Operation prediction in wireless communication network

Abstract

A method of managing a wireless communication network, comprising: obtaining data regarding performance of a cell of the wireless communication network; and generating a prediction of values of a plurality of key performance indicators (KPIs) related to uplink (UL) throughput for the cell of the wireless communication network based on the obtained data. The method comprises: generating a prediction of UL throughput that the cell will experience degradation at a future time based on a prediction value of the KPI; and determining, from among the plurality of KPIs, a set of candidate root cause KPIs associated with the predicted degraded UL throughput. The method further includes selecting an activity based on the determined set of candidate KPIs; and applying the activity to the wireless communication network.

Description

Technical Field

[0001] The present disclosure relates generally to network management of wireless communication networks, and in particular, to systems and methods for predicting uplink (UL) throughput with proactive repair in fifth generation (5G) networks. Background Art

[0002] With the large-scale commercialization of 5G networks, operators may increasingly turn their attention to opportunities in the business-to-business market. A high-quality 5G network should accommodate the massive data capacity requirements for uplink (UL) data transmission to support high-definition video, online gaming, big data collection, smart surveillance, alternate reality / virtual reality, live video, and other uplink data-intensive services. The growing demand for uplink data transmission requires continuous improvements in network capacity and throughput.

[0003] 5G implementations can be non-standalone (NSA) implementations that interwork with non-5G network elements, or standalone (SA) implementations that include only 5G network elements. In addition, 5G networks may involve multiple radio access technology (RAT) components, especially in NSA architectures. In addition to the architecture, dynamic spectrum sharing allows instant sharing of spectrum resources between fourth generation (4G) and 5G in NSA implementations. Spectrum sharing in Figure 1 As shown in the figure. Figure 1 As shown in, for example, a wireless communication network such as a 4G network can occupy the entire 20 MHz bandwidth. Alternatively, the 4G network can share the 20 MHz bandwidth with the 5G network, where each network occupies a separate 10 MHz bandwidth, or the two networks can use dynamic spectrum sharing (DSS) to share the same 20 MHz bandwidth.

[0004] Due to the increased demand for UL data transmission in 5G networks, it is important for 5G network operators to ensure that sufficient UL resources are available to users.

[0005] Some efforts have been made to use machine learning to predict UL throughput in communication networks. However, previous efforts have not addressed the management of UL throughput in multi-RAT environments presented by 5G NSA or SA architecture implementations or spectrum sharing components introduced in 5G technology. In addition, many ML solutions are not proactive ML solutions, but are predictive in nature. Summary of the invention

[0006] Some embodiments provide a method of managing a wireless communication network, comprising: obtaining data about performance of a cell of the wireless communication network; and generating, based on the obtained data, a prediction of values ​​of a plurality of key performance indicators (KPIs) related to uplink (UL) throughput for the cell of the wireless communication network. The method comprises: generating a prediction of a degraded UL throughput that the cell will experience at a future time based on the predicted values ​​of the KPIs; and determining, from among the plurality of KPIs, a set of candidate root cause KPIs associated with the predicted degraded UL throughput. The method further comprises: selecting an activity based on the determined set of candidate KPIs; and applying the activity to the wireless communication network.

[0007] The data regarding the performance of the cell may include performance measurement data and / or configuration management data.

[0008] Predicting the value of the KPI may include generating features for a machine learning (ML) model based on the data regarding the performance of the cell.

[0009] Generating the feature may include generating a moving average of the KPI over one or more past time periods.

[0010] Generating the feature may include generating a lagged value of the KPI from a past time period.

[0011] The method may further comprise generating a threshold value for the KPI based on historical values ​​of the KPI.Generating the prediction that the cell will experience degraded uplink throughput may be based on a comparison of the predicted value of the KPI and the threshold value for the KPI.

[0012] The threshold value of the KPI may be obtained by performing exploratory data analysis on historical data from the wireless communication network. In some embodiments, the threshold value of the KPI may be obtained based on an operating frequency band and bandwidth of the wireless communication network.

[0013] The method may also include: classifying the KPIs into multiple KPI categories; selecting the ML model from among multiple ML models based on the classification of at least one of the KPIs; and applying the ML model to generate the prediction of the value of the at least one of the KPIs.

[0014] The method may further include: selecting a plurality of ML models based on the classification of the KPI; and applying the plurality of ML models.

[0015] The KPIs may be categorized as being related to availability, accessibility, retainability, mobility, coverage, quality, utilization, transmission ratio, and / or accessibility of the cell.

[0016] The ML model may include a plurality of ML models, and the prediction that the cell will experience degraded UL throughput is based on outputs of the plurality of models.

[0017] The prediction that the cell will experience degraded UL throughput may be based on a weighted average of the outputs of the plurality of models.

[0018] The plurality of ML models may include one or more of an XGBoost model, a random forest model, a long short-term memory model, a CatBoost model, and a light gradient boosting model.

[0019] Determining the set of candidate root cause KPIs may include generating a ranking of the plurality of KPIs based on importance of the predicted UL throughput degradation; and selecting the set of candidate root cause KPIs based on the ranking of the KPIs by importance.

[0020] The ranking of the plurality of KPIs may be generated by applying a Tree Shapley Additional Interpretation (Tree SHAP) algorithm to the KPIs and UL throughput degradation prediction.

[0021] The method may further include classifying each candidate root cause KPI in the set of candidate root cause KPIs into one of a plurality of KPI categories.The activity may be selected based on the KPI category of the set of candidate root cause KPIs.

[0022] Selecting the activity may include: determining whether at least one of the candidate root cause KPI set is classified according to a first KPI category; when determining that at least one of the candidate root cause KPI set is classified according to the first KPI category, checking the operating conditions of the wireless communication network associated with the first KPI category; and selecting the activity based on the operating conditions of the wireless communication network associated with the first KPI category.

[0023] The method may also include repeating the following steps for multiple KPI categories: determining whether at least one of the candidate root cause KPI set is classified according to a KPI category, checking the operating conditions of the wireless communication network associated with the KPI category, and selecting the activity based on the operating conditions of the wireless communication network associated with the KPI category.

[0024] Some embodiments provide a network management system, comprising a processor and a memory coupled to the processor. The memory comprises computer program instructions, which when executed by the processor, cause the network management system to perform operations, the operations comprising: obtaining data regarding performance of a cell of a wireless communication network, generating a prediction of values ​​of a plurality of KPIs related to UL throughput for the cell of the wireless communication network based on the obtained data, generating a prediction that the cell will experience degraded UL throughput at a future time based on the predicted values ​​of the KPIs, determining a set of candidate root cause KPIs associated with the predicted degraded UL throughput from among the plurality of KPIs, selecting an action based on the determined set of candidate KPIs, and applying the action to the wireless communication network.

[0025] Some embodiments provide a computer program comprising computer code to be executed by a network management system, the network management system being configured to perform operations comprising: obtaining data regarding performance of a cell of a wireless communication network, generating a prediction of values ​​of a plurality of KPIs related to UL throughput for the cell of the wireless communication network based on the obtained data, generating a prediction that the cell will experience degraded UL throughput at a future time based on the predicted values ​​of the KPIs, determining a set of candidate root cause KPIs associated with the predicted degraded UL throughput from among the plurality of KPIs, selecting an activity based on the determined set of candidate KPIs, and applying the activity to the wireless communication network.

[0026] Some embodiments provide a computer program product comprising a non-transitory storage medium, the non-transitory storage medium comprising program code to be executed by a processing circuit of a network management system, whereby execution of the program code causes an apparatus to perform operations comprising: obtaining data regarding performance of a cell of a wireless communication network, generating a prediction of values ​​of a plurality of KPIs related to UL throughput of the cell of the wireless communication network based on the obtained data, generating a prediction of a degraded UL throughput that the cell will experience at a future time based on the predicted values ​​of the KPIs, determining a set of candidate root cause KPIs associated with the predicted degraded UL throughput from among the plurality of KPIs, selecting an activity based on the determined set of candidate KPIs, and applying the activity to the wireless communication network.

[0027] Some embodiments described herein may provide certain advantages. For example, they may enable proactive identification and correction of UL throughput problems before they occur. By predicting UL throughput problems in advance, the network may have sufficient time to implement activities to prevent degradation. This may provide an improved customer experience for services that require high UL throughput, such as video conferencing, video games, virtual reality applications, etc. Some embodiments may improve the performance of cells in a wireless communication network, which may improve quality of service and optimize utilization of network resources. In addition, some embodiments may scale with network size, geography, and complexity, and may be applied to both SA and NSA architectures. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The accompanying drawings, which are included to provide a further understanding of the present disclosure and are incorporated in and constitute a part of this application, illustrate certain non-limiting embodiments of the inventive concept. In the drawings:

[0029] Figure 1 Illustrated is spectrum sharing in a wireless communication system.

[0030] Figure 2 Operations for predicting low UL throughput and self-healing of a cell in accordance with some embodiments are illustrated.

[0031] Figure 3 The operation of the system / method according to some embodiments is illustrated.

[0032] Figure 4 A cell marking operation according to some embodiments is illustrated.

[0033] Figure 5 Illustrated are examples of rankings of key performance indicators (KPIs) that may be output by a root cause mapping operation in accordance with some embodiments.

[0034] Figure 6 Illustrated are example operations for activity selection based on root cause identification operations in accordance with some embodiments.

[0035] Figure 7 Illustrated is a self-healing operation of a network management system according to some embodiments.

[0036] Figure 8 is a block diagram of a network management system according to some embodiments.

[0037] Fig. 9 is a flow chart illustrating the operation of a network management system according to some embodiments.

[0038] Fig.10 is a block diagram of a communication system according to some embodiments. DETAILED DESCRIPTION

[0039] The inventive concept will now be described more fully below with reference to the accompanying drawings, in which examples of embodiments of the inventive concept are shown. However, the inventive concept can be embodied in many different forms and the inventive concept should not be interpreted as being limited to the embodiments set forth herein. On the contrary, these embodiments are provided so that this disclosure will be exhaustive and complete, and will fully convey the scope of the inventive concept to those skilled in the art. It should also be noted that these embodiments are not mutually exclusive. A component from one embodiment may be assumed by default to be present in / used in another embodiment.

[0040] The following description presents various embodiments of the disclosed subject matter. These embodiments are presented as teaching examples and should not be interpreted as limiting the scope of the disclosed subject matter. For example, certain details of the described embodiments may be modified, omitted or elaborated without departing from the scope of the described subject matter.

[0041] The following explanations of potential problems and some approaches are current implementations as part of the present disclosure and should not be construed as being previously known by others.

[0042] Some embodiments described herein utilize machine learning (ML) concepts to predict in advance which cells of a wireless communication network may suffer from symptoms of low UL throughput. In addition, some embodiments may select activities that can be applied to the network before the cell suffers a predicted UL degradation. The selection of activities to be applied to the network may be based on an analysis of factors that contribute to the prediction of UL degradation. The selection and application of activities based on predicted UL throughput degradation is referred to as "self-repair" in this article. However, it will be understood that the systems and / or methods as described herein can avoid UL throughput degradation before it occurs.

[0043] Figure 2 A general overview of operations for predicting low UL throughput and self-healing of a cell according to some embodiments is provided. Figure 2 As shown in FIG. 2 , a wireless communication network 200 may include multiple cells in which wireless communication devices may obtain service.

[0044] A secondary key performance indicator (KPI) is a KPI that may contribute to a primary KPI of interest, such as UL throughput. The secondary KPI may be defined by a standard such as the New Radio (NR) standard. Secondary KPIs that may be considered include quantities such as UL received signal strength indicator (RSSI), block error rate (BLER), channel quality indicator (CQI), signal to interference plus noise ratio (SINR), reference signal received power (RSRP), ranking indicator, path loss, packet loss, dual connectivity establishment scheduling request (SR), modulation KPI, resource block (RB) symbol utilization, number of users, availability, etc.

[0045] Secondary KPIs may be classified as belonging to categories such as accessibility, retainability, mobility, integrity, availability, and quality. Secondary KPIs may be measured for each cell, and the measured KPI data may be processed by an overall ML algorithm as described in more detail below to predict which cells may suffer from degraded UL throughput at a future time (e.g., two hours in advance). Figure 2 In the example shown in , based on analysis of the KPI measurements, it is predicted that Cell 2 and Cell 4 will suffer degradation within two hours.

[0046] In the self-healing process, one or more activities may be selected based on the KPIs that are important to the prediction of UL degradation and applied to the network. The activities may include actions that may be taken automatically or may be taken by the network operator to change the operation of the network, and the activities are selected to avoid the predicted UL throughput degradation. The activities may include actions such as optimizing network parameters, optimizing thresholds, checking various network conditions, etc. The system / method continues to monitor the KPIs to ensure that KPI degradation does not occur. Figure 2 As shown in , after application of the activity(s), cells 2 and 4 are considered "recovered" (even though they may never actually experience the predicted UL throughput degradation).

[0047] In some embodiments, the ML algorithm is trained on the secondary KPI to identify in advance the cells that may show UL throughput degradation issues in the future (eg, within a few hours).

[0048] The ML algorithm uses performance management (PM) counters and configuration management (CM) data to study the cell behavior under different conditions. The overall algorithm correlates the performance of important KPIs in the past hours with the behavior of the cell in the future.

[0049] Figure 3 The operation of the system / method according to the embodiments described herein is illustrated. Figure 3 At block 302, a network management system according to some embodiments collects data related to a wireless communication network 300. In particular, data may be collected from an operations support system (OSS) of the network 300. The collected data may include performance management (PM) data, configuration management (CM) data, and / or data regarding the operation of a particular site or node within the network. The data may include PM counters, CM counters, cell / site databases, and / or other information.

[0050] At block 304, the system / method performs feature engineering to generate a set of features from the collected data, which can be input into the ML model to generate a prediction of UL throughput degradation for a given cell of the network. Feature engineering involves developing appropriate relationships between key counters collected during the data collection process. In wireless communication networks, the values ​​of KPIs have time dependence. That is, future values ​​of the KPI are statistically correlated with and dependent on past values ​​of the KPI. For this reason, time-dependent features (such as the average and lag of the KPI value) can be selected as some of the features to be input into the ML model.

[0051] The inventors have found that in a wireless communication network, there is a good correlation between hysteresis values ​​(e.g., 15-minute and 30-minute hysteresis values) and moving averages of KPIs related to UL throughput. For example, moving averages of KPIs calculated over the past few hours, the past day, and the past week can provide insights into upcoming cell behavior.

[0052] "Lagging" a time series means shifting its values ​​forward by one or more time steps, or equivalently, shifting the time in its index backward by one or more steps. For example, Table 1 shows the variable "y" along with the lagged values ​​"y_lag_1" and "y_lag_2" that are lagged by one and two time steps (months in this example), respectively. Table 1 - Examples of data lags

[0053] In Table 1, column “y” is taken as reference and the lagged values ​​are generated by shifting the values ​​by one and two time steps.

[0054] Hysteresis values ​​of the primary and secondary KPIs may help provide accurate UL throughput predictions. For example, the following formula may help predict uplink throughput based on a performance measurement (PM) KPI associated with information about the UL data volume (mac-volume) and PUSCH scheduling activity (pusch-sched-activity) in the MAC entity: Uplink throughput = (64*(mac-volume / (pusch-sched- activity*1000)))

[0055] A combination of frequency band and bandwidth from the cell definition may be used to determine a dynamic threshold setting for a target label for UL throughput degradation.

[0056] Once the relevant features have been defined, the feature data is processed by the ML model 306, which generates a prediction of UL degradation for the cell. In particular, the ML model may be a classification model that labels a cell as degraded or not degraded based on an analysis of the relevant features. In particular, the ML model may be a regression model that predicts the UL throughput value of a cell if Figure 4 If a cell is downgraded as illustrated in and described below, the cell may be marked.

[0057] In some embodiments, the ML model 306 may be an overall ML model that combines the outputs of several different ML algorithms. For example, the outputs of the different ML algorithms may be combined as a weighted average, and the labeling of the cell may be based on the combined outputs of the ML algorithms. This approach is based on the finding that different features may provide different levels of predictive performance for UL throughput degradation when processed by different models.

[0058] Therefore, some embodiments use different classification algorithms trained for specific features. Some algorithms considered include XGBoost, CatBoost Random Forest, Long Short-Term Memory (LSTM), and Light Gradient Boosting Machine (LightGBM).

[0059] The overall ML algorithm is based on the following assumption: UL throughput degradation occurs when there is degradation in the following categories of KPIs: Availability Accessibility (ENDC setting SR) Retainability Mobility (Change SR) Coverage & Quality (RSRP / UL RSSI / BLER / CQI / RANK / SINR / Path Loss) Utilization (payload / number of users / RB symbol utilization / RRC connections & active users) Transmission ratio Integrity (packet loss)

[0060] To obtain a single prediction based on these different types of features, different ML models may be used, and the outputs of the different models may be combined (e.g., as a weighted average) to provide a final prediction of future cell classification as degraded or not degraded in terms of UL throughput.

[0061] The performance of several different models including Random Forest, CatBoost and Long Short Term Memory (LSTM) in terms of prediction accuracy of UL throughput degradation was measured. The results are shown in Table 2, along with the results of an aggregate model that combines the outputs of the different models as described above. As seen in Table 2, the aggregate model performs better than any single model in each of the R2, Root Mean Square Error (RMSE), Precision, F1 score and Recall metrics. Table 2 — Comparison of metrics for different ML models measure Random Forest Catboost LSTM polymerization R2 0.65 0.73 0.72 0.82 RMSE 0.356 0.67 0.14 0.01 Accuracy 81.56 81.71 86.56 85.56 F1 score 85.63 82.63 88.93 92.93 Recall 75.60 65.74 80.76 86.76

[0062] Some systems / methods generate a prediction of cell degradation two hours in the future. Two hours was selected as the target future time for the prediction because it has been found that accurate predictions of UL throughput degradation are difficult to obtain based on KPI analysis of time periods greater than two hours, but relatively accurate predictions of UL throughput degradation within two hours can be obtained. In addition, knowing in advance that UL throughput degradation may occur within two hours thus provides sufficient time to implement activities on the network, which can help prevent the predicted UL throughput degradation from occurring. However, it will be appreciated that other time lags (including time lags greater than two hours and time lags less than two hours) may be used for predictions, depending on the nature of the particular system being managed, without departing from the scope of the inventive concept.

[0063] like Figure 3 As illustrated in , feedback from the system may be used to train the ML model(s) 306. It has been found that the ML model(s) may be adequately trained using performance data collected at 15 minute intervals over a month. Once the ML model(s) 306 have been adequately trained, a final prediction 308 is generated that marks the cell as predicted to be degraded or not degraded in the future.

[0064] In some embodiments, the decision to mark a cell as degraded or not degraded may be based on both primary KPI data and secondary KPI data, where "primary KPI data" refers to the KPI being predicted, i.e., UL throughput degradation, and "secondary KPI data" refers to the KPI that contributes to the value of the primary KPI. That is, in terms of UL throughput, marking a cell as predictively degraded or not degraded may not be based solely on the prediction of the UL throughput at the target time.

[0065] For example, Figure 4 For a brief reference, Figure 4 A cell marking operation according to some embodiments is shown, which is based on exploratory data analysis (EDA) results of the 10th and 90th percentile values ​​of throughput, symbol utilization, and RSSI KPI, respectively.

[0066] Generally speaking, data analysis involves finding trends in data through statistics and probability. EDA is a well-known step in data science that involves studying the properties of data, performing outlier analysis, checking for missing values, analyzing correlations, etc.

[0067] In particular, after obtaining the input feature list (block 402) and processing it using the overall model (block 404), the predicted values ​​of the relevant KPIs are obtained. Separately, at block 408, EDA is performed on the historical data collected from the network 300 and analyzed to obtain statistical measures of the data, such as the 10th percentile and 90th percentile values ​​associated with the data. A determination of whether a cell is predicted to be degraded or not degraded can be obtained by comparing various KPIs with the 10th percentile and / or 90th percentile values ​​associated with the KPIs. For example, at block 410, the system / method checks to see if the UL throughput (KPI_throughput) predicted by the overall model 404 for the cell is less than the 10th percentile of the UL throughput. If so, the system / method checks at block 412 to see if the predicted symbol utilization KPI (KPI_SMBL_UTIL) of the cell is less than the 10th percentile of the KPI. If yes, the system / method checks to see if the predicted KPI (KPI_RSSI) of the received signal strength indicator for the cell is less than the 90th percentile RSSI at block 414. If yes, the cell is marked as degraded at block 418. Otherwise, if any of the above tests are negative, the cell is marked as not degraded at block 416.

[0068] Reference again Figure 3 Once a prediction of future cell degradation has been generated, and a cell has been predictively marked as degraded, the system / method performs root cause identification at block 310 to determine the root cause of the predicted UL degradation. That is, root cause identification 310 attempts to determine which of the analyzed KPIs is most significant to the prediction of UL throughput degradation. By determining which KPI is most significant to the prediction, insight into the cause of the predicted UL throughput degradation may be gained. This insight then informs the selection of activities that may be applied to the network 300 in an effort to avoid the predicted degradation of UL throughput.

[0069] The root cause identification process may determine that a KPI is a significant cause for predicting throughput degradation based on whether the KPI exhibits abnormal behavior, such as by having a value that is far above or below the normal range of the KPI. To this end, a tree SHAP (Shapley appended explanation) algorithm may be used to discover the root cause KPIs. In the tree SHAP algorithm, the KPIs used as input to the ML ensemble model may be ranked by importance, and a subset of KPIs representing those KPIs determined to be most important for the prediction of UL throughput degradation may be selected as candidate root causes.

[0070] The technique for identifying the important features that lead to the prediction of UL throughput degradation can be based on the probability of each ML model producing a result. That is, from among the ML models deployed in the overall model, the model with the highest probability of producing the target classification can be selected and future analysis can be performed on the features input to that model.

[0071] right Figure 5 For a brief reference, Figure 5 An example of a ranking of KPIs that may be output by a root cause mapping operation based on the KPIs input to the LightGBM model is illustrated in accordance with some embodiments. Figure 5 In the example shown in , the ranked KPIs include KPIs related to packet loss (ul_packet_loss), average channel quality indicator (hpi_nr_nsa_cqi_avg), number of active UL users (kpi_nr_nsa_avg_active_ul_users), uplink block error rate (kpi_nr_nsa_mac_ul_bler), PDCCH blocking rate (kpi_nr_nsa_pdcch_blocking_ul), transmission ratio (kpi_nr_nsa_ul_16qam_transport_ratio), average rank (kpi_nr_nsa_ul_cqi64qamrank_avg), number of connected users (kpi_nr_nsa_max_endc_connection_users), setup scheduling request (kpi_nr_nsa_nr_endc_setup_sr), and cell downtime (kpi_nr_nsa_cell_downtime). From these KPIs, a subset (eg, the top five) may be selected as the most likely candidate causes for prediction.It will be appreciated that many other KPIs may be considered in the root cause mapping operation.

[0072] Reference again Figure 3, the subset of KPIs identified as candidate root causes in the root cause identification operation are then provided to the recommendation engine / activity selection block 312, which generates recommended actions to be applied to the network 300 based on the identified root causes to avoid the predicted UL throughput degradation. In particular, the recommended actions can be selected by considering the correlation of secondary KPIs (such as CQI, RSSI, packet loss, path loss, etc.) with UL throughput degradation. In certain embodiments, the system / method can select an action based on consideration of whether a particular type of KPI is considered a root cause factor. In determining the recommended action, the recommendation engine 312 can consider factors such as confidence scores for each prediction, consecutive counts of predictions, and business rules.

[0073] Figure 6 Example operations for activity selection based on root cause identification operations according to some embodiments are illustrated. Recommendations for activities, such as parameter adjustments, can be developed with input from subject matter experts. Some example activities may include modifying event-based thresholds (e.g., handover thresholds), offset adjustments (such as cell individual offset (CIO) adjustments), traffic balancing, remote electrical tilt (RET) changes, power changes, RRC connected user permission extensions, enabling UL-256QAM features, etc.

[0074] like Figure 6 As shown in , at block 602, features associated with the ML model with the highest prediction confidence are identified. Then, at block 604, the identified features are ranked, for example, using the tree SHAP algorithm as described above.

[0075] The identified features can then be classified according to type. For example, features can be classified into categories such as warning, cell downtime, traffic, radio conditions, DL packet loss, retainability, DL transmission ratio, new parameter changes, NR capacity in DSS, PDCCH restrictions, UL packet fragmentation, core network problems, etc.

[0076] At block 606, the system / method determines whether the important feature is classified as an availability feature. If so, the system / method checks to see if there is downtime in the service or neighboring (NBR) node (block 608), and if so, generates a recommended action to check for downtime in the node (block 610). If not, the operation proceeds to block 614 below.

[0077] Otherwise, at block 612, the system / method determines whether the important feature is classified as a traffic feature. If so, the system / method checks to see if the user count is above a threshold, and if the UL symbol utilization is above a threshold (block 614), and if so, generates a recommended action to optimize event-based thresholds and check for inter-PSCell and intra-PSCell change scheduling request (SR) degradation (block 616). If not, the operation proceeds to block 620 below.

[0078] Otherwise, at block 618, the system / method determines whether the significant feature is classified as a downlink packet loss feature. If so, the system / method checks to see if there is interference due to overshooting cells (block 620), and if so, generates a recommended action to optimize cell parameters to avoid overshooting (block 622). If not, operation proceeds to block 626 below.

[0079] Otherwise, at block 624, the system / method determines whether the important feature is classified as a radio condition feature. If so, the system / method checks to see if a KPI such as RSSI, SINR, CQI, or ranking is out of an optimal range (block 626), and if so, generates a recommended action to optimize cell parameters to correct the KPI (block 628). If not, operation proceeds to block 632 below.

[0080] Otherwise, at block 630, the system / method determines whether the significant feature is classified as a UL packet segment feature. If so, the system / method checks to see if the modulation and coding scheme (MCS) is out of range (block 632), and if so, generates a recommended action to optimize cell parameters to correct the MCS (block 634). If not, operation proceeds to block 638 below.

[0081] Otherwise, at block 636, the system / method determines whether the significant feature is classified as a UL transmission ratio feature. If so, the system / method checks to see if the QPSK samples are outside the optimal range (block 638), and if so, generates a recommended action to check the UL 256QAM feature (block 640).

[0082] If not, then at block 642 it is determined that a root cause could not be identified.

[0083] Reference again Figure 3 , at block 314, the selected actions are applied to the network 300 in an effort to avoid the predicted degradation of UL throughput. The system / method may continue to monitor the KPIs to ensure that the cell in question is repaired, i.e., it is no longer predicted to suffer UL throughput degradation.

[0084] Figure 7The self-healing operation according to some embodiments is illustrated. As shown therein, the system / method according to some embodiments can predict the cell UL throughput x hours in advance (box 702). In response to the predicted decrease in throughput in the cell, the system / method can cause the cell to switch / handover one or more user equipment connections from a candidate cell to a candidate cell to avoid the predicted degradation. To achieve this, the system / method can identify a healthy target cell based on KPIs of one or more candidate target cells, such as handover (HO) attempts and HO success ratio (HOSR) (box 704). Once the target cell is identified, the service is switched to the target cell (box 706), and after the switch, the system / method continues to monitor the KPIs in both the source cell and the target cell to predict future UL throughput degradation (box 708).

[0085] Thus, some embodiments provide operations that may be particularly suitable for managing the operation of 5G networks. According to some embodiments, UL throughput degradation can be predicted, and preemptive action can be taken in response to the prediction to mitigate potential performance issues.

[0086] Although primarily described herein with reference to 5G networks, some embodiments described herein may be used for other types of wireless communication networks, such as 4G / Long Term Evolution (LTE). In addition, some embodiments described herein may be applicable to both 5G bands FR1 (below 6 GHz) and FR2 (millimeter wave).

[0087] According to some embodiments, cells in a wireless communication network are classified as being in one of a plurality of operating modes, including pure LTE, ENDC-LTE, ENDC-NR, or spectrum sharing in NSA. The same approach can also be extended to include CA (carrier aggregation) cells in each of LTE / NR. Multiple KPIs can be considered in an active (or passive) model to obtain a holistic view of UL degradation. In addition, continuous degradation can be considered in both modeling and activities.

[0088] Some embodiments enable early detection of degraded cells (eg, up to two hours in advance), which allows operators to take preventive actions to improve network performance.

[0089] According to some embodiments, a degraded secondary KPI may be identified and the degradation of the secondary KPI may be used to predict degradation of a primary KPI, such as UL throughput degradation.Some embodiments enable self-healing of the network by proactively adjusting cell parameters to avoid predicted cell degradation.

[0090] The embodiments described herein may provide certain technical advantages. For example, proactively identifying and resolving UL throughput issues may improve cell performance by improving / balancing utilization of network resources. This in turn may improve customer experience for applications requiring high UL throughput, such as video conferencing, video gaming, virtual reality, and the like.

[0091] Figure 8 800 for managing a wireless communication network according to some embodiments. The apparatus 800 may be provided by, for example, an apparatus in the cloud running software on cloud computing hardware; or a software function / service that manages or controls a wireless communication network. That is, the apparatus may be implemented as part of a communication system (e.g., as described below with respect to Fig.10 The device may be a device that is part of the communication system 1000 discussed, or implemented on the device as a standalone functionality / service hosted in the cloud. The device may also be provided as standalone software for managing a wireless communication network; and the device may be in a deployment that may include virtual or cloud-based network functions (VNFs or CNFs) and even physical network functions (PNFs). The cloud may be public, private (e.g., on-premises or hosted), or hybrid.

[0092] As shown, the apparatus may include transceiver circuitry 800 (e.g., RF transceiver circuitry) including a transmitter and a receiver configured to provide uplink and downlink radio communications with an apparatus (e.g., a controller for automatically performing an activity). The apparatus may include network interface circuitry 808 (also referred to as a network interface) configured to provide communications with other apparatuses (e.g., a controller for automatically performing an activity). The apparatus may also include processing circuitry 803 (also referred to as a processor) coupled to the transceiver circuitry, memory circuitry 805 (also referred to as a memory) coupled to the processing circuitry.

[0093] As discussed herein, the operation of the device may be performed by the processing circuit 803, the network interface 808, and / or the transceiver 801. For example, the processing circuit 803 may control the device 800 to perform operations according to the embodiments disclosed herein. The processing circuit 803 may also control the transceiver 801 to transmit downlink communications to one or more devices on the radio interface through the transceiver 801 and / or receive uplink communications from one or more devices on the radio interface through the transceiver 801. Similarly, the processing circuit 803 may control the network interface 808 to transmit communications to one or more devices through the network interface 808 and / or receive communications from one or more devices through the network interface. In addition, modules may be stored in the memory 805, and these modules may provide instructions so that when the instructions of the modules are executed by the processing circuit 803, the processing circuit 803 performs corresponding operations (e.g., operations discussed below with respect to example embodiments involving devices). According to some embodiments, the device 800 and / or its (one or more) elements / (one or more) functions may be embodied as one or more virtual devices and / or one or more virtual machines.

[0094] According to some other embodiments, the apparatus may be implemented without a transceiver. In such embodiments, transmissions to the wireless device may be initiated by the apparatus 800 such that the transmissions to the wireless device are provided by the apparatus including the transceiver (e.g., by a base station). According to embodiments in which the apparatus includes a transceiver, initiating the transmission may include transmitting through the transceiver.

[0095] Fig. 9 is a flow chart illustrating a computer-implemented method of a network management system according to some embodiments. Fig. 9 , a method of managing a wireless communication network includes: obtaining (block 902) data regarding performance of a cell of the wireless communication network; and generating (block 904) a prediction of values ​​of a plurality of KPIs related to UL throughput for the cell of the wireless communication network based on the obtained data. The method generates (block 906) a prediction of a UL throughput that the cell will experience degraded at a future time based on the predicted values ​​of the KPIs; and determining (block 908) a set of candidate root cause KPIs from among the plurality of KPIs. The method selects (block 910) an action based on the determined set of candidate KPIs; and applies (block 912) the action to the wireless communication network.

[0096] In some embodiments, said data regarding the performance of said cell comprises performance measurement data and / or configuration management data.

[0097] In some embodiments, predicting the value of the KPI comprises generating features for an ML model based on the data regarding the performance of the cell.

[0098] In some embodiments, generating the feature includes generating a moving average of the KPI over one or more past time periods. Generating the feature may include generating a lagged value of the KPI from past time periods.

[0099] The method may further include generating a threshold value for the KPI based on historical values ​​of the KPI, wherein generating the prediction that the cell will experience degraded uplink throughput is based on a comparison of the predicted value of the KPI with the threshold value for the KPI. The threshold value for the KPI may be obtained by performing exploratory data analysis on historical data from the wireless communication network.

[0100] In some embodiments, the threshold value of the KPI may be obtained based on an operating frequency band and a bandwidth of a wireless communication system.

[0101] The method may also include: classifying the KPIs into multiple KPI categories; selecting the ML model from among multiple ML models based on the classification of at least one of the KPIs; and applying the ML model to generate the prediction of the value of the at least one of the KPIs.

[0102] In some embodiments, the method may further include: selecting a plurality of ML models based on the classification of the KPI; and applying the plurality of ML models.

[0103] The KPIs may be categorized as being related to availability, accessibility, retainability, mobility, coverage, quality, utilization, transmission ratio, and / or accessibility of the cell.

[0104] The ML model may include a plurality of ML models, and the prediction that the cell will experience degraded UL throughput may be based on outputs of the plurality of models. In particular, the prediction that the cell will experience degraded UL throughput may be based on a weighted average of the outputs of the plurality of models.

[0105] The plurality of ML models may include one or more of an XGBoost model, a CatBoost model, a random forest model, a long short-term memory model, and a light gradient boosting model.

[0106] In some embodiments, determining the set of candidate root cause KPIs comprises: generating a ranking of the plurality of KPIs based on importance of the predicted UL throughput degradation; and selecting the set of candidate root cause KPIs based on the ranking of the KPIs by importance.

[0107] The ranking of the plurality of KPIs may be generated by applying a Tree Shapley Additional Interpretation (Tree SHAP) algorithm to the KPIs and UL throughput degradation prediction.

[0108] The method may further include classifying each candidate root cause KPI in the set of candidate root cause KPIs into one of a plurality of KPI categories, and selecting the activity based on the KPI category of the set of candidate root cause KPIs.

[0109] Selecting the activity may include determining whether at least one of the set of candidate root cause KPIs is classified according to a first KPI category. Upon determining that at least one of the set of candidate root cause KPIs is classified according to the first KPI category, the method may examine operating conditions of the wireless communication network associated with the first KPI category and select the activity based on the operating conditions of the wireless communication network associated with the first KPI category.

[0110] The method may also include repeating the following steps for multiple KPI categories: determining whether at least one of the candidate root cause KPI set is classified according to a KPI category, checking the operating conditions of the wireless communication network associated with the KPI category, and selecting the activity based on the operating conditions of the wireless communication network associated with the KPI category.

[0101] Some embodiments provide a network management system (800) comprising a processor (803) and a memory (805) coupled to the processor, wherein the memory comprises computer program instructions, which when executed by the processor cause a network control device to execute Fig. 9 The operations shown in the figure.

[0102] Some embodiments provide a computer program comprising computer code to be executed by a network management system (800), the network management system (800) being configured to execute Fig. 9 The operations shown in the figure.

[0103] Some embodiments provide a computer program product comprising a non-transitory storage medium (805) comprising program code to be executed by a processing circuit (803) of a network management system (800), whereby execution of the program code causes an apparatus to perform Fig. 9 The operations shown in the figure.

[0114] Fig.10 An example of a communication system 900 is shown in accordance with some embodiments.

[0115] In an example, the communication system 1000 includes a telecommunications network 1002, which includes an access network 1004 such as a RAN and a core network 1006, which includes one or more core network nodes 1008. The access network 1004 includes one or more access network nodes, such as network nodes 1010a and 1010b (one or more of which may be generally referred to as network nodes 1010), or any other similar third generation partnership project (3GPP) access nodes or non-3GPP access points. The network node 1010 facilitates direct or indirect connection of user equipment (UE), such as by connecting UE 1012a, 1012b, 1012c, and 1012d (one or more of which may be generally referred to as UE 1012) to the core network 1006 via one or more wireless connections.

[0116] Exemplary wireless communications over wireless connections include transmitting and / or receiving wireless signals using electromagnetic waves, radio waves, infrared waves, and / or other types of signals suitable for conveying information without the use of wires, cables, or other material conductors. In addition, in different embodiments, the communication system 1000 may include any number of wired or wireless networks, network nodes, UEs, and / or any other components or systems that can facilitate or participate in the transfer of data and / or signals, whether via a wired connection or a wireless connection. The communication system 1000 may include any type of communication, telecommunication, data, cellular, radio network, and / or other similar types of systems and / or be connected to any type of communication, telecommunication, data, cellular, radio network, and / or other similar types of systems via an interface.

[0117] UE 1012 may be any of a wide variety of communication devices, including wireless devices arranged, configured and / or operable to wirelessly communicate with network node 1010 and other communication devices. Similarly, network node 1010 is arranged, capable, configured and / or operable to communicate directly or indirectly with UE 1012 and / or with other network nodes or devices in telecommunication network 1002 to enable and / or provide network access (such as wireless network access) and / or perform other functions (such as management in telecommunication network 1002).

[0118] In the depicted example, the core network 1006 connects the network node 1010 to one or more hosts (such as the host 1016). These connections may be direct or indirect via one or more intermediate networks or devices. In other examples, the network node may be directly coupled to the host. The core network 1006 includes one or more core network nodes (e.g., core network node 1008) constructed by hardware and software components. The features of these components may be substantially similar to those described with respect to the UE, network node, and / or host, so that their description is generally applicable to the corresponding components of the core network node 1008. The example core network node includes a mobile switching center (MSC), a mobility management entity (MME), a home subscriber server (HSS), an access and mobility management function (AMF), a session management function (SMF), an authentication server function (AUSF), a subscription identifier de-hiding function (SIDF), a unified data management (UDM), a security edge protection agent (SEPP), a network open function (NEF), and / or a user plane function (UPF) One or more functions.

[0119] The host 1016 may be under the ownership or control of a service provider other than the operator or provider of the telecommunications network 1002 and / or the access network 1004, and may be operated by or on behalf of the service provider. The host 1016 may host various applications to provide one or more services. Examples of such applications include real-time and pre-recorded audio / video content, data collection services (such as retrieving and compiling data about various environmental conditions detected by multiple UEs), analysis functionality, social media, functions for controlling or otherwise interacting with remote devices, functions for alarm and monitoring centers, or any other such functions performed by a server.

[0120] on the whole, Fig.10 The communication system 1000 enables connectivity between UEs, network nodes, hosts, and devices. In that sense, the communication system can be configured to operate according to predefined rules or procedures, such as specific standards, including but not limited to: Global System for Mobile Communications (GSM); Universal Mobile Telecommunications System (UMTS); Long Term Evolution (LTE), and / or other suitable 2G, 3G, 4G, 5G standards, or any applicable future generation standards (e.g., 6G); Wireless Local Area Network (WLAN) standards, such as the Institute of Electrical and Electronics Engineers (IEEE) 802.11 standard (WiFi); and / or any other suitable wireless communication standards, such as Worldwide Interoperability for Microwave Access (WiMax), Bluetooth, Z-wave, Near Field Communication (NFC), ZigBee, LiFi, and / or any Low Power Wide Area Network (LPWAN) standards such as LoRa and Sigfox.

[0121] In some examples, the telecommunication network 1002 is a cellular network implementing 3GPP standardized features. Therefore, the telecommunication network 1002 can support network slicing to provide different logical networks to different devices connected to the telecommunication network 1002. For example, the telecommunication network 1002 can provide ultra-reliable low-latency communication (URLLC) services to some UEs, while providing enhanced mobile broadband (eMBB) services to other UEs, and / or providing massive machine type communication (mMTC) / massive IoT services to yet other UEs.

[0122] In some examples, UE 1012 is configured to transmit and / or receive information without direct human interaction. For example, the UE may be designed to transmit information to access network 1004 on a predetermined schedule when triggered by an internal or external event or in response to a request from access network 1004. In addition, the UE may be configured to operate in single-RAT or multi-RAT or multi-standard mode. For example, the UE may operate with any one or a combination of Wi-Fi, NR (new air interface), and LTE, i.e., be configured for multi-radio dual connectivity (MR-DC), such as E-UTRAN (Evolved UMTS Terrestrial Radio Access Network) New Air Interface-Dual Connectivity (EN-DC).

[0123] In an example, the hub 1014 communicates with the access network 1004 to facilitate indirect communications between one or more UEs (eg, UEs 1012c and / or 1012d) and a network node (eg, network node 1010b).

[0124] Although the devices described herein may include a combination of hardware components shown, other examples may include computing devices with different combinations of components. It should be understood that these devices may include any suitable combination of hardware and / or software required to perform the tasks, features, functions, and methods disclosed herein. The determination, calculation, acquisition, or similar operations described herein may be performed by a processing circuit, which may process information by, for example, the following operations: converting the obtained information into other information, comparing the obtained information or the converted information with the information stored in the device, and / or performing one or more operations based on the obtained information or the converted information, and making a determination as a result of the processing. In addition, although the components are depicted as being located within a larger box or nested within multiple boxes, in fact, the device may include multiple different physical components constituting a single illustrated component, and functionality may be divided between the various components. For example, a communication interface may be configured to include any component described herein, and / or the functionality of the component may be divided between the processing circuit and the communication interface. In another example, the non-computationally intensive functions of any such component may be implemented in software or firmware, and the computationally intensive functions may be implemented in hardware.

[0125] In certain embodiments, some or all of the functions described herein may be provided by a processing circuit that executes instructions stored in a memory, which in certain embodiments may be a computer program product in the form of a non-transitory computer-readable storage medium. In alternative embodiments, some or all of the functionality may be provided by a processing circuit without executing instructions stored on a separate or discrete device-readable storage medium, such as in a hardwired manner. In any of those specific embodiments, the processing circuit may be configured to perform the described functionality, regardless of whether instructions stored on a non-transitory computer-readable storage medium are executed. The benefits provided by such functionality are not limited to separate processing circuits or other components of the device, but are enjoyed by the device as a whole and / or generally by end users and wireless networks.

[0126] Further definitions and embodiments are discussed below.

[0127] In the above description of various embodiments of the inventive concept, it is to be understood that the terms used herein are only for the purpose of describing specific embodiments and are not intended to be limitations of the inventive concept. Unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meanings as commonly understood by ordinary technicians in the field to which the inventive concept belongs. It will be further understood that terms such as those defined in common dictionaries should be interpreted as having meanings consistent with their meanings in the context of this specification and related fields, and unless explicitly defined as such in this article, the terms will not be interpreted in an idealized or overly formal sense.

[0128] When an element is referred to as being "connected" to, "coupled" to, "responsive to" or its variant to another element, it can be directly connected to, directly coupled to or directly respond to the other element or there can be an intermediate element. On the contrary, when an element is referred to as being "directly connected" to, "directly coupled" to, "directly responding to" or its variant to another element, there is no intermediate element. The same number refers to the same element throughout. In addition, "coupled", "connected", "responsive" or its variant as used in this article may include wireless coupling, connection or response. Unless the context clearly indicates otherwise, as used in this article, the singular form "a, an" and "the" are intended to also include plural forms. For the sake of brevity and / or clarity, well-known functions or structures may not be described in detail. The term "and / or" (abbreviated as " / ") includes any and all combinations of one or more of the associated listed items.

[0129] It will be understood that, although the terms first, second, third, etc. can be used to describe various elements / operations in this article, these elements / operations should not be limited by these terms. These terms are only used to distinguish one element / operation from another element / operation. Therefore, without departing from the teaching of the present invention, the first element / operation in some embodiments can be referred to as the second element / operation in other embodiments. Throughout the specification, the same figure mark or the same reference numeral represents the same or similar element.

[0130] As used herein, the terms "comprise, comprising, comprises," "include, including, includes," "have, has, having," or variations thereof, are open ended and include one or more stated features, integers, elements, steps, components, or functions, but do not exclude the presence or addition of one or more other features, integers, elements, steps, components, functions, or groups thereof. In addition, as used herein, the common abbreviation "eg," derived from the Latin phrase "exempligratia," may be used to introduce or specify one or more general examples of previously mentioned items and is not intended to be a limitation of such items. The common abbreviation "ie," derived from the Latin phrase "idest," may be used to specify a specific item from a more general narrative.

[0131] Example embodiments are described herein with reference to block diagrams and / or flowchart illustrations of computer-implemented methods, devices (systems and / or apparatuses), and / or computer program products. It is understood that the blocks of the block diagrams and / or flowchart illustrations and the combination of blocks in the block diagrams and / or flowchart illustrations can be implemented by computer program instructions executed by one or more computer circuits. These computer program instructions can be provided to processor circuits of general-purpose computer circuits, special-purpose computer circuits, and / or other programmable data processing circuits to produce a machine, so that instructions executed via processors of computers and / or other programmable data processing devices transform and control transistors, values ​​stored in memory locations, and other hardware components within such circuits to implement the functions / actions specified in one or more block diagrams and / or flowchart boxes, and thereby create components (functionality) and / or structures for implementing the functions / actions specified in (one or more) block diagrams and / or flowchart boxes.

[0132] These computer program instructions that can direct a computer or other programmable data processing device to operate in a specific manner may also be stored in a tangible computer-readable medium, so that the instructions stored in the computer-readable medium produce an article of manufacture including instructions for implementing the functions / actions specified in one or more block diagrams and / or flowchart blocks. Thus, embodiments of the inventive concept may be embodied in hardware and / or in software (including firmware, resident software, microcode, etc.) running on a processor such as a digital signal processor, which may be collectively referred to as "circuits," "modules," or variations thereof.

[0133] It should also be noted that, in some alternative implementations, the function / action annotated in the frame may not occur in the order annotated in the flow chart. For example, depending on the functionality / action involved, the actual two frames shown in succession may be performed substantially simultaneously or the frames may be performed in reverse order sometimes. In addition, the functionality of a given frame of a flow chart and / or block diagram may be divided into a plurality of frames and / or may at least partially integrate the functionality of two or more frames of a flow chart and / or block diagram. Finally, without departing from the scope of the inventive concept, other frames may be added / inserted between the frames shown, and / or frames / operations may be omitted. In addition, although some figures in the figure include arrows on the communication path to illustrate the main direction of communication, it is to be understood that communication may occur in the direction opposite to the arrows depicted.

[0134] In the case of not departing from the principle of the inventive concept substantially, many changes and modifications may be made to the embodiments. All such changes and modifications are intended to be included in the scope of the inventive concept herein. Therefore, the subject matter disclosed above is to be considered illustrative, rather than restrictive, and the examples of the embodiments are intended to cover all such modifications, enhancements and other embodiments that fall within the spirit and scope of the inventive concept. Therefore, to the maximum extent permitted by law, the scope of the inventive concept shall be determined by the most extensive permissible interpretation of the disclosure including the examples of the embodiments and their equivalents, and the scope of the inventive concept shall not be restricted or limited by the above detailed description.

Claims

1. A method for managing a wireless communication network, comprising: obtaining (902) data regarding performance of a cell of the wireless communication network; generating (904) a prediction of values ​​of a plurality of key performance indicators KPIs related to uplink UL throughput for the cell of the wireless communication network based on the obtained data; generating (906) a prediction of a degraded UL throughput that the cell will experience at a future time based on the predicted value of the KPI; determining (908) a set of candidate root cause KPIs associated with the predicted degraded UL throughput from among the plurality of KPIs; selecting (910) an activity based on the determined set of candidate KPIs; and The activity is applied (912) to the wireless communication network.

2. The method of claim 1, wherein: The data regarding the performance of the cell comprises performance measurement data and / or configuration management data.

3. The method of claim 1, wherein: Predicting the value of the KPI comprises generating features for a machine learning (ML) model based on the data regarding the performance of the cell.

4. The method of claim 3, wherein: Generating the feature includes generating a moving average of the KPI over one or more past time periods. The method of claim 3 , generating the feature comprises generating a lagged value of the KPI from a past time period.

6. The method of claim 3, further comprising: A threshold value for the KPI is generated based on historical values ​​of the KPI, wherein generating the prediction that the cell will experience degraded uplink throughput is based on a comparison of the predicted value of the KPI and the threshold value for the KPI.

7. The method of claim 6, wherein: The threshold of the KPI is obtained by performing exploratory data analysis on historical data from the wireless communication network.

8. The method of claim 6, wherein: The threshold value of the KPI is obtained based on an operating frequency band and a bandwidth of the wireless communication network.

9. The method of claim 3, further comprising: categorizing the KPI into a plurality of KPI categories; selecting the ML model from among a plurality of ML models based on the classification of at least one of the KPIs; as well as The ML model is applied to generate the prediction of the value of the at least one of the KPIs.

10. The method of claim 9, further comprising: selecting a plurality of ML models based on the classification of the KPI; as well as The plurality of ML models are applied.

11. The method of claim 10, wherein: The KPIs are categorized as being related to availability, accessibility, retainability, mobility, coverage, quality, utilization, transmission ratio and / or accessibility of the cell.

12. The method of claim 3, wherein: The ML model includes a plurality of ML models, and the prediction that the cell will experience degraded UL throughput is based on outputs of the plurality of models.

13. The method of claim 12, wherein: The prediction that the cell will experience degraded UL throughput is based on a weighted average of the outputs of the plurality of models.

14. The method of claim 10, wherein: The multiple ML models include one or more of an XGBoost model, a random forest model, a long short-term memory model, a CatBoost model, and a light gradient boosting model.

15. A method as claimed in any preceding claim, wherein: Determining the candidate root cause KPI set includes: generating a ranking of the plurality of KPIs based on the importance of the predicted UL throughput degradation; and Based on the ranking of KPIs by importance, the set of candidate root cause KPIs is selected.

16. The method of claim 15, wherein: The ranking of the plurality of KPIs is generated by applying a Tree Shapley Additional Interpretation (Tree SHAP) algorithm to the KPIs and UL throughput degradation prediction.

17. The method of any preceding claim, further comprising: classifying each candidate root cause KPI in the set of candidate root cause KPIs into one of a plurality of KPI categories; Wherein, the activity is selected based on the KPI category of the candidate root cause KPI set.

18. The method of claim 17, wherein: Selected activities include: determining whether at least one of the set of candidate root cause KPIs is classified according to a first KPI category; Upon determining that at least one of the set of candidate root cause KPIs is classified according to the first KPI category, examining an operating condition of the wireless communication network associated with the first KPI category; and The activity is selected based on the operating condition of the wireless communication network associated with the first KPI category.

19. The method of claim 18, further comprising repeating the following steps for multiple KPI categories: determining whether at least one of the candidate root cause KPI set is classified according to a KPI category, checking the operating conditions of the wireless communication network associated with the KPI category, and selecting the activity based on the operating conditions of the wireless communication network associated with the KPI category.

20. A network management system (800), comprising: Processor (803); as well as a memory (805) coupled to the processor; The memory comprises computer program instructions, which, when executed by the processor, cause the network management system to perform operations according to any one of claims 1 to 19.

21. A computer program comprising computer code to be executed by a network management system (800), the network management system being configured to perform the operations according to any one of claims 1 to 19.

22. A computer program product comprising a non-transitory storage medium (805), the non-transitory storage medium (805) comprising program code to be executed by a processing circuit (803) of a network management system (800), whereby execution of the program code causes the apparatus to perform operations according to any one of claims 1 to 19.

Images