodiment of the present invention provides an electronic device, including a processor, a communication interface, a memory, and a bus, where the processor and the communication interface, the memory complete communication with each other through the bus, and the processor may call a logic instruction in the memory to perform the steps of the method provided in the first aspect.
In a fourth aspect, an embodiment of the present invention provides a non-transitory computer readable storage medium, on which a computer program is stored, which when executed by a processor, implements the steps of the method as provided in the first aspect.
According to the RRC fault detection method and device provided by the embodiment of the invention, the DBSCAN clustering is applied to the RRC fault detection in the wireless network, the DBSCAN clustering is an unsupervised learning process, manual intervention is not needed, and a threshold value is not needed to be set, so that the addition of subjective factors in the RRC fault detection process is effectively avoided, and the reliability and accuracy of the RRC fault detection are improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and those skilled in the art can also obtain other drawings according to the drawings without creative efforts.
Fig. 1 is a schematic flowchart of an RRC fault detection method according to an embodiment of the present invention;
fig. 2 is a schematic flow diagram of a dbs can parameter adaptive method according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of an RRC failure detection apparatus according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The existing RRC fault detection judges whether the RRC needs to be alarmed or not through a fixed threshold value, detection is carried out completely by subjective experience, and reliability is low. In view of the above, an embodiment of the present invention provides a method for detecting an RRC failure. Fig. 1 is a schematic flow chart of an RRC failure detection method according to an embodiment of the present invention, as shown in fig. 1, the method includes:
Specifically, the RRC performance index data set includes multiple sets of RRC performance index data, each set of RRC performance index data corresponds to an RRC that needs to perform fault detection, the RRC performance index data is a specific data value of a corresponding RRC performance index, each set of RRC performance index data may include specific data values of one or more performance indexes, for example, any RRC performance index data corresponding to an RRC includes RRC request times and RRC request failure times of the RRC within a preset time, and for example, any RRC performance index data corresponding to an RRC includes an RRC access success rate of the RRC within a preset time.
Specifically, in an RRC fault detection scenario, RRC performance index data corresponding to different base stations and at different times will reach orders of magnitude of tens of thousands. After the RRC performance index data set reaches a sufficiently large scale both in time and in geographical scale, most of the RRC performance index data in the RRC performance index data set reacts to the RRC in a normal state, and only a very small number of the RRC performance index data reacts to the RRC in a failure state. Further, the distribution of the RRC performance indicator data in the RRC normal state has aggregation, that is, the distance between the RRC performance indicator data in the RRC normal state is closer than the spatial density of the entire RRC performance indicator data set, and the distribution of the RRC performance indicator data in the RRC failure state is relatively dispersed. Therefore, the DBSCAN is adopted to cluster the RRC performance index data set in the embodiment of the invention.
DBSCAN (sensitivity-Based Spatial Clustering of Applications with Noise) is a Density-Based Clustering algorithm. In this algorithm, the maximum set of density-connected points is one cluster. The algorithm utilizes a clustering concept based on density, and the number of points contained in a certain area in a clustering space is required to be not less than a preset neighborhood sample number threshold. The method can find clusters in any shape in a noisy spatial database, can connect adjacent regions with high enough density, can effectively process abnormal data, and is mainly used for clustering spatial data.
Clustering the RRC performance index data set by using the DBSCAN, namely using the RRC performance index data set as a spatial database, using each RRC performance index data as a point, randomly selecting one RRC performance index data from the RRC performance index data set, judging whether the number of the RRC performance index data contained in the neighborhood of the RRC performance index data is more than or equal to a neighborhood sample number threshold MinPts or not based on a preset distance threshold Eps, and further determining whether the RRC performance index data is a core point or not; if the RRC performance index data is a core point, acquiring each RRC performance index data with the reachable density of the RRC performance index data in the RRC performance index data set to form a cluster; otherwise, the above operation is performed for the next RRC performance indicator data.
Wherein the neighborhood of any RRC performance indicator data is the region within the distance threshold Eps of that RRC performance indicator data. The core point is RRC performance index data in which the number of pieces of RRC performance index data included in the neighborhood is equal to or greater than the neighborhood sample number threshold value MinPts, the boundary point is RRC performance index data that does not belong to the core point but falls within the neighborhood of a certain core point, and the noise point is RRC performance index data that does not belong to either the core point or the boundary point. Direct density reachability means that if any RRC performance indicator data is in the neighborhood of any core point, the RRC performance indicator data is directly density reachable from the core point. The density reachable means that an object chain exists, a plurality of RRC performance index data are sequentially arranged in the object chain, any RRC performance index data in the object chain is directly density reachable from the next RRC performance index data, and the first RRC performance index data is density reachable from the last RRC performance index data.
By performing DBSCAN clustering on the RRC performance index data set, the clustering result of each RRC performance index data can be obtained. Here, the clustering result is used to indicate whether the RRC performance indicator data is a noise point, and the clustering result may be a core point, a boundary point, or a noise point, or may be a non-noise point or a noise point, which is not specifically limited in this embodiment of the present invention.
Specifically, because the distribution of the RRC performance index data in the normal RRC state has aggregation, and under DBSCAN clustering, noise points are distributed and dispersed RRC performance index data, after the clustering result is obtained, it is determined that the RRC corresponding to the RRC performance index data of which the clustering result is a noise point has a fault.
According to the method provided by the embodiment of the invention, the DBSCAN clustering is applied to the RRC fault detection in the wireless network, the DBSCAN clustering is an unsupervised learning process, manual intervention is not needed, and a threshold value is not needed to be set, so that the addition of subjective factors in the RRC fault detection process is effectively avoided, and the reliability and the accuracy of the RRC fault detection are improved.
DBSCAN is an unsupervised learning algorithm, which is essentially an approximation to the true model of the problem. What is expected to be obtained in the RRC fault detection is a model relationship between RRC performance index data and RRC alarm rate, and since the above model relationship is unknown, an actual error of the fault detection cannot be obtained only by the existing DBSCAN. To this end, based on any of the above embodiments, in the method, the step 120 specifically includes:
and step 121, acquiring a distance threshold and a neighborhood sample number threshold for DBSCAN clustering based on the RRC performance index data set, the contour coefficient and the estimated alarm rate.
Here, the contour Coefficient (Silhouette Coefficient) is an evaluation method for evaluating the clustering effect. The value of the contour coefficient is between-1, and the closer to 1, the better the cohesion and separation of the cluster, and the better the fault detection effect based on DBSCAN.
The estimated alarm rate is a preset alarm rate interval, and the alarm rate is the ratio of the number of the failed RRC to the total number of the RRC, namely the ratio of the number of the RRC performance index data corresponding to the failed RRC to the data quantity of the RRC performance index data set. And if the alarm rate obtained according to the DBSCAN clustering meets the estimated alarm rate, the actual error of fault detection based on the DBSCAN is smaller. Here, the predicted alarm rate may be obtained based on an alarm rate of the existing RRC fault detection.
On the basis of the RRC performance index data set, the DBSCAN is limited through the profile coefficient and the estimated alarm rate, and then the distance threshold Eps and the neighborhood sample number threshold MinPts which have good clustering effect and meet the estimated alarm rate are obtained.
And step 122, performing DBSCAN clustering on the RRC performance index data sets based on the distance threshold and the neighborhood sample number threshold, and acquiring a clustering result of each RRC performance index data.
The method provided by the embodiment of the invention obtains the distance threshold and the neighborhood sample number threshold based on the profile coefficient and the estimated alarm rate and uses the distance threshold and the neighborhood sample number threshold for DBSCAN clustering, so that the experience risk and the confidence risk in the clustering process can be effectively reduced, and the accuracy of the RRC detection result is improved.
In addition, in the DBSCAN clustering, different parameter combinations of the distance threshold value Eps and the neighborhood sample number threshold value MinPts have a large influence on the final clustering result. Artificially specifying the parameter values of the distance threshold value Eps and the neighborhood sample number threshold value MinPts may cause that the DBSCAN clustering cannot be applied to different data sets, and even cause that the clustering result is not available. In contrast, based on any of the above embodiments, step 121 specifically includes:
step 1211, obtaining a neighborhood sample number threshold interval based on the data quantity and the data dimension of the RRC performance index data set; the interval includes a plurality of candidate neighborhood sample number thresholds.
Here, the data size of the RRC performance indicator data set refers to the number of RRC performance indicator data in the RRC performance indicator data set, and the data dimension refers to the number of performance indicators included in the RRC performance indicator data, for example, if the RRC performance indicator data is composed of the number of RRC request times and the number of RRC request failure times, the corresponding data dimension is 2, and if the RRC performance indicator data is composed of the number of RRC request times, the number of RRC request failure times, and the RRC access success rate, the corresponding data dimension is 3.
The neighborhood sample number threshold interval is a range of neighborhood sample number thresholds, and the candidate neighborhood sample number threshold is a threshold belonging to the range of neighborhood sample number thresholds, for example, when the neighborhood sample number threshold interval is [3,7], the corresponding candidate neighborhood sample number thresholds are 3, 4, 5, 6, and 7. There are various methods for obtaining the interval of the neighborhood sample number threshold. For example, the relationship between the data amount, the data dimension, and the interval is preset, the corresponding interval may be directly obtained according to the data amount and the data dimension, and for example, the relationship between the data amount and the interval upper limit and the relationship between the data dimension and the interval lower limit are preset, the interval may be obtained according to the data amount, the interval upper limit, the interval lower limit, and the interval, which is not specifically limited in the embodiment of the present invention.
In step 1212, the alarm rate corresponding to any candidate neighborhood sample number threshold is obtained to satisfy the maximum distance threshold of the estimated alarm rate.
In particular, there may be multiple candidate distance thresholds for any of the candidate neighborhood sample number thresholds. Based on any candidate neighborhood sample number threshold and any candidate distance threshold, DBSCAN clustering can be performed on the RRC performance index data set, and then RRC performance index data with corresponding RRC faults are obtained, and therefore the alarm rate is obtained.
Aiming at any candidate neighborhood sample number threshold, the alarm rate under the combination of the candidate neighborhood sample number threshold and each corresponding candidate distance threshold can be obtained, the combination of which the alarm rate meets the estimated alarm rate is selected from the candidate neighborhood sample number threshold, and the maximum candidate distance threshold is selected from the combination, namely the alarm rate meets the maximum distance threshold of the estimated alarm rate.
Step 1213, obtain the contour coefficients corresponding to the candidate neighborhood sample number threshold and the maximum distance threshold.
Specifically, for any candidate neighborhood sample number threshold, after the maximum distance threshold corresponding to the candidate neighborhood sample number threshold is obtained, the profile coefficient for performing the DBSCAN clustering under the candidate neighborhood sample number threshold and the maximum distance threshold is obtained.
Step 1214, selecting the threshold of the number of candidate neighborhood samples and the maximum distance threshold corresponding to the maximum contour coefficient as the threshold of the number of neighborhood samples and the distance threshold.
Specifically, after obtaining the contour coefficient of each candidate neighborhood sample number threshold and the corresponding maximum distance threshold, the candidate neighborhood sample number threshold with the largest contour coefficient and the maximum distance threshold are selected from the contour coefficients, the candidate neighborhood sample number threshold is used as a neighborhood sample number threshold MinPts, and the maximum distance threshold is used as a distance threshold Eps for RRC fault detection.
According to the method provided by the embodiment of the invention, the distance threshold and the neighborhood sample number threshold are obtained under the limitation of the contour coefficient and the estimation alarm rate, so that the distance threshold and the neighborhood sample number threshold can be determined in a self-adaptive manner according to specific requirements in an RRC fault detection scene, and the problems of over-strong subjectivity and reduction of reliability and accuracy of fault detection caused by artificial parameter determination are solved.
Based on any of the above embodiments, in the method, step 1211 specifically includes: taking the product of the data volume and the preset percentage as the maximum value of the neighborhood sample number threshold; acquiring a first minimum value of a preset neighborhood sample number threshold; adding 1 to the data dimension to serve as a second minimum value of the neighborhood sample number threshold; and acquiring an interval of the neighborhood sample number threshold value based on the maximum value, the first minimum value and the second minimum value of the neighborhood sample number threshold value.
Specifically, the preset percentage is a preset percentage, for example, 2%, that is, 2% of the data amount is used as the maximum value of the neighborhood sample number threshold. The first minimum value is preset, and in DBSCAN, it is not reasonable to set the threshold value MinPts to 1, because if it is set to 1, each RRC performance indicator data is a cluster, and if MinPts is less than or equal to 2, the result is the same as the nearest neighbor result of the hierarchy distance, therefore, MinPts must be selected to be greater than or equal to 3, and the first minimum value can be set to 3. The second minimum value is +1, for example, when the data dimension is 2, the second minimum value is 3. Of the first minimum value and the second minimum value, the larger value is taken as the minimum value of the interval, and the interval can be obtained based on the maximum value.
Based on any of the above embodiments, in the method, step 1212 specifically includes: obtaining the estimated clustering quantity corresponding to the estimated alarm rate; acquiring a correlation curve of a candidate distance threshold value and a cluster number corresponding to any candidate neighborhood sample number threshold value; selecting a candidate distance threshold corresponding to each estimated clustering quantity from the association curve; and selecting the maximum candidate distance threshold as the maximum distance threshold corresponding to the candidate neighborhood sample number threshold.
Specifically, given an estimated alarm rate, a corresponding estimated cluster number may be obtained based on the estimated alarm rate and the data volume of the RRC performance indicator data set. Here, the estimated cluster number is a cluster number in which the corresponding alarm rate can satisfy the estimated alarm rate, and the estimated cluster number may be one or more.
And aiming at any candidate neighborhood sample number threshold, acquiring a correlation curve of the candidate distance threshold and the cluster number corresponding to the candidate neighborhood sample number threshold, wherein the correlation curve comprises the cluster number under the combination of the candidate neighborhood sample number threshold and each corresponding candidate distance threshold. On the basis, the candidate distance threshold corresponding to the estimated clustering number can be selected from the association curve, namely the alarm rate can meet each candidate distance threshold of the estimated alarm rate.
After each candidate distance threshold value of which the alarm rate can meet the estimated alarm rate is obtained, the maximum candidate distance threshold value is selected from the candidate distance threshold values, namely the maximum distance threshold value corresponding to the candidate neighborhood sample number threshold value.
Based on any of the above embodiments, in the method, step 1213 specifically includes:
based on the candidate neighborhood sample number threshold and the maximum distance threshold, carrying out DBSCAN clustering on the RRC performance index data set to obtain a candidate clustering result; here, the candidate clustering result is a clustering result obtained by performing DBSCAN clustering based on the candidate neighborhood sample number threshold and the maximum distance threshold thereof.
Based on the candidate clustering result, obtaining the contour coefficient by the following formula:
wherein s (i) is a contour coefficient of the RRC performance indicator data i, a (i) is an average value of distances from i to each of the rest of the data in the cluster to which the data belongs, and b (i) is a minimum value of the distances from i to each of the rest of the clusters.
Wherein, a (i) average (i is the distance from each RRC performance indicator data in the cluster to which the RRC performance indicator data belongs);
and (i) min (the average distance from i to each RRC performance indicator data in each of the remaining clusters).
It should be noted that, in the calculation of b (i), each RRC performance indicator data in each of the remaining clusters includes not only RRC performance indicator data in other clusters, but also noise, so that the above calculation can be performed even if the DBSCAN cluster forms only one cluster.
In addition, in a wireless network RRC fault detection scenario, the scale of performance indexes related to RRC is extremely large, and selecting different key performance indexes has a great influence on the RRC fault detection result, and how to select an index that can best reflect the RRC performance as a key performance index is still a problem to be solved by those skilled in the art. In this regard, based on any of the above embodiments, in the method, the RRC performance indicator data includes an RRC access success rate and a number of failures.
The RRC access success rate is a core performance index of RRC fault detection, and is equal to RRC access success number/total number of RRC requests 100%. However, if RRC failure detection is performed based on only the RRC access success rate, there may be the following problems:
in certain event or region areas, because there are fewer network access users, when calculating the performance index, fewer network connection failures cause a significant attenuation in the performance index data. For example, when the access number is 20, 1 RRC request failure number may result in as low as 95% RRC access success rate, and if only DBSCAN is performed at the RRC access success rate, the above situation is determined to be a failure, and an alarm is triggered. In practice, however, there is a discrepancy in the reflection of the network performance status for an abnormal attenuation of the index data due to fewer user connections.
Therefore, the failure times are used as another performance index of the RRC fault detection, the failure times can be set to achieve high tolerance to low request numbers, and the lower the total request number is, the lower the threshold of the call completing rate required when the specified failure times are reached is. Therefore, in the embodiment of the present invention, the RRC performance indicator data includes an RRC access success rate and a failure number, which may be expressed as (RRC access success rate, failure number).
Based on any one of the above embodiments, an embodiment of the present invention provides a RRC fault detection method, including the following steps:
an RRC performance indicator data set including a plurality of sets of RRC performance indicator data is obtained, wherein each RRC performance indicator is represented as (RRC access success rate, failure times). And constructing a coordinate system with the RRC access success rate as an abscissa and the failure times as an ordinate, and taking each RRC performance index data in the RRC performance index data set as a point in the coordinate system.
Then, the DBSCAN parameter self-adapting method is executed. Fig. 2 is a schematic flow diagram of a DBSCAN parameter adaptive method provided in an embodiment of the present invention, and as shown in fig. 2, the method includes:
presetting the percentage as 2%, and taking 2% of the data volume as the maximum value of the neighborhood sample number threshold; assuming that the data size is 1000, the maximum value is 20;
the preset first minimum value is 3; the data dimension of the RRC performance indicator data set is 2, so the second minimum value is 2+1, and therefore the minimum value of the obtained neighborhood sample number threshold is 3;
the interval to get MinPts is [3,20] based on the maximum and minimum values.
updating the current candidate neighborhood sample number threshold, namely candidate MinPts, and acquiring a correlation curve of the candidate distance threshold and the cluster number corresponding to the candidate neighborhood sample number threshold, wherein the correlation curve comprises the cluster number under the combination of the candidate neighborhood sample number threshold and each corresponding candidate distance threshold.
considering the existing RRC fault detection alarm threshold (the access success rate is less than 95%, and the failure times are greater than 300%), the fault alarm rate is about 0.06%. When the access success rate is less than the threshold of 95%, the fault alarm rate is about 8.38%. Here, 1% to 3% is taken as the estimated alarm rate.
Under the condition of the known estimated alarm rate, selecting a candidate distance threshold value with the number of noise points being 1% -3% as the candidate distance threshold value meeting the estimated alarm rate. Here, the number of noise points may be obtained based on a correlation curve. And selecting a maximum candidate distance threshold from the candidate distance thresholds meeting the estimated alarm rate, and ensuring that the maximum candidate distance threshold is positioned at the stable position of the association curve so as to ensure the stability of the algorithm. And finally, taking the selected maximum candidate distance threshold as the maximum Eps value corresponding to the current candidate MinPts.
In step 204, the contour coefficients at the candidate MinPts and the maximum Eps values are calculated.
after the contour coefficient of each candidate neighborhood sample number threshold and the corresponding maximum distance threshold is obtained, the candidate neighborhood sample number threshold with the maximum contour coefficient and the maximum distance threshold are selected from the contour coefficients.
After the neighborhood sample number threshold MinPts and the distance threshold Eps are obtained based on the DBSCAN parameter adaptive method, the RRC performance index data sets are clustered by the DBSCAN based on the neighborhood sample number threshold MinPts and the distance threshold Eps, and a clustering result of each RRC performance index data is obtained.
And if the clustering result of any RRC performance index data is a noise point, confirming that the RRC corresponding to any RRC performance index data has a fault.
The method provided by the embodiment of the invention uses the DBSCAN algorithm for RRC fault detection, determines the feasibility of the DBSCAN algorithm in the scene, uses the RRC access success rate and the failure times as performance indexes, solves the problems that the detection result performance cannot be judged and the judgment is completely determined by subjective experience in the existing RRC fault detection fixed threshold method, ensures that the judgment of whether to alarm has clear performance index constraint, and reduces the experience risk in the classification process. In addition, in the embodiment of the invention, the self-adaptive determination of the threshold value MinPts of the number of samples and the distance threshold value Eps of the neighborhood is limited by the contour coefficient and the estimated alarm rate, so that the influence of subjective factors is avoided to the maximum extent, and the accuracy and the reliability of RRC fault detection are improved.
Based on any of the above embodiments, fig. 3 is a schematic structural diagram of an RRC fault detection apparatus according to an embodiment of the present invention, as shown in fig. 3, the apparatus includes a data set obtaining unit 310, a clustering unit 320, and a fault determining unit 330;
the data set obtaining unit 310 is configured to obtain an RRC performance indicator data set including multiple sets of RRC performance indicator data;
the clustering unit 320 is configured to perform DBSCAN clustering on the RRC performance indicator data sets to obtain a clustering result of each RRC performance indicator data;
the failure determining unit 330 is configured to determine that an RRC corresponding to any one of the RRC performance indicator data has a failure if the clustering result of the any one of the RRC performance indicator data is a noise point.
According to the device provided by the embodiment of the invention, the DBSCAN clustering is applied to the RRC fault detection in the wireless network, the DBSCAN clustering is an unsupervised learning process, manual intervention is not needed, and a threshold value is not needed to be set, so that the addition of subjective factors in the RRC fault detection process is effectively avoided, and the reliability and the accuracy of the RRC fault detection are improved.
Based on any of the above embodiments, in the apparatus, the clustering unit 320 includes a parameter obtaining subunit and a clustering subunit:
the parameter obtaining subunit is configured to obtain a distance threshold and a neighborhood sample number threshold for DBSCAN clustering based on the RRC performance indicator data set, the profile coefficient and the estimated alarm rate;
and the clustering subunit is used for performing DBSCAN clustering on the RRC performance index data sets based on the distance threshold and the neighborhood sample number threshold to obtain a clustering result of each RRC performance index data.
Based on any one of the above embodiments, in the device, the parameter obtaining subunit includes an interval obtaining module, a maximum distance threshold obtaining module, a contour coefficient obtaining module, and a selecting module;
the interval acquisition module is used for acquiring the interval of the neighborhood sample number threshold value based on the data quantity and the data dimension of the RRC performance index data set; a plurality of candidate neighborhood sample number thresholds are included within the interval;
a maximum distance threshold obtaining module, configured to obtain that an alarm rate corresponding to any one of the candidate neighborhood sample number thresholds satisfies a maximum distance threshold of the estimated alarm rate;
a contour coefficient obtaining module, configured to obtain a contour coefficient corresponding to the threshold of the number of candidate neighborhood samples and the threshold of the maximum distance;
a selecting module, configured to select the candidate neighborhood sample number threshold and the maximum distance threshold corresponding to the maximum contour coefficient as the neighborhood sample number threshold and the distance threshold.
Based on any of the above embodiments, in the apparatus, the interval obtaining module is specifically configured to:
taking the product of the data volume and a preset percentage as the maximum value of the neighborhood sample number threshold;
acquiring a first minimum value of the preset neighborhood sample number threshold;
adding 1 to the data dimension as a second minimum of the neighborhood sample number threshold;
obtaining an interval of the neighborhood sample count threshold based on the maximum value, the first minimum value, and the second minimum value of the neighborhood sample count threshold.
Based on any of the above embodiments, in the apparatus, the maximum distance threshold acquisition module is specifically configured to:
obtaining the estimated clustering quantity corresponding to the estimated alarm rate;
acquiring a correlation curve of a candidate distance threshold value and a cluster number corresponding to any one candidate neighborhood sample number threshold value;
selecting a candidate distance threshold corresponding to each estimated clustering quantity from the association curve;
and selecting the maximum candidate distance threshold as the maximum distance threshold corresponding to any one of the candidate neighborhood sample number thresholds.
Based on any of the above embodiments, in the apparatus, the contour coefficient obtaining module is specifically configured to:
based on any candidate neighborhood sample number threshold and the maximum distance threshold, carrying out DBSCAN clustering on the RRC performance index data set to obtain a candidate clustering result;
obtaining the contour coefficient based on the candidate clustering result by the following formula:
wherein s (i) is a contour coefficient of the RRC performance indicator data i, a (i) is an average value of distances from i to each of the rest of the data in the cluster to which the data belongs, and b (i) is a minimum value of the distances from i to each of the rest of the clusters.
Based on any of the above embodiments, in the apparatus, the RRC performance indicator data includes an RRC access success rate and a failure number.
Fig. 4 is a schematic entity structure diagram of an electronic device according to an embodiment of the present invention, and as shown in fig. 4, the electronic device may include: a processor (processor)401, a communication Interface (communication Interface)402, a memory (memory)403 and a communication bus 404, wherein the processor 401, the communication Interface 402 and the memory 403 complete communication with each other through the communication bus 404. The processor 401 may call a computer program stored in the memory 403 and operable on the processor 401 to perform the RRC failure detection method provided by the above embodiments, for example, including: acquiring an RRC performance index data set containing a plurality of groups of RRC performance index data; performing DBSCAN clustering on the RRC performance index data set to obtain a clustering result of each RRC performance index data; and if the clustering result of any RRC performance index data is a noise point, confirming that the RRC corresponding to any RRC performance index data has a fault.
In addition, the logic instructions in the memory 403 may be implemented in the form of software functional units and stored in a computer readable storage medium when the software functional units are sold or used as independent products. Based on such understanding, the technical solutions of the embodiments of the present invention may be essentially implemented or make a contribution to the prior art, or may be implemented in the form of a software product stored in a storage medium and including 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 methods described in the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
An embodiment of the present invention further provides a non-transitory computer-readable storage medium, on which a computer program is stored, where the computer program is implemented to perform the RRC fault detection method provided in the foregoing embodiments when executed by a processor, for example, the method includes: acquiring an RRC performance index data set containing a plurality of groups of RRC performance index data; performing DBSCAN clustering on the RRC performance index data set to obtain a clustering result of each RRC performance index data; and if the clustering result of any RRC performance index data is a noise point, confirming that the RRC corresponding to any RRC performance index data has a fault.
The above-described embodiments of the apparatus are merely illustrative, and 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 modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. With this understanding in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the embodiments or some parts of the embodiments.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (8)
1. A method for RRC fault detection, comprising:
acquiring an RRC performance index data set containing a plurality of groups of RRC performance index data;
performing DBSCAN clustering on the RRC performance index data set to obtain a clustering result of each RRC performance index data;
if the clustering result of any RRC performance index data is a noise point, confirming that the RRC corresponding to any RRC performance index data has a fault;
the performing DBSCAN clustering on the RRC performance indicator data set to obtain a clustering result of each RRC performance indicator data specifically includes:
acquiring a distance threshold and a neighborhood sample number threshold for DBSCAN clustering based on the RRC performance index data set, the profile coefficient and the estimated alarm rate;
based on the distance threshold and the neighborhood sample number threshold, carrying out DBSCAN clustering on the RRC performance index data set to obtain a clustering result of each RRC performance index data;
the acquiring a distance threshold and a neighborhood sample number threshold for DBSCAN clustering based on the RRC performance indicator dataset, the profile coefficient and the estimated alarm rate specifically includes:
acquiring an interval of the neighborhood sample number threshold based on the data quantity and the data dimension of the RRC performance index data set; a plurality of candidate neighborhood sample number thresholds are included within the interval;
acquiring the alarm rate corresponding to any candidate neighborhood sample number threshold value and meeting the maximum distance threshold value of the estimated alarm rate;
acquiring a contour coefficient corresponding to the threshold of the number of any candidate neighborhood samples and the maximum distance threshold;
and selecting the candidate neighborhood sample number threshold and the maximum distance threshold corresponding to the maximum contour coefficient as the neighborhood sample number threshold and the distance threshold.
2. The RRC fault detection method according to claim 1, wherein the obtaining the interval of the neighborhood sample number threshold based on the data size and the data dimension of the RRC performance indicator data set specifically includes:
taking the product of the data volume and a preset percentage as the maximum value of the neighborhood sample number threshold;
acquiring a first minimum value of the preset neighborhood sample number threshold;
adding 1 to the data dimension as a second minimum of the neighborhood sample number threshold;
obtaining an interval of the neighborhood sample count threshold based on the maximum value, the first minimum value, and the second minimum value of the neighborhood sample count threshold.
3. The RRC fault detection method according to claim 1, wherein the obtaining of the alarm rate corresponding to any one of the candidate neighborhood sample number thresholds satisfies a maximum distance threshold of the estimated alarm rate, specifically includes:
obtaining the estimated clustering quantity corresponding to the estimated alarm rate;
acquiring a correlation curve of a candidate distance threshold value and a cluster number corresponding to any one candidate neighborhood sample number threshold value;
selecting a candidate distance threshold corresponding to each estimated clustering quantity from the association curve;
and selecting the maximum candidate distance threshold as the maximum distance threshold corresponding to any one of the candidate neighborhood sample number thresholds.
4. The RRC fault detection method according to claim 1, wherein the obtaining of the profile coefficient corresponding to any one of the candidate neighborhood sample number threshold and the maximum distance threshold specifically includes:
based on any candidate neighborhood sample number threshold and the maximum distance threshold, carrying out DBSCAN clustering on the RRC performance index data set to obtain a candidate clustering result;
obtaining the contour coefficient based on the candidate clustering result by the following formula:
wherein s (i) is a contour coefficient of the RRC performance indicator data i, a (i) is an average value of distances from i to each of the rest of the data in the cluster to which the data belongs, and b (i) is a minimum value of the distances from i to each of the rest of the clusters.
5. The RRC fault detection method of any one of claims 1 to 4, wherein the RRC performance indicator data comprises an RRC access success rate and a number of failures.
6. An apparatus for RRC failure detection, comprising:
a data set obtaining unit, configured to obtain an RRC performance indicator data set including multiple sets of RRC performance indicator data;
the clustering unit is used for carrying out DBSCAN clustering on the RRC performance index data set to obtain a clustering result of each RRC performance index data;
a fault determining unit, configured to determine that a fault exists in an RRC corresponding to any one of the RRC performance indicator data if the clustering result of any one of the RRC performance indicator data is a noise point;
the clustering unit comprises a parameter obtaining subunit and a clustering subunit:
the parameter obtaining subunit is configured to obtain a distance threshold and a neighborhood sample number threshold for DBSCAN clustering based on the RRC performance indicator data set, the profile coefficient and the estimated alarm rate;
the clustering subunit is configured to perform DBSCAN clustering on the RRC performance indicator data sets based on the distance threshold and the neighborhood sample number threshold, and obtain a clustering result of each RRC performance indicator data;
the parameter acquisition subunit comprises an interval acquisition module, a maximum distance threshold acquisition module, a contour coefficient acquisition module and a selection module;
the interval acquisition module is used for acquiring the interval of the neighborhood sample number threshold value based on the data quantity and the data dimension of the RRC performance index data set; a plurality of candidate neighborhood sample number thresholds are included within the interval;
a maximum distance threshold obtaining module, configured to obtain that an alarm rate corresponding to any candidate neighborhood sample number threshold satisfies a maximum distance threshold of the estimated alarm rate;
a contour coefficient obtaining module, configured to obtain a contour coefficient corresponding to the threshold of the number of candidate neighborhood samples and the threshold of the maximum distance;
a selecting module, configured to select the candidate neighborhood sample number threshold and the maximum distance threshold corresponding to the maximum contour coefficient as the neighborhood sample number threshold and the distance threshold.
7. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor when executing the program performs the steps of the RRC fault detection method of any of claims 1 to 5.
8. A non-transitory computer readable storage medium having stored thereon a computer program, which when executed by a processor, performs the steps of the RRC fault detection method of any of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910560786.5A CN112153685B (en) | 2019-06-26 | 2019-06-26 | RRC fault detection method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910560786.5A CN112153685B (en) | 2019-06-26 | 2019-06-26 | RRC fault detection method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112153685A CN112153685A (en) | 2020-12-29 |
| CN112153685B true CN112153685B (en) | 2022-02-25 |
Family
ID=73869806
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910560786.5A Active CN112153685B (en) | 2019-06-26 | 2019-06-26 | RRC fault detection method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112153685B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101714952A (en) * | 2009-12-22 | 2010-05-26 | 北京邮电大学 | Method and device for identifying traffic of access network |
| CN106254153A (en) * | 2016-09-19 | 2016-12-21 | 腾讯科技（深圳）有限公司 | A kind of Network Abnormal monitoring method and apparatus |
| CN107333294A (en) * | 2017-07-31 | 2017-11-07 | 南昌航空大学 | A kind of combination AdaBoost and SVMs link quality prediction method |
| CN108737147A (en) * | 2017-04-25 | 2018-11-02 | 中国移动通信集团广东有限公司 | A kind of network alarm event-handling method and device |
| CN108923952A (en) * | 2018-05-31 | 2018-11-30 | 北京百度网讯科技有限公司 | Method for diagnosing faults, equipment and storage medium based on service monitoring index |
| CN109257762A (en) * | 2018-09-12 | 2019-01-22 | 南方电网科学研究院有限责任公司 | Match electric terminal illegal wireless communication link detection method based on wireless signal strength Density Clustering Analysis |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWI331868B (en) * | 2007-06-11 | 2010-10-11 | Univ Nat Pingtung Sci & Tech | Detecting method of network invasion |
| US9213590B2 (en) * | 2012-06-27 | 2015-12-15 | Brocade Communications Systems, Inc. | Network monitoring and diagnostics |
| US10063575B2 (en) * | 2015-10-08 | 2018-08-28 | Cisco Technology, Inc. | Anomaly detection in a network coupling state information with machine learning outputs |
| US9979619B2 (en) * | 2015-11-24 | 2018-05-22 | Cisco Technology, Inc. | Identifying a source of packet drops in a network |
-
2019
- 2019-06-26 CN CN201910560786.5A patent/CN112153685B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101714952A (en) * | 2009-12-22 | 2010-05-26 | 北京邮电大学 | Method and device for identifying traffic of access network |
| CN106254153A (en) * | 2016-09-19 | 2016-12-21 | 腾讯科技（深圳）有限公司 | A kind of Network Abnormal monitoring method and apparatus |
| CN108737147A (en) * | 2017-04-25 | 2018-11-02 | 中国移动通信集团广东有限公司 | A kind of network alarm event-handling method and device |
| CN107333294A (en) * | 2017-07-31 | 2017-11-07 | 南昌航空大学 | A kind of combination AdaBoost and SVMs link quality prediction method |
| CN108923952A (en) * | 2018-05-31 | 2018-11-30 | 北京百度网讯科技有限公司 | Method for diagnosing faults, equipment and storage medium based on service monitoring index |
| CN109257762A (en) * | 2018-09-12 | 2019-01-22 | 南方电网科学研究院有限责任公司 | Match electric terminal illegal wireless communication link detection method based on wireless signal strength Density Clustering Analysis |
Non-Patent Citations (5)
| Title |
|---|
| 《DBSCAN算法在无线网络优化中的应用》;鲁飞，王任，翁维波，陈明，涂志龙;《移动通信》;20181215;正文第1-2小节 * |
| 《自适应确定DBSCAN算法参数的算法研究》;李文杰，闫世强，蒋莹，张松芝，王成良;《计算机工程与应用》;20190301;全文 * |
| DBSCAN算法研究及并行化实现;宋董飞等;《计算机工程与应用》;20181215(第24期);全文 * |
| Siti Monalisa.《Analysis Outlier Data on RFM and LRFM Models to Determining Customer Loyalty with DBSCAN Algorithm》.《2018 International Symposium on Advanced Intelligent Informatics (SAIN)》.2019, * |
| 一种基于改进DBSCAN算法的光伏故障检测方法;叶进等;《广西大学学报(自然科学版)》;20190425(第02期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112153685A (en) | 2020-12-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109586952B (en) | Server capacity expansion method and device | |
| CN110879981B (en) | Face key point quality assessment method, device, computer equipment and storage medium | |
| CN108989075A (en) | A kind of network failure locating method and system | |
| TWI684139B (en) | System and method of learning-based prediction for anomalies within a base station | |
| CN111294819A (en) | Network optimization method and device | |
| CN111122222B (en) | Sample point position determining method and system | |
| CN112308345A (en) | Communication network load prediction method, device and server | |
| CN115081969A (en) | Abnormal data determination method and related device | |
| CN110162379A (en) | Virtual machine migration method, device and computer equipment | |
| US10805186B2 (en) | Mobile communication network failure monitoring system and method | |
| CN113486134B (en) | Rainfall anomaly detection method, rainfall anomaly detection device, computer equipment and storage medium | |
| CN113112126B (en) | Artificial intelligence-based water supply pipe leakage early warning method and system | |
| CN111259007B (en) | Electric vehicle information monitoring method and device, server and electric vehicle management system | |
| CN115705282A (en) | Cell network anomaly detection method and device and computer readable storage medium | |
| CN112153685B (en) | RRC fault detection method and device | |
| CN112836843B (en) | Base station out-of-service alarm prediction method and device | |
| CN111258968B (en) | Enterprise redundant data cleaning method and device and big data platform | |
| CN117009115A (en) | Fault positioning method and device, electronic equipment and computer storage medium | |
| CN114036711A (en) | Network quality degradation detection method and system | |
| CN111954232B (en) | Wireless network index classification method and device | |
| CN112580908A (en) | Wireless performance index evaluation method and device | |
| CN112749751B (en) | Detector fusion method and system based on probability perception | |
| CN112905419B (en) | Index data monitoring threshold range determining method and device and readable storage medium | |
| CN115437812A (en) | Application fault risk influence range analysis method and system | |
| CN117632366A (en) | Method, device, equipment and medium for determining resource occupation state of simulation accelerator |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |