Method of identifying problem cell, electronic device, computer readable medium

Abstract

The embodiment of the present disclosure provides a method for identifying a problem cell, which comprises: determining abnormal contribution degrees of each cell in a subnet; determining at least one cell as a problem cell according to the abnormal contribution degrees of each cell in the subnet; wherein the abnormal contribution degree of each cell is the degree of association between the performance index of the cell and an abnormality of the performance index of the subnet when the abnormality occurs; the abnormality of the performance index of the subnet is that the performance index of the subnet is out of a first threshold range; the performance index of the subnet is determined according to parameter statistics of each cell in the subnet; and the performance index of the cell is determined according to parameter statistics of the cell. The embodiment of the present disclosure also provides an electronic device and a computer readable medium.

Description

Technical Field

[0001] This disclosure relates to the field of mobile communication network technology, and in particular to methods, electronic devices, and computer-readable media for identifying problematic cells. Background Technology

[0002] In mobile communication networks (such as mobile broadband networks, MBB), a subnet is considered to have experienced an anomaly when its key performance indicators (KPIs) such as call completion rate, call drop rate, congestion rate, handover success rate, traffic volume, and data rate exceed predetermined ranges. Subnet anomalies are typically caused by problems in one or more cells; these cells are called problem cells, or top N worst cells, and are usually those with the worst performance indicators.

[0003] Therefore, accurately identifying the problematic cell causing the anomaly is crucial for network optimization and improving user experience. However, current technologies cannot accurately pinpoint the problem cell. Summary of the Invention

[0004] This disclosure provides a method, electronic device, and computer-readable medium for identifying problematic cells.

[0005] In a first aspect, embodiments of this disclosure provide a method for identifying problematic cells, comprising:

[0006] Determine the abnormal contribution level of each cell in the subnet;

[0007] Based on the abnormal contribution of each cell in the subnet, at least one cell is identified as a problem cell;

[0008] in,

[0009] The abnormal contribution of each cell is the degree of correlation between the cell's performance index and the abnormality when the subnet's performance index is abnormal; the abnormality of the subnet's performance index is when the subnet's performance index exceeds a first threshold range; the subnet's performance index is determined based on the parameter statistics of each cell; the cell's performance index is determined based on the parameter statistics therein.

[0010] In some embodiments, the first threshold range includes:

[0011] The lower limit of the first threshold range, and / or the upper limit of the first threshold range.

[0012] In some embodiments, the first threshold range is determined by dynamic threshold detection technology.

[0013] In some embodiments, the performance metric is a ratio-type performance metric, which is the ratio of a first parameter statistic to a second parameter statistic.

[0014] The abnormal contribution of each cell represents the degree to which the performance index of the subnet shifts within a first threshold range relative to the performance index of the subnet before removal of the cell.

[0015] In some embodiments, determining at least one cell as a problem cell based on the abnormal contribution of each cell in the subnet includes:

[0016] The cells with positive abnormal contribution are sorted in descending order to obtain the sequence to be screened, and N is set to 1;

[0017] Determine the optimized performance index; the optimized performance index is the difference between the subnet's performance index and the subnet's temporary performance index, and the temporary performance index is the overall performance index of the remaining cells in the subnet after removing the top N cells in the to-be-screened sequence.

[0018] If the optimized performance index is within the first threshold range, the top N cells in the screening sequence are determined as candidate cells; if the optimized performance index exceeds the first threshold range, N is increased by 1 and the process returns to the step of determining the optimized performance index.

[0019] At least some of the candidate cells were identified as problematic cells.

[0020] In some embodiments, determining at least a subset of candidate cells as problem cells includes:

[0021] Determine multiple historical performance metrics for each candidate cell; each historical performance metric is the performance metric of that cell at a historical moment before the subnet's performance metric became abnormal;

[0022] Determine the historical anomaly ratio of each candidate cell; the historical anomaly ratio of each cell is the proportion of historical performance indicators that exceed the first threshold range among all historical performance indicators of that cell;

[0023] Cells with a historical anomaly ratio higher than the second threshold are removed from the candidate cells.

[0024] In some embodiments, after removing cells with a historical anomaly ratio higher than a second threshold from the candidate cells, the method further includes:

[0025] If there are remaining candidate cells, all remaining candidate cells are identified as problem cells;

[0026] If there are no remaining candidate cells, the cell with the largest abnormal contribution is identified as the first pre-positioned cell and is considered the problem cell.

[0027] In some embodiments, the performance index is a numerical performance index, which is a parameter statistic;

[0028] The abnormal contribution of each cell is determined based on the ratio of the deviation of the cell's performance index to the deviation of the subnet's performance index;

[0029] Wherein, the performance index deviation value of each cell is the difference between the cell's performance index and the predicted performance index, the performance index deviation value of the subnet is the difference between the subnet's performance index and the predicted performance index, the predicted performance index of each cell is the normal value of the predicted performance index of that cell, and the predicted performance index of the subnet is the sum of the predicted performance indices of all cells therein.

[0030] In some embodiments, the predicted performance metric for each cell is the average of the cell's performance metric over a predetermined period before the subnet's performance metric becomes abnormal.

[0031] In some embodiments, determining at least one cell as a problem cell based on the abnormal contribution of each cell in the subnet includes:

[0032] A third threshold is determined based on the abnormal contribution values ​​of all cells with positive abnormal contribution values, wherein the third threshold is a positive number;

[0033] If there are cells whose abnormal contribution exceeds the third threshold, all cells whose abnormal contribution exceeds the third threshold are identified as problem cells.

[0034] If there are no cells whose abnormal contribution exceeds the third threshold, the second pre-positioned cell with the largest abnormal contribution is identified as the problem cell.

[0035] In some embodiments, the third threshold is calculated by the following formula: Third threshold = average of the abnormal contribution of all cells with positive abnormal contribution + k * standard deviation of the abnormal contribution of all cells with positive abnormal contribution;

[0036] Where k is a value greater than 0.

[0037] In a second aspect, embodiments of this disclosure provide an electronic device, which includes:

[0038] One or more processors;

[0039] A memory having stored one or more programs, which, when executed by one or more processors, enable the one or more processors to implement any of the methods described above for identifying problem cells;

[0040] One or more I / O interfaces are connected between the processor and the memory and configured to enable information interaction between the processor and the memory.

[0041] Thirdly, embodiments of this disclosure provide a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements any of the methods described above for identifying problematic cells.

[0042] In this embodiment, the anomaly contribution of each cell in the subnet is determined based on the performance metrics (or parameter statistics of all cells) of the subnet and cells when the anomaly occurs. This means determining the likelihood that each cell in the subnet is causing the anomaly. Therefore, compared to techniques that determine the anomaly contribution solely based on the performance metrics of each cell in the subnet, this embodiment can more accurately identify the problematic cell (i.e., the cell actually causing the anomaly) based on the anomaly contribution, enabling targeted network optimization and improved user experience.

[0043] Furthermore, the embodiments disclosed herein can be implemented entirely automatically without human intervention, are unaffected by human factors, and are fast, intelligent, and accurate. Attached Figure Description

[0044] The accompanying drawings are provided to further illustrate the embodiments of this disclosure and form part of the specification. They are used together with the embodiments of this disclosure to explain the disclosure and do not constitute a limitation thereof. The above and other features and advantages will become more apparent to those skilled in the art from the detailed description of exemplary embodiments with reference to the accompanying drawings, in which:

[0045] Figure 1 This is a schematic block diagram of the composition of a mobile communication network to which embodiments of this disclosure apply;

[0046] Figure 2 This is a schematic diagram showing the real-time value of the uplink rate performance index and the range of the first threshold in an embodiment of this disclosure;

[0047] Figure 3 A flowchart illustrating a method for identifying problematic cells provided in this disclosure embodiment;

[0048] Figure 4 A flowchart of some steps in another method for identifying problem cells provided in this disclosure embodiment;

[0049] Figure 5 A flowchart of some steps in another method for identifying problem cells provided in this disclosure embodiment;

[0050] Figure 6 A flowchart of some steps in another method for identifying problem cells provided in this disclosure embodiment;

[0051] Figure 7 A logical process diagram of another method for identifying problematic cells provided in this disclosure embodiment;

[0052] Figure 8 A block diagram of an electronic device provided in an embodiment of this disclosure;

[0053] Figure 9 This is a block diagram illustrating the composition of a computer-readable medium provided in an embodiment of the present disclosure. Detailed Implementation

[0054] To enable those skilled in the art to better understand the technical solutions of the embodiments of this disclosure, the method, electronic device, and computer-readable medium for identifying problematic cells provided in the embodiments of this disclosure will be described in detail below with reference to the accompanying drawings.

[0055] Embodiments of this disclosure will be described more fully below with reference to the accompanying drawings; however, the embodiments shown may be embodied in different forms and should not be construed as limited to the embodiments set forth in this disclosure. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will enable those skilled in the art to fully understand the scope of this disclosure.

[0056] Embodiments of this disclosure can be described with reference to plan views and / or cross-sectional views, taking into account the ideal schematic diagrams of this disclosure. Therefore, the example illustrations may be modified according to manufacturing techniques and / or tolerances.

[0057] Where there is no conflict, the various embodiments of this disclosure and the features thereof in the embodiments may be combined with each other.

[0058] The terminology used in this disclosure is for the purpose of describing particular embodiments only and is not intended to limit the disclosure. The term "and / or" as used in this disclosure includes any and all combinations of one or more of the associated enumerated entries. The singular forms "a" and "the" as used in this disclosure are also intended to include the plural forms, unless the context clearly indicates otherwise. The terms "comprising," "made of," etc., as used in this disclosure specify the presence of the stated feature, integral, step, operation, element, and / or component, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or groups thereof.

[0059] Unless otherwise specified, all terms used in this disclosure (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art. It will also be understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant art and this disclosure, and will not be interpreted as having an idealized or overly formal meaning, unless expressly so defined in this disclosure.

[0060] This disclosure is not limited to the embodiments shown in the accompanying drawings, but includes modifications to the configuration based on the manufacturing process. Therefore, the areas illustrated in the drawings are schematic, and the shapes of the areas shown illustrate specific shapes of the areas of an element, but are not intended to be limiting.

[0061] Glossary

[0062] In this disclosure, unless otherwise specified, the following technical terms shall be interpreted as follows:

[0063] A mobile communication network is a network that enables communication between mobile users and fixed users, or between mobile users themselves. Specific examples of mobile communication networks include Global System for Mobile Communications (GSM), Universal Mobile Telecommunications System (UMTS), Long Term Evolution (LTE), Code Division Multiple Access (CDMA), and New Radio (NR).

[0064] A cell is the smallest independently controllable area in a mobile communication network. For example, a cell can be the area covered by a base station, or the area covered by a single antenna within a base station.

[0065] A subnet is an area that integrates multiple cells for unified management, and it may include multiple cells corresponding to a single network element device.

[0066] Performance metrics are parameters derived from statistical parameters in a mobile communication network that represent the network's performance. Examples of performance metrics include call completion rate, call drop rate, congestion rate, handover success rate, traffic volume, and data rate.

[0067] Parametric statistics refer to the direct statistical values ​​of parameters generated during the operation of a mobile communication network, but do not include further values ​​calculated from the direct statistical values ​​(such as ratio-type performance indicators). For example, parametric statistics may include traffic, rate, number of connected calls, number of connection attempts, number of dropped calls, number of calls, number of congestion events, number of data transmissions, number of successful handovers, and number of handover attempts.

[0068] Application Environment

[0069] The embodiments disclosed herein are used in a mobile communication network environment.

[0070] For example, refer to Figure 1 The mobile communication networks applicable to the embodiments of this disclosure may include: wireless network equipment, core network equipment, network equipment management server, network performance monitoring server, etc.

[0071] The specific calculations in this embodiment can be implemented through a network performance monitoring server. The network performance monitoring server can periodically obtain performance indicator data related to subnets and cells from the network device management server for use in implementing this embodiment.

[0072] Specific embodiments of this disclosure

[0073] In mobile communication networks (such as mobile broadband networks, MBB), the main management area is the subnet, and each subnet is further divided into multiple cells. Therefore, the performance indicators (KPIs) of each subnet, such as call completion rate, call drop rate, congestion rate, handover success rate, traffic, and speed, are actually a combination of the performance indicators of all the cells within it.

[0074] When a performance metric of a subnet exceeds a predetermined range, it indicates that the performance metric is clearly unreasonable, meaning the subnet is experiencing an anomaly, which will severely impact user experience. Subnet anomalies are usually caused by problems with one or more cells; these cells are called problem cells, or top N worst cells, and are typically cells with poor performance across various metrics.

[0075] Therefore, accurately identifying the problematic cells causing the anomalies is crucial for optimizing the network and improving user experience.

[0076] In some related technologies, problematic cells are identified directly by ranking performance indicators. For example, cells with high call drop rates can be directly identified as problematic cells. However, due to the significant differences in traffic volume (such as call volume) between different cells, the performance indicators of cells with low traffic volume fluctuate greatly. Therefore, performance indicators alone often cannot reflect the true performance of a cell, and analyzing a cell's performance indicators alone may not be sufficient to identify the actual cell causing the subnet anomaly.

[0077] Another related technology uses performance indicators and manually set absolute numbers as criteria for screening problem cells. For example, taking call drop rate as an example, staff (such as on-site network optimization engineers or customers) manually set thresholds based on the cell's geographical environment, frequently complained-about issues, etc., stipulating that only cells with a call drop rate greater than the corresponding threshold are considered problem cells. However, this method involves subjective judgment, has a large human factor, cannot guarantee accuracy, is time-consuming, and lacks universality.

[0078] Firstly, referring to Figure 3 This disclosure provides a method for identifying problematic cells, comprising:

[0079] S101. Determine the abnormal contribution of each cell in the subnet.

[0080] The abnormal contribution of each cell is the degree of correlation between the cell's performance index and the abnormality when the subnet's performance index is abnormal; an abnormality in the subnet's performance index is defined as the subnet's performance index exceeding a first threshold range; the subnet's performance index is determined based on the parameter statistics of each cell; and the cell's performance index is determined based on the parameter statistics of its cells.

[0081] When a certain performance indicator of a subnet becomes abnormal (i.e., the overall performance indicator of the subnet exceeds the first threshold range), the abnormal contribution of each cell can be determined based on the current performance indicator of the subnet and the performance indicators of each cell within it. That is, the degree of correlation between the performance indicator of each cell and the abnormality that occurred in the subnet, or the probability that each cell in the subnet caused the abnormality.

[0082] Of course, it should be understood that the first threshold range is different for different performance indicators; that is, each performance indicator should have its own corresponding first threshold range.

[0083] S102. Based on the abnormal contribution of each cell in the subnet, identify at least one cell as a problem cell.

[0084] Based on the probability (anomaly contribution) of each cell in the subnet causing the anomaly, the problematic cell can be identified, that is, the cell that actually caused the anomaly.

[0085] Of course, it should be understood that the method of this disclosure embodiment is for calculations in the event of a subnet anomaly, and therefore can be calculated in real time when a subnet anomaly occurs, but this does not mean that the method of this disclosure embodiment must be performed in real time. For example, the relevant data when the subnet is anomaly occurs can also be saved and processed later in accordance with the method of this disclosure embodiment.

[0086] In this embodiment, the anomaly contribution of each cell in the subnet is determined based on the performance metrics (or parameter statistics of all cells) of the subnet and cells when the anomaly occurs. This means determining the likelihood that each cell in the subnet is causing the anomaly. Therefore, compared to techniques that determine the anomaly contribution solely based on the performance metrics of each cell in the subnet, this embodiment can more accurately identify the problematic cell (i.e., the cell actually causing the anomaly) based on the anomaly contribution, enabling targeted network optimization and improved user experience.

[0087] Furthermore, the embodiments disclosed herein can be implemented entirely automatically without human intervention, are unaffected by human factors, and are fast, intelligent, and accurate.

[0088] In some embodiments, the first threshold range includes: a lower limit of the first threshold range, and / or, an upper limit of the first threshold range.

[0089] The first threshold range used to determine whether a subnet is abnormal can include an upper and lower limit value; that is, performance indicators that are too high or too low may indicate abnormality. For example, excessively high or low speeds may both represent abnormalities.

[0090] Alternatively, the first threshold range may include only one of the upper and lower limits, meaning that the performance metric can only become abnormal due to either being too high or too low. For example, a dropout rate greater than a certain value is clearly abnormal, but even if the dropout rate is the minimum value of 0, it should not be considered abnormal.

[0091] In some embodiments, the first threshold range is determined by dynamic threshold detection technology.

[0092] In other words, the first threshold range mentioned above can be a real-time value calculated based on dynamic threshold detection technology (such as the automatically learning Holtwinters algorithm).

[0093] For example, refer to Figure 2 The dark line represents the real-time value of the uplink speed performance index, while the light-colored areas above and below it represent the value range of the first threshold range at various times. It can be seen that the first threshold range changes in real time according to the time.

[0094] The process of determining the normal range (first threshold range) of performance indicators using dynamic threshold detection technology can be implemented using some related technologies, and will not be described in detail here.

[0095] As one embodiment of this disclosure, the following describes a specific method for identifying problematic cells when ratio-type performance indicators are abnormal.

[0096] Among them, ratio-type performance indicators refer to the ratio formed by two different parameter statistics, such as percentages, that is, the ratio of the first parameter statistic to the second parameter statistic.

[0097] For example, ratio-based performance metrics may include connection rate, call drop rate, congestion rate, and handover success rate; correspondingly, the first parameter statistics corresponding to connection rate, call drop rate, congestion rate, and handover success rate are the number of connection attempts, the number of call drops, the number of congestion attempts, and the number of successful handover attempts, respectively; while the second parameter statistics are the number of connection attempts, the number of calls, the number of data transmission attempts, and the number of handover attempts, respectively.

[0098] Therefore, the performance index (ratio-type performance index) of the i-th cell can be calculated using the following formula:

[0099] The performance index of cell i = the first parameter statistic of cell i / the second parameter statistic of cell i.

[0100] Correspondingly, the performance metric (ratio-based performance metric) of each subnet is equal to the ratio of the first parameter statistic to the second parameter statistic of all its cells, i.e.,

[0101]

[0102] Where n is the total number of cells in the subnet.

[0103] It is evident that, compared to ratio-based performance metrics, the performance metrics of a subnet are not simply the sum of the performance metrics of its individual cells.

[0104] In some embodiments, for ratio-based performance metrics, the abnormal contribution of each cell represents the degree to which the subnet's performance metric shifts relative to the subnet's performance metric before removal of the cell within a first threshold range after the cell is removed from the subnet.

[0105] Compared to ratio-based performance metrics, the anomaly contribution of each cell is represented as follows:

[0106] After removing a cell from the subnet, the remaining cells in the subnet can obtain a new subnet performance index. The degree to which the performance index of the new subnet shifts relative to the original performance index of the subnet before removal is within the first threshold range is the abnormal contribution of the cell.

[0107] The term "degree of shift towards the first threshold range" refers to the following: if the new performance index after removal is closer to the first threshold range than the original performance index before removal, the degree of shift is positive (i.e., shift towards the first threshold range); conversely, if the new performance index after removal is farther away from the first threshold range than the original performance index before removal, the degree of shift is negative (i.e., shift away from the first threshold range).

[0108] Since the current state is abnormal, the subnet's performance metrics must exceed the first threshold range. Therefore, when the new performance metrics after removal are closer to the first threshold range than the original performance metrics before removal, it indicates that the shift of the new performance metrics relative to the original performance metrics towards the first threshold range is a "positive value." That is, after removing a cell, the performance metrics become closer to or even enter the first threshold range (i.e., biased towards normal), so the role of the "removed" cell should be to move the performance metrics away from the first threshold range. Therefore, the abnormal contribution of this cell is a "positive contribution," or in other words, the role of this cell is to cause anomalies in the subnet.

[0109] Conversely, when the new performance metric after removal is further away from the first threshold range than the original performance metric before removal, it indicates that the degree of shift of the new performance metric relative to the original performance metric towards the first threshold range is "negative." That is, after removing a cell, the performance metric becomes further away from the first threshold range (i.e., biased towards anomalies), so the role of the "removed" cell should be to bring the performance metric closer to the first threshold range. Therefore, the anomaly contribution of this cell is "negative contribution," or in other words, the role of this cell is to prevent anomalies in the subnet.

[0110] Of course, the new performance indicators after the above removal may also be equal to the original performance indicators. In this case, the anomaly contribution of the cell is 0, that is, the cell has no impact on the anomalies generated by the subnet.

[0111] Specifically, the abnormal contribution of the j-th cell can be calculated using the following formula:

[0112]

[0113] Where n is the total number of cells in the subnet; D is the abnormal direction, defined as:

[0114] When the performance index of the subnet is greater than the upper limit of the first threshold range, D = 1;

[0115] When the performance index of the subnet is less than the lower limit of the first threshold range, D = -1.

[0116] Of course, it should be understood that when the first threshold range has only one of the upper and lower limits, the performance index can only exceed the first threshold range in one direction, so the abnormal direction can be a fixed value.

[0117] In addition, the absolute value of the abnormal direction does not necessarily have to be 1, as long as it is one of two values ​​with the same absolute value, either positive or negative.

[0118] As can be seen from the above algorithm, when a ratio-based performance indicator becomes abnormal, the cell's abnormal contribution is not calculated solely based on the cell's performance indicator and parameter statistics (such as the number of occurrences), but rather based on the cell's own first parameter statistics and second parameter statistics, as well as the first parameter statistics and second parameter statistics of other cells in the subnet. Therefore, this abnormal contribution can more realistically reflect the degree of correlation between the cell and the subnet's anomalies, and the problem cells identified based on this abnormal contribution are more accurate, which can achieve better network optimization and improve user experience.

[0119] Reference Figure 4 In some embodiments, determining at least one cell as a problem cell based on the abnormal contribution of each cell in the subnet (S102) includes:

[0120] S1021. Sort the cells with positive abnormal contribution in descending order to obtain the sequence to be screened, and determine N to be 1.

[0121] S1022. Determine the optimized performance index; the optimized performance index is the difference between the subnet's performance index and the subnet's temporary performance index. The temporary performance index is the overall performance index of the remaining cells in the subnet after removing the top N cells in the to-be-screened sequence.

[0122] S1023. If the optimized performance index is within the first threshold range, determine the top N cells in the screening sequence as candidate cells; if the optimized performance index exceeds the first threshold range, increase N by 1 and return to the step of determining the optimized performance index.

[0123] S1024. Identify at least some candidate cells as problematic cells.

[0124] After determining the anomaly contribution of each cell in the subnet, it is necessary to further identify the problem cells. However, the above anomaly contribution is a relative value. Although it can indicate the degree of correlation between each cell in the subnet and the anomaly, it cannot directly determine how many cells should be selected as problem cells based on the size of the anomaly contribution.

[0125] Therefore, we can first select cells with positive abnormal contribution rates to form a screening sequence. This is because only cells with positive abnormal contribution rates may cause anomalies, while the role of other cells is to eliminate anomalies or is unrelated to anomalies. Thus, only cells with positive abnormal contribution rates are likely to be problem cells.

[0126] Then, cells are excluded in descending order of abnormal contribution, with the number of cells excluded gradually increasing. Specifically, the cell with the largest abnormal contribution is excluded the first time (N=1), the cells with the top two abnormal contributions are excluded the second time (N=2), the cells with the top three abnormal contributions are excluded the third time (N=3), and so on.

[0127] After each cell is excluded, the temporary performance index of the subnet is calculated, which is the performance index of the "small subnet" composed of the remaining cells in the subnet after removing the excluded cells.

[0128] The difference between the subnet's performance index and its temporary performance index is then calculated as the optimized performance index, which is the difference between the performance index of the "atomic network" (the overall performance index of all cells) and the performance index of the "small subnet" after excluding the current number of cells (i.e., the overall performance index of the remaining cells after excluding N cells with the largest abnormal contributions).

[0129] For example, when N cells are excluded, the performance optimization metric can be calculated using the following formula:

[0130]

[0131] Where n is the total number of cells in the subnet, and N is the number of cells currently excluded.

[0132] After any elimination, the following situations should be handled:

[0133] (1) If the optimized performance index is not within the range of the first threshold above, then exclude one more cell (N = N + 1) and return to the step of determining the optimized performance index (S1022) to recalculate the optimized performance index;

[0134] (2) If the optimized performance index is within the range of the first threshold above, the exclusion process ends, and the N cells currently excluded (i.e. the N cells with the largest abnormal contribution) are used as "candidate cells", and at least some of the candidate cells are further identified as problem cells (S1024).

[0135] It should be understood that since anomalies can only be caused by cells with positive anomaly contribution, when the widest area network excludes all cells with positive anomaly contribution (i.e., all cells in the screening sequence), the performance index will inevitably be within the first threshold range, meaning that the candidate cells are at most all cells in the screening sequence.

[0136] In some embodiments, refer to Figure 5 Identifying at least some candidate cells as problem cells (S1024) includes:

[0137] S10241. Determine multiple historical performance indicators for each candidate cell; each historical performance indicator is the performance indicator of the cell at a historical moment before the subnet's performance indicator became abnormal.

[0138] S10242. Determine the historical anomaly ratio of each candidate cell; the historical anomaly ratio of each cell is the proportion of historical performance indicators that exceed the first threshold range among all historical performance indicators of that cell.

[0139] S10243. Remove cells with a historical anomaly ratio higher than the second threshold from the candidate cells.

[0140] Based solely on the abnormal contribution rate, the candidate cells identified above are the problematic cells that caused the anomalies. However, the historical status of these candidate cells has not been taken into account.

[0141] For example, some cells may have consistently poor performance metrics (e.g., performance metrics frequently exceed the first threshold range); these cells are called "poor metric cells." Poor metric cells typically have a large abnormal contribution, and therefore are likely to be identified as candidate cells.

[0142] However, the performance indicators of cells with poor performance metrics have been consistently low, while the subnet has remained normal. This suggests that the anomalies in this case may not be caused by the cells with poor performance metrics, but rather by a sudden "deterioration" in the performance metrics of other cells. In other words, the cells with poor performance metrics should be filtered out when identifying the problem cells.

[0143] Therefore, for each candidate cell, its performance metrics (historical performance metrics) can be calculated at multiple historical moments before the subnet anomaly (multiple historical moments can all be within a predetermined time range before the anomaly occurs). For example, if the metrics are calculated every 15 minutes in the 5 hours before the anomaly, a total of 20 historical performance metrics at 20 historical moments can be obtained.

[0144] Next, it is determined whether each historical performance indicator exceeds the first threshold range, that is, whether the cell's performance indicator is poor at each historical moment. The first threshold range can be a fixed value; or if the first threshold range is the above real-time value, then each historical performance indicator can be compared with the first threshold range of its corresponding historical moment.

[0145] Furthermore, the proportion of historical performance indicators that exceed the first threshold range among all historical performance indicators of each candidate cell is determined, i.e., the historical anomaly ratio. For example, if a cell has 20 historical performance indicators, and 16 of them exceed the first threshold range, then the historical anomaly ratio of that cell = 18 / 20 = 90%.

[0146] Clearly, the higher the historical anomaly rate of a cell, the worse its overall performance indicators have been in the past, and the more likely it is to be classified as a poor-performing cell. Therefore, the historical anomaly rate of each candidate cell can be compared with a preset second threshold (e.g., 80%). If the historical anomaly rate of a cell (e.g., 90%) is greater than the second threshold (e.g., 80%), then the candidate cell is confirmed to have poor long-term performance indicators and is classified as a poor-performing cell, and can be filtered out from the candidate cells.

[0147] By using the above methods, cells with poor long-term performance indicators are filtered out, and the remaining candidate cells should be the problem cells that are actually causing the anomalies, making the problem cells identified more accurate.

[0148] In some embodiments, after removing cells with a historical anomaly ratio higher than a second threshold from the candidate cells (S10243), the method further includes:

[0149] S10244. If there are remaining candidate cells, all remaining candidate cells are identified as problem cells; if there are no remaining candidate cells, the cell with the largest abnormal contribution is identified as the first pre-positioned cell.

[0150] After excluding cells with poor quality indicators, if at least one candidate cell remains (i.e., it has not been filtered out), the remaining candidate cell can be directly regarded as the problem cell.

[0151] However, if there are no remaining candidate cells after excluding cells with poor index quality, it indicates that the above filtering process is unreasonable. Therefore, the specific cells with the largest abnormal contribution (i.e., the first pre-positioned cells) should be selected directly from all cells (of course, all cells before filtering) as problem cells. For example, the three cells with the largest abnormal contribution should be selected as problem cells (i.e., the first pre-positioned cells are the three).

[0152] For example, the following is a specific example of an anomaly determination problem cell for ratio-based performance metrics.

[0153] Suppose that subnet 370801 of a certain LTE network experiences an abnormal cell availability performance index at 02:15 on December 5, 2019, that is, the cell availability is lower than the lower limit of the first threshold range, as shown in the table below:

[0154] The cell availability rate is equal to the ratio of the actual available time within the statistical period to the total duration of the statistical period. When the anomaly occurred, the actual available time was 614,700 seconds, and the total duration of the period was 722,700 seconds. Therefore, the cell availability rate at this time was 0.85056, which is lower than the lower limit of the first threshold range.

[0155] Accordingly, the actual available time and total statistical period of each cell in the subnet are obtained, and their abnormal contribution is calculated. Taking cell (210517, 210517, 12) as an example, the actual available time of this cell is 0 seconds, and the statistical period is 900 seconds, then we can obtain:

[0156] The abnormal contribution of the cell (210517, 210517, 12) = -1 * [614700 / 722700 - (614700 - 0) / (722700 - 900)] = 0.106055 * 10 -2 .

[0157] Cells with anomaly contribution values ​​greater than 0 are retained and sorted in descending order to obtain the following sequence for screening:

[0158]

[0159] Following the above method, starting with N=1, cells are excluded. The optimized performance metrics obtained after excluding different numbers of cells are as follows: ...、0.903225、0.904352、0.90548、.....、0.956639、0.957359、0.958079

[0161] After excluding 66 cells, the optimized performance index of 0.958079 becomes greater than the lower limit of the first threshold range of 0.957941, thus yielding a total of 66 candidate cells.

[0162] Among the 66 candidate cells, some cells have a cell availability rate of 0. Therefore, the contribution of these cells to the anomalies cannot be reflected by the anomaly contribution alone.

[0163] Therefore, it is necessary to filter out cells with poor performance indicators. It is necessary to determine whether the proportion of each cell whose availability is not within the first threshold range exceeds the second threshold (e.g., 80%) within a certain period before the abnormal moment. If it does, the cell will be removed from the candidate cells.

[0164] Therefore, the remaining candidate cells can be identified as problem cells.

[0165] Moreover, these problematic cells are not those with consistently poor availability over a long period. Instead, they are cells whose availability suddenly and significantly deteriorated during or shortly before a subnet anomaly. In other words, they are the cells that actually caused the availability of the subnet to fall below the first threshold (the current anomaly).

[0166] As another embodiment of this disclosure, the following describes a specific method for identifying problematic cells when numerical performance indicators are abnormal.

[0167] Numerical performance metrics are directly parametric statistics, rather than being calculated from parametric statistics.

[0168] For example, numerical performance indicators may include rate, flow rate, etc., and the corresponding parameter statistics are rate and flow rate.

[0169] Of course, the first and second parameter statistics used to calculate ratio-type performance indicators can also be regarded as numerical performance indicators, such as the number of connected calls, the number of dropped calls, the number of congestion events, the number of successful handovers, the number of connection attempts, the number of calls, the number of data transmissions, and the number of handover attempts.

[0170] Therefore, the performance index of the subnet at this time is directly equal to the sum of the performance indices of all its cells.

[0171] In some embodiments, for numerical performance metrics, the anomaly contribution of each cell is determined based on the ratio of the deviation of the cell's performance metric to the deviation of the subnet's performance metric.

[0172] The performance index deviation of each cell is the difference between its performance index and the predicted performance index. The performance index deviation of a subnet is the difference between its performance index and the predicted performance index. The predicted performance index of each cell is the normal value of the predicted performance index of that cell. The predicted performance index of a subnet is the sum of the predicted performance indices of all its cells.

[0173] In other words, when the performance metric is a numerical performance metric, the anomaly contribution of each cell can be calculated in the following way:

[0174] A predicted performance metric is set for each cell, which is the "normal value" of the predicted performance metric for the cell. Correspondingly, the sum of the predicted performance metrics of all cells in the subnet is the predicted performance metric of the subnet, which is also the "normal value" of the predicted performance metric for the subnet.

[0175] Obviously, when a subnet is abnormal, the subnet's performance metrics will inevitably deviate from its predicted performance metrics (i.e., deviate from the normal value); at the same time, at least some cells in the subnet will also have their predicted performance metrics deviate from their predicted performance metrics. That is, the subnet's performance metrics deviating from the normal value must be caused by the performance metrics of its individual cells deviating from their respective normal values.

[0176] Therefore, based on the difference between the cell's performance index and the predicted performance index, the deviation value of the cell's performance index can be obtained, that is, the degree to which the cell's performance index deviates from the normal value. Similarly, based on the difference between the subnet's performance index and the predicted performance index, the deviation value of the subnet's performance index can be obtained, that is, the degree to which the subnet's performance index deviates from the normal value.

[0177] Furthermore, the ratio of the performance index deviation of the cell to the performance index deviation of the subnet represents the proportion of the cell's performance index deviation in the subnet's performance index deviation, or the extent to which the subnet's performance index deviation (i.e., anomaly) is caused by the cell, hence the cell's anomaly contribution.

[0178] Specifically, the anomalous contribution of the i-th cell can be calculated using the following formula:

[0179] The anomaly contribution of cell i = (performance index of cell i - predicted performance index of cell i) / (performance index of subnet - predicted performance index of subnet).

[0180] Of course, it should be understood that since the performance indicators of cells and subnets may deviate from their respective predicted performance indicators (normal values) in different directions (may be larger or smaller), the abnormal contribution calculated above may also be positive, negative or 0.

[0181] Correspondingly, if the anomaly contribution of a cell is positive, it indicates that the cell has a "positive contribution" to anomalies, or in other words, the cell's role is to cause anomalies in the subnet.

[0182] If the anomaly contribution of a cell is negative, it indicates that the cell has a "negative contribution" to anomalies, or that the cell's role is to prevent anomalies from occurring in the subnet.

[0183] If the anomaly contribution of a cell is 0, it indicates that the cell has no impact on anomalies generated by the subnet.

[0184] In some embodiments, the predicted performance metric for each cell is the average of the cell's performance metric over a predetermined period of time before the subnet's performance metric becomes abnormal.

[0185] In other words, the average performance index of a cell over a period of time before the subnet anomaly occurs can be used as the predicted performance index of that cell. For example, the average performance index of the cell can be taken as the predicted performance index of the cell during the same time period each day within the 15 days before the anomaly occurs.

[0186] Correspondingly, the predicted performance index of the subnet is actually the average of the subnet's performance index over the above time period.

[0187] Of course, it should be understood that it is also feasible to determine the predictive performance indicators of each cell in the subnet by manually setting them based on experience.

[0188] In some embodiments, refer to Figure 6 Based on the abnormal contribution of each cell in the subnet, at least one cell is identified as a problem cell (S102), including:

[0189] S1025. Determine the third threshold based on the abnormal contribution of all cells with positive abnormal contribution. The third threshold is a positive number.

[0190] S1026. If there are cells with abnormal contribution exceeding the third threshold, all cells with abnormal contribution exceeding the third threshold are identified as problem cells; if there are no cells with abnormal contribution exceeding the third threshold, the second pre-positioned cell with the largest abnormal contribution is identified as a problem cell.

[0191] As mentioned before, the abnormal contribution of each cell in the subnet only represents the relative value of its correlation with the abnormality. However, based solely on the abnormal contribution of each cell in the subnet, it is not possible to directly determine how many cells should be selected as problem cells.

[0192] Therefore, a third threshold can be calculated based on the abnormal contribution of all cells with positive abnormal contribution (i.e., all cells that may cause abnormalities), and cells whose abnormal contribution exceeds the third threshold (if they exist) can be identified as problem cells.

[0193] Of course, if it is found that no cell actually has an abnormal contribution exceeding the third threshold, then the abnormality can be considered to be caused by a common problem in multiple cells, and the specific cells with the largest abnormal contribution (i.e., the first two pre-positioned cells) (i.e., all cells) can be directly identified as problem cells. For example, the 10 cells with the largest abnormal contribution can be selected as problem cells (i.e., the second pre-position is 10).

[0194] In some embodiments, the third threshold is calculated using the following formula:

[0195] The third threshold = the average of the abnormal contribution of all cells with positive abnormal contribution + k * the standard deviation of the abnormal contribution of all cells with positive abnormal contribution; where k is a value greater than 0.

[0196] As a specific method, the average and standard deviation of the abnormal contribution of all cells with positive abnormal contribution can be calculated, and the average plus a certain multiple of the standard deviation can be used as the third threshold mentioned above.

[0197] For example, a cell whose abnormal contribution exceeds the average plus three times the standard deviation (i.e., k=3) can be defined as a problem cell.

[0198] Of course, it should be understood that the algorithm for the third threshold above is merely an example, and its specific calculation method may differ. For instance, the third threshold could be equal to the average of the abnormal contribution values ​​of all cells with positive abnormal contribution values.

[0199] For example, the following is a specific example of a problem cell for anomaly determination of numerical performance metrics.

[0200] Suppose that at 21:00 on December 9, 2019, subnet 370400 of a certain LTE network experienced an anomaly in the numerical performance indicators of the user plane uplink data volume.

[0201] First, calculate the outlier contribution of each cell in the subnet using the method described above, retain the positive outlier contributions, and sort them. The results are as follows:

[0202]

[0203]

[0204] Furthermore, the average and standard deviation of the abnormal contribution values ​​of the above-mentioned positive abnormal contribution values ​​are determined, that is, the following is determined:

[0205] The third threshold = mean + 3 * standard deviation = 0.018652;

[0206] Therefore, cells with an abnormal contribution rate exceeding 0.018652 can be selected as problem cells.

[0207] As can be seen, in this embodiment of the disclosure, the specific algorithms for determining problem cells are different when numerical performance indicators and ratio performance indicators are abnormal.

[0208] Therefore, refer to Figure 7 Logically, this embodiment of the disclosure needs to first obtain the performance indicators of the subnet in order to determine whether the performance indicators of the subnet are abnormal; when an abnormality occurs, it is necessary to first determine whether the performance indicator exceeding the first threshold range is a numerical performance indicator or a ratio performance indicator, and then process it in the corresponding manner.

[0209] The main differences between the algorithms for numerical performance metrics and ratio-based performance metrics include:

[0210] (1) The methods for calculating the abnormal contribution of a cell are different. This is because the ratio-type performance index of a subnet is not equal to the sum of the ratio-type performance indices of each cell, so the ratio-type performance indices of each cell cannot be added together.

[0211] (2) The calculation of anomalies in numerical performance indicators does not include the step of filtering out “poor quality cells”. This is because the calculation of the anomaly contribution of numerical performance indicators already takes into account the historical performance indicators (predicted performance indicators) of the cells, while the calculation of the anomaly contribution of ratio-based performance indicators only considers the data of each cell at the time of the anomaly; therefore, when there is an anomaly in the numerical performance indicator, it is not necessary to filter out cells with historically poor performance indicators (poor quality cells) separately.

[0212] As can be seen, in this embodiment of the disclosure, different specific algorithms for determining problem cells are proposed for numerical performance indicators and ratio performance indicators, respectively. Each algorithm is particularly suitable for the corresponding performance indicator, so that accurate problem cells can be obtained for various performance indicators.

[0213] Furthermore, the embodiments disclosed herein can be implemented entirely automatically without human intervention, are unaffected by human factors, and are fast, intelligent, and accurate.

[0214] Secondly, referring to Figure 8 This disclosure provides an electronic device, which includes:

[0215] One or more processors;

[0216] A memory that stores one or more programs, which, when executed by one or more processors, enable the one or more processors to implement any of the methods for identifying problem cells described above.

[0217] One or more I / O interfaces are connected between the processor and memory and configured to enable information exchange between the processor and memory.

[0218] Among them, the processor is a device with data processing capabilities, including but not limited to the central processing unit (CPU); the memory is a device with data storage capabilities, including but not limited to random access memory (RAM, more specifically SDRAM, DDR, etc.), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), and flash memory (FLASH); the I / O interface (read-write interface) is connected between the processor and the memory, enabling information exchange between the memory and the processor, including but not limited to the data bus (Bus).

[0219] Thirdly, referring to Figure 9 This disclosure provides a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements any of the methods described above for identifying problematic cells.

[0220] Those skilled in the art will understand that all or some of the steps, systems, and devices disclosed above, as functional modules / units, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0221] In hardware implementations, the division between functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be executed by several physical components working together.

[0222] Some or all of the physical components may be implemented as software executed by a processor, such as a central processing unit (CPU), digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit (ASIC). Such software may be distributed on a computer-readable medium, which may include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technique for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, random access memory (RAM, more specifically SDRAM, DDR, etc.), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory (FLASH) or other disk storage; read-only optical disc (CD-ROM), digital versatile disc (DVD) or other optical disc storage; magnetic cartridges, magnetic tapes, disk storage or other magnetic storage; and any other media that can be used to store desired information and can be accessed by a computer. Furthermore, as is known to those skilled in the art, communication media typically contain computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0223] This disclosure has disclosed exemplary embodiments, and although specific terminology has been used, it is for general illustrative purposes only and should not be construed as limiting. In some instances, it will be apparent to those skilled in the art that features, characteristics, and / or elements described in conjunction with particular embodiments may be used alone, or in combination with features, characteristics, and / or elements described in conjunction with other embodiments, unless otherwise expressly indicated. Therefore, those skilled in the art will understand that various changes in form and detail may be made without departing from the scope of this disclosure as set forth by the appended claims.

Claims

1. A method for identifying problem cells, comprising: determining abnormal contribution degrees of cells in a subnet; determining at least one cell as a problem cell according to the abnormal contribution degrees of the cells in the subnet; wherein, the abnormal contribution degree of each cell is a degree of association between a performance indicator of the cell and an abnormality of a performance indicator of the subnet when the abnormality occurs, the abnormality of the performance indicator of the subnet is that the performance indicator of the subnet is out of a first threshold range, the performance indicator of the subnet is determined according to parameter statistics of the cells in the subnet, and the performance indicator of each cell is determined according to parameter statistics of the cell; the performance indicator is a ratio type performance indicator, and the ratio type performance indicator is a ratio of a first parameter statistic to a second parameter statistic; the abnormal contribution degree of each cell represents a degree of shift of the performance indicator of the subnet after removing the cell from the subnet to the first threshold range relative to the performance indicator of the subnet before removing the cell; 2. The method of claim 1, wherein, the first threshold range comprises: a first threshold range lower limit and / or a first threshold range upper limit. 3.The method of claim 1, wherein, the first threshold range is determined by a dynamic threshold detection technique.

4. The method of claim 1, wherein, the determining at least one cell as a problem cell according to the abnormal contribution degrees of the cells in the subnet comprises: arranging cells with positive abnormal contribution degrees in descending order to obtain a to-be-screened sequence, and determining N as 1; determining an optimized performance indicator, the optimized performance indicator being a difference between the performance indicator of the subnet and a temporary performance indicator of the subnet, the temporary performance indicator being a performance indicator of the whole of the remaining cells in the subnet after removing the top N cells in the to-be-screened sequence; if the optimized performance indicator is within the first threshold range, determining the top N cells in the to-be-screened sequence as candidate cells, and if the optimized performance indicator is out of the first threshold range, increasing N by 1 and returning to the step of determining the optimized performance indicator; determining at least part of the candidate cells as problem cells.

5. The method of claim 4, wherein, the determining at least part of the candidate cells as problem cells comprises: determining a plurality of historical performance indicators of each candidate cell, each historical performance indicator being a performance indicator of the cell at a historical time before the abnormality of the performance indicator of the subnet occurs; determining a historical abnormality proportion of each candidate cell, the historical abnormality proportion of each cell being a proportion of historical performance indicators out of the first threshold range in all historical performance indicators of the cell; removing a cell with a historical abnormality proportion higher than a second threshold from the candidate cells.

6. The method of claim 5, wherein, after the removing a cell with a historical abnormality proportion higher than a second threshold from the candidate cells, further comprising: if there are remaining candidate cells, determining all the remaining candidate cells as problem cells; if there are no remaining candidate cells, determining the top first predetermined number of cells with the largest abnormal contribution degrees as problem cells.

7. The method of claim 1, wherein, the performance indicator is a numerical value type performance indicator, and the numerical value type performance indicator is a parameter statistic; the abnormal contribution degree of each cell is determined according to a ratio of a performance indicator deviation value of the cell to a performance indicator deviation value of the subnet. The performance index deviation of each cell is the difference between the performance index of the cell and the predicted performance index, the performance index deviation of the subnet is the difference between the performance index of the subnet and the predicted performance index, the predicted performance index of each cell is the normal value of the predicted performance index of the cell, and the predicted performance index of the subnet is the sum of the predicted performance indexes of the cells.

8. The method of claim 7, wherein, The predicted performance index of each cell is the average value of the performance index of the cell within a predetermined time before the performance index of the subnet is abnormal.

9. The method of claim 7, wherein, The determining of the at least one cell as the problem cell according to the abnormal contribution degrees of the cells of the subnet comprises: determining a third threshold according to the abnormal contribution degrees of all the cells with positive abnormal contribution degrees, the third threshold being a positive number; if there is a cell with an abnormal contribution degree exceeding the third threshold, determining all the cells with abnormal contribution degrees exceeding the third threshold as the problem cells; if there is no cell with an abnormal contribution degree exceeding the third threshold, determining the cells with the top second predetermined positions of the abnormal contribution degrees as the problem cells.

10. The method of claim 9, wherein, The third threshold is calculated by the following formula: Third threshold = average value of the abnormal contribution degrees of all the cells with positive abnormal contribution degrees + k * standard deviation of the abnormal contribution degrees of all the cells with positive abnormal contribution degrees; wherein k is a value greater than 0.

11. An electronic device, comprising: one or more processors; a memory having one or more programs stored thereon, when the one or more programs are executed by the one or more processors, the one or more processors implement the method for identifying a problem cell according to any one of claims 1 to 10; one or more I / O interfaces connected between the processor and the memory, configured to realize the information interaction between the processor and the memory.

12. A computer readable medium having a computer program stored thereon, when the program is executed by a processor, the method for identifying a problem cell according to any one of claims 1 to 10 is implemented.

Images