Traffic suppression evaluation method and device and computer equipment

Abstract

The invention discloses a traffic depression assessment method and device and computer equipment. The method comprises the following steps: collecting data of a sample point of a target cell, the data of the sample point at least comprising one of the following: a physical resource block (PRB) utilization rate, uplink and downlink traffic, and an average number of activated users; at least adopting a pre-trained isolated forest model to clean abnormal sample points in the data of the sample points to obtain data of effective sample points; pre-judging the target cell according to the data of the effective sample points to obtain a first judgment result, and the first judgment result is used for representing a pre-judgment result of a traffic suppression state of the target cell; and fitting the traffic data of the target cell by adopting a piecewise linear regression function to obtain a fitting result, and correcting the first judgment result by adopting the fitting result to obtain a second judgment result which is used for representing a final judgment result of the traffic depression state of the target cell.

Description

Technical Field

[0001] This application relates to the field of wireless communication technologies, and in particular, to a method, apparatus, and computer device for evaluating traffic depression. Background Art

[0002] With the continuous increase in 5G users and the growing 5G traffic, the traffic shows an unbalanced development trend. Due to the mobility and aggregation of user service behaviors in the mobile network, combined with the characteristics of the mobile network structure, there must be some cases in the mobile network where the traffic volume of some mobile network cellular cells exceeds the maximum traffic volume capacity. In hot cells, there is even a full-load state that lasts for several hours, that is, the traffic depression state. The shortage of available time-frequency resources for a single user directly leads to a decline in the user experience, resulting in a reduction in the rate of users in the cell. When users use mobile phones, they may encounter situations such as unable to open web pages, too long waiting time, video stuttering, and excessive game latency, which affect the user perception. Generally, there are two ways in the current network capacity evaluation technical solutions: one is to first calculate the theoretical maximum traffic volume value through wireless spectrum efficiency and device capability parameters, and then give the evaluation result in combination with the operation indicators of the cell; the other is to directly perform multi-dimensional analysis on the network operation indicators of the existing network cells, and directly judge whether there is a traffic depression situation by setting relevant thresholds, and then give a solution for capacity expansion according to the thresholds. The current two solutions have the following problems: the evaluation solutions are based on static thresholds given by existing parameters or empirical values, and the thresholds are unified values for the whole network. Due to the different user distributions and traffic characteristics between cells, it is difficult for a unified standard to take into account the uniqueness of each cell; and it is impossible to give a specific value for the traffic depression value of a single cellular cell, which lacks strong support for the refined formulation of the solution. Summary of the Invention

[0003] Embodiments of this application provide a method, apparatus, and computer device for evaluating traffic depression to at least solve the technical problem that the evaluation method in related technologies is relatively single and has low adaptability to different cells.

[0004] According to one aspect of the embodiments of the present application, a method for evaluating traffic depression is provided, including: collecting data of sample points in a target cell, where the data of the sample points includes at least one of the following: Physical Resource Block (PRB) utilization rate, uplink and downlink traffic, and average number of active users; cleaning abnormal sample points in the data of the sample points at least by using a pre-trained isolation forest model to obtain data of valid sample points; making a preliminary determination on the target cell according to the data of the valid sample points to obtain a first determination result, where the first determination result is used to represent the preliminary determination result of the traffic depression state of the target cell; fitting the traffic data of the target cell by using a piecewise linear regression function to obtain a fitting result, and using the fitting result to correct the first determination result to obtain a second determination result, where the second determination result is used to represent the final determination result of the traffic depression state of the target cell.

[0005] Optionally, cleaning abnormal sample points in the data of the sample points at least by using a pre-trained isolation forest model to obtain data of valid sample points includes: screening out sample points with invalid values of the average number of active users and the uplink and downlink traffic from the data of the sample points and determining them as sample points to be processed; zero-padding the average number of active users and the uplink and downlink traffic in the sample points to be processed to obtain data of initial sample points; obtaining a training data set, where the training data set contains data of multiple sample points in the target cell; generating multiple isolation trees based on the training data set, and forming an isolation forest with the multiple isolation trees. The isolation tree is used to represent the distribution of each sample point in the sample point subset in the target feature dimension. The sample point subset is a set of sample points randomly obtained from the training data set, and the target feature dimension is one of the Physical Resource Block (PRB) utilization rate, uplink and downlink traffic, and average number of active users; screening out abnormal sample points from the data of the initial sample points based on the isolation forest and removing the abnormal sample points from the initial sample points to obtain the data of the valid sample points.

[0006] Optionally, screening out abnormal sample points from the data of the initial sample points based on the isolation forest includes: sequentially traversing each sample point in the data of the initial sample points through each isolation tree in the isolation forest to obtain the average value of the path lengths of each sample point in each isolation tree in the isolation forest; obtaining the average path length of each isolation tree; determining the score of each sample point according to the average value of the path lengths of each sample point in each isolation tree in the isolation forest and the average path length of each isolation tree; and determining the sample points with scores higher than a preset threshold in the data of the initial sample points as the abnormal sample points.

[0007] Optionally, the method further includes: obtaining the data features of each sample point in the data of the valid sample points; determining the type of the target cell based on the data features corresponding to the data of the valid sample points; when the target cell is the first cell, clustering the data of the valid sample points in the target cell to obtain the data of the clustered sample points; pre-determining the target cell based on the data of the clustered sample points to obtain a first determination result; when the target cell is the second cell, pre-determining the target cell according to the PRB utilization rate in the data of the valid sample points, where the distribution trends of the sample points in the first cell and the second cell are different.

[0008] Optionally, pre-determining the target cell according to the PRB utilization rate in the data of the valid sample points includes: obtaining the per-second PRB utilization rate of all valid sample points within a preset period; determining the valid sample points with the proportion of the duration in which the per-second PRB utilization rate is not less than a first preset ratio in the preset period being greater than a second preset ratio as severely suppressed sample points; determining the valid sample points with the proportion of the duration in which the per-second PRB utilization rate is not less than a first preset ratio in the preset period being less than a third preset ratio as non-suppressed sample points; determining the other valid sample points among all valid sample points as target sample points; counting the severely suppressed sample points and the non-suppressed sample points in the target cell to obtain a statistical result, and determining the first determination result of the target cell based on the statistical result.

[0009] Optionally, determining the first determination result of the target cell based on the statistical result includes: dividing the valid sample points into multiple bins based on the average number of active users corresponding to each valid sample point, where the average number of active users corresponding to the valid sample points within each bin is the same; obtaining the proportion of the severely suppressed sample points in all valid sample points within all bins, and determining the bins with the proportion of the severely suppressed sample points in all valid sample points within the bin being greater than a first threshold as the first bins, determining the bins with the proportion of the non-suppressed sample points in all valid sample points within the bin being greater than a second threshold as the second bins, and determining the bins with the proportion of the severely suppressed sample points in all valid sample points within the bin being less than the first threshold as the third bins; when there is at least one first bin among the multiple bins, determining the first determination result as the target cell being a severely suppressed cell; when all the multiple bins are the second bins, determining the first determination result as the target cell being a non-suppressed cell; when there is no first bin among the multiple bins and there is at least one third bin, determining the target cell as a mildly suppressed cell.

[0010] Optionally, a piecewise linear regression function is used to fit the traffic data of the target cell, and a fitting result is obtained, including: when the first determination result of the target cell indicates that the target cell is a severely depressed cell, determining the piecewise linear regression function as a three-segment regression function; when the first determination result of the target cell indicates that the target cell is a mildly depressed cell, determining the piecewise linear regression function as a two-segment regression function; when the first determination result of the target cell indicates that the target cell is a non-depressed cell, determining the piecewise linear regression function as a one-segment regression function; respectively fitting the traffic data of the target cell based on different piecewise linear regression functions to obtain a fitting result.

[0011] Optionally, the first determination result is corrected by using the fitting result to obtain a second determination result, including: when the target cell is the severely depressed cell, dividing the traffic data of the target cell into three sections by using an initial segmentation point, where the three sections include: a first section, a second section, and a third section; when the slope of the fitting curve corresponding to the traffic data in the third section is greater than zero, correcting the target cell to the mildly depressed cell; when the target cell is the mildly depressed cell, obtaining whether the slope of the fitting curve corresponding to the traffic data in the second section is continuously decreasing, and when the slope of the fitting curve corresponding to the traffic data in the second section does not decrease, correcting the target cell to the non-depressed cell.

[0012] Optionally, the method further includes: obtaining the slope of the fitting curve corresponding to the traffic data in the first section, and determining a linear fitting function based on the slope of the fitting curve corresponding to the traffic data in the first section; obtaining the average number of active users corresponding to the traffic data, and determining the required traffic based on the average number of active users corresponding to the traffic data and the linear fitting function; determining the difference between the required traffic and the actual traffic as the traffic depression value.

[0013] According to another aspect of the embodiments of the present application, there is also provided an evaluation device for traffic depression, including: a collection module, configured to collect data of sample points of a target cell, where the data of the sample points includes at least one of the following: physical resource block (PRB) utilization rate, uplink and downlink traffic, average number of active users; a cleaning module, configured to clean abnormal sample points in the data of the sample points by at least using a pre-trained isolation forest model to obtain data of valid sample points; a determination module, configured to pre-determine the target cell according to the data of the valid sample points to obtain a first determination result, where the first determination result is used to represent the pre-determination result of the traffic depression state of the target cell; a correction module, configured to fit the traffic data of the target cell by using a piecewise linear regression function to obtain a fitting result, and use the fitting result to correct the first determination result to obtain a second determination result, where the second determination result is used to represent the final determination result of the traffic depression state of the target cell.

[0014] According to yet another aspect of the embodiments of the present application, there is also provided a computer device, including: a memory and a processor, where the memory is used to store program instructions; the processor is connected to the memory and is configured to execute the above-mentioned evaluation method for traffic depression.

[0015] According to still another aspect of the embodiments of the present application, there is also provided a non-volatile storage medium, which includes a stored computer program, where the device where the non-volatile storage medium is located executes the above-mentioned evaluation method for traffic depression by running the computer program.

[0016] According to still another aspect of the embodiments of the present application, there is also provided a computer program product, including computer instructions, where the computer instructions implement the above-mentioned evaluation method for traffic depression when executed by a processor.

[0017] In an embodiment of the present application, data of sample points in a target cell is collected, and the data of the sample points includes at least one of the following: Physical Resource Block (PRB) utilization rate, uplink and downlink traffic, average number of active users; at least an isolated forest model trained in advance is used to clean abnormal sample points in the data of the sample points to obtain data of valid sample points; a pre-determination is made on the target cell according to the data of the valid sample points to obtain a first determination result, and the first determination result is used to represent a pre-determination result of the traffic depression state of the target cell; a piecewise linear regression function is used to fit the traffic data of the target cell to obtain a fitting result, and the first determination result is corrected by using the fitting result to obtain a second determination result, and the second determination result is used to represent a final determination result of the traffic depression state of the target cell, thereby achieving the purpose of cleaning the traffic data of the target cell by using a machine learning method and making two determinations based on the data of the cleaned valid sample points, thereby realizing the technical effect of accurately evaluating according to the actual traffic data of different cells, and further solving the technical problem that the evaluation method in the related art is relatively single and has low adaptability to different cells. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings described herein are used to provide a further understanding of the present application and form a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation to the present application. In the drawings:

[0019] Figure 1 is a hardware structure block diagram of a computer terminal for implementing an evaluation method for traffic depression according to an embodiment of the present application;

[0020] Figure 2 is a flowchart of an evaluation method for traffic depression according to an embodiment of the present application;

[0021] Figure 3 is a schematic diagram of the distribution of sample points after clustering according to an embodiment of the present application;

[0022] Figure 4 is a schematic diagram of a clustering control interface according to an embodiment of the present application;

[0023] Figure 5 is a schematic diagram of sample point classification according to an embodiment of the present application;

[0024] Figure 6 is a schematic diagram of a fitting curve according to an embodiment of the present application;

[0025] Figure 7 is another schematic diagram of a fitting curve according to an embodiment of the present application;

[0026] Figure 8It is a flowchart of another method for evaluating traffic depression according to an embodiment of the present application;

[0027] Figure 9 It is a structural diagram of an apparatus for evaluating traffic depression according to an embodiment of the present application. Detailed implementation manners

[0028] To enable those skilled in the art to better understand the solutions of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0029] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily need to be used to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present application described here can be implemented in an order other than those illustrated or described here. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily have to be limited to those clearly listed steps or units, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0030] The information collected in the embodiments of the present application is information and data authorized by the user or fully authorized by all parties, and the processing of the relevant data such as collection, storage, use, processing, transmission, provision, disclosure, and application complies with the relevant laws, regulations, and standards of the relevant regions, takes necessary confidentiality measures, does not violate public order and good customs, and provides a corresponding operation entry for the user to choose to authorize or reject the automated decision result; if the user chooses to reject, the expert decision-making process will be entered.

[0031] To solve the problems existing in the related art, an embodiment of the present application provides a method for evaluating traffic depression, and this method can run on Figure 1 the computer terminal shown below, and the following is an explanation of this computer terminal.

[0032] The embodiment of the method for evaluating traffic depression provided by the embodiment of the present application can be executed on a mobile terminal, a computer terminal, or a similar computing device. Figure 1 The following shows a hardware structure block diagram of a computer terminal for implementing the method for evaluating traffic depression. As Figure 1As shown, the computer terminal 10 may include one or more processors (illustrated as 102a, 102b, ……, 102n in the figure) (the processor may include, but is not limited to, a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 104 for storing data, and a transmission module 106 for communication functions connected through wired and / or wireless networks. In addition, it may further include: a display, a keyboard, a cursor control device, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the I / O interface), a network interface, and a BUS bus. Those of ordinary skill in the art can understand that Figure 1 the structure shown is only schematic and does not limit the structure of the above-mentioned electronic device. For example, the computer terminal 10 may further include more or fewer components than Figure 1 shown in, or have a different configuration from Figure 1 that shown.

[0033] It should be noted that the above one or more processors and / or other data processing circuits may generally be referred to as "data processing circuits" herein. The data processing circuit may be embodied in software, hardware, firmware, or any combination thereof, in whole or in part. In addition, the data processing circuit may be a single independent processing module, or be incorporated in whole or in part into any one of other elements in the computer terminal 10. As involved in the embodiments of the present application, the data processing circuit is a kind of processor control (such as the selection of a variable resistance terminal path connected to an interface).

[0034] The memory 104 may be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the traffic suppression evaluation method in the embodiments of the present application. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, implements the above-mentioned traffic suppression evaluation method. The memory 104 may include a high-speed random access memory, and may further include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memories. In some instances, the memory 104 may further include a memory remotely set relative to the processor, and these remote memories may be connected to the computer terminal 10 through a network. Examples of the above network include, but are not limited to, the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0035] The transmission module 106 is used to receive or send data via a network. Specific examples of the above-mentioned network may include a wireless network provided by the communication provider of the computer terminal 10. In one example, the transmission module 106 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission module 106 can be a Radio Frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0036] The display can be, for example, a touch-screen liquid crystal display (LCD), which enables the user to interact with the user interface of the computer terminal 10.

[0037] It should be noted here that in some alternative embodiments, the above Figure 1 illustrated computer terminal may include hardware elements (including circuits), software elements (including computer code stored on a computer-readable medium), or a combination of both hardware elements and software elements. It should be pointed out that Figure 1 is only an example of a specific specific instance and is intended to illustrate the types of components that may exist in the above computer terminal.

[0038] Under the above operating environment, an embodiment of a method for evaluating traffic depression is provided in an embodiment of the present application. It should be noted that the steps illustrated in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although the logical order is illustrated in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0039] Figure 2 is a flowchart of a method for evaluating traffic depression according to an embodiment of the present application. As Figure 2 shown, the method includes the following steps:

[0040] Step S202, collecting data of sample points of the target cell, where the data of the sample points includes at least one of the following: Physical Resource Block (PRB) utilization rate, uplink and downlink traffic, average number of active users;

[0041] In step S202, the sample point is the network status operating state within a preset period of the target cell, and is associated with multiple metric data, such as: PRB utilization rate per hour, PRB utilization rate per second, uplink and downlink traffic, and average number of active users.

[0042] In an actual application scenario, the network management data of the target cell at a granularity of every hour or every 15 minutes within 1 month can be collected through a professional network management system (the network management data at each granularity can be used as a sample point), including: traffic / user count / resource utilization rate / single-user perceived rate, etc., and data import is performed.

[0043] Step S204, at least use a pre-trained isolation forest model to clean the abnormal sample points in the data of the sample points to obtain the data of valid sample points;

[0044] In step S204, the sample points with an average active UE user count and uplink and downlink traffic of 0 can be screened out first, and the missing values are filled with zeros, and then by introducing the isolation forest algorithm, the abnormal value sample points with low density and far from the valid data in the massive original sampling point data are removed to reduce the interference of abnormal data.

[0045] Step S206, perform a preliminary determination on the target cell according to the data of the valid sample points to obtain a first determination result, and the first determination result is used to represent the preliminary determination result of the traffic depression state of the target cell;

[0046] In step S206, before performing a preliminary determination on the target cell according to the data of the valid sample points, it is also necessary to distinguish the type of the target cell and select different determination methods according to different cell types.

[0047] Step S208, use a piecewise linear regression function to fit the traffic data of the target cell to obtain a fitting result, and use the fitting result to correct the first determination result to obtain a second determination result, and the second determination result is used to represent the final determination result of the traffic depression state of the target cell.

[0048] Through the above steps S202 to S208, data of sample points of the target cell is collected. The data of the sample points includes at least one of the following: Physical Resource Block (PRB) utilization rate, uplink and downlink traffic, and average number of active users. At least an isolated forest model pre-trained is used to clean abnormal sample points in the data of the sample points, and data of valid sample points is obtained. Based on the data of the valid sample points, a preliminary determination of the target cell is made to obtain a first determination result, where the first determination result is used to represent the preliminary determination result of the traffic depression state of the target cell. A piecewise linear regression function is used to fit the traffic data of the target cell to obtain a fitting result, and the first determination result is corrected using the fitting result to obtain a second determination result, where the second determination result is used to represent the final determination result of the traffic depression state of the target cell. Thus, the purpose of cleaning the traffic data of the target cell by means of machine learning and making two determinations based on the data of the cleaned valid sample points is achieved, thereby realizing the technical effect of accurately evaluating according to the actual traffic data of different cells, and further solving the technical problem that the evaluation method in the related technology is relatively single and has low adaptability to different cells. The following is a detailed description.

[0049] In some embodiments of the present application, the specific steps of at least using an isolated forest model pre-trained to clean abnormal sample points in the data of the sample points and obtaining data of valid sample points are as follows: Sample points with invalid values for the average number of active users and the uplink and downlink traffic are screened out from the data of the sample points and determined as sample points to be processed; The average number of active users and the uplink and downlink traffic in the sample points to be processed are filled with zeros to obtain data of initial sample points; A training data set is obtained, and the training data set contains data of multiple sample points in the target cell; Based on the training data set, multiple isolated trees are generated, and the multiple isolated trees are formed into an isolated forest. The isolated tree is used to represent the distribution of each sample point in the sample point subset in the target feature dimension. The sample point subset is a set of sample points randomly obtained from the training data set, and the target feature dimension is one of the Physical Resource Block (PRB) utilization rate, uplink and downlink traffic, and average number of active users; Based on the isolated forest, abnormal sample points are screened out from the data of the initial sample points, and the abnormal sample points are removed from the initial sample points to obtain the data of the valid sample points.

[0050] Among them, the specific steps for screening out abnormal sample points from the data of the initial sample points based on the isolation forest are as follows: Traverse each sample point in the data of the initial sample points through each isolation tree in the isolation forest in turn, and obtain the average value of the path lengths of each sample point in each isolation tree in the isolation forest; Obtain the average path length of each isolation tree; Determine the score of each sample point according to the average value of the path lengths of each sample point in each isolation tree in the isolation forest and the average path length of each isolation tree; Determine the sample points with scores higher than the preset threshold in the data of the initial sample points as the abnormal sample points.

[0051] Specifically, first screen out the sample points with the average number of active users and the uplink and downlink traffic being 0 (invalid values) in the sample points and fill the missing values with zero, and then introduce the isolation forest algorithm to remove the abnormal value sample points with low density and far from the valid data in the massive original sampling point data, reducing the interference of abnormal data.

[0052] Among them, filling missing values with zero: Screen out the sample points with the average number of active UE users and the uplink and downlink traffic being 0 (invalid values) and fill the missing values with zero;

[0053] Taking the data of three dimensions, namely the average number of active users in the cell, the number of downlink service bytes (downlink traffic) of PDCP (Packet Data Convergence Protocol), and the PRB (PRB utilization rate) used by DTCH (Dedicated Traffic Channel) as an example, the IsolationForest (isolation forest) algorithm of machine learning is used to remove the abnormal value sample points with low density and far from the valid data, reducing the interference of abnormal data. The operation process of this algorithm is as follows:

[0054] Step 1) Randomly select n point sample points from the training data as a sample subset and put them into the root node of the tree.

[0055] Step 2) Randomly specify one dimension (feature) among the three dimensions of the average number of active users in the cell, the number of downlink service bytes (MByte) of PDCP, and the PRB (%) used by DTCH, and randomly generate a cut point p in the data of the current node in the isolation tree (the cut point is generated between the maximum value and the minimum value of the specified dimension in the data of the current node).

[0056] Step 3) Generate a hyperplane with the cut point p, and then divide the current node data space into 2 subspaces: Place the data less than p in the specified dimension in the left child node of the current node, and place the data greater than or equal to p in the right child node of the current node.

[0057] Step 4) recursively repeat steps 2) and 3) in the child node, and continuously construct new child nodes until there is only one data in the child node (no further cutting is possible) or the child node has reached a limited height, thus generating an isolated tree.

[0058] Step 5) loops through step 1) to step 4) until t isolated trees iTree are generated.

[0059] Step 6) After obtaining t isolated trees iTree, the isolation forest training is completed, and then the generated isolation forest is used to evaluate the test data. For each sample point xi, it is traversed through each isolated tree (iTree), and the average height of point xi in the forest is calculated h(xi), and the average height of all points is normalized. The average path length of the isolation tree can be determined by the following formula:

[0060]

[0061] H(i) is the harmonic number, which can be approximated as ln(i)+0.5772156649, i=1.2…n, where n is the number of samples. For a given data set size n, the expectation of the average path length is a constant. This formula provides a standardized benchmark for standardizing the path length.

[0062] The calculation formula for the sample point score is as follows:

[0063]

[0064] Among them, E(h(x)) is the average path length of data point x in all trees, n is the sample size of training data, and c(n) is the average path length of the tree, which is used for standardization. The outlier value range is [0,1]. The closer to 1, the easier it is to isolate the sample and the more likely it is to be abnormal.

[0065] Based on the calculated anomaly score, a threshold is set to determine whether the data point is abnormal and to remove abnormal sampling points.

[0066] In actual application scenarios, there may be significant differences in user behavior in a cell at different time periods, resulting in different characteristics of sample points or the existence of some special sample points that do not conform to the traffic depression model. Such cells need to be identified and processed separately, and traffic depression determination cannot be directly carried out. Therefore, it is necessary to determine the type of the target cell, and the specific differentiation process is as follows: Obtain the data characteristics of each sample point in the data of the valid sample points; Determine the type of the target cell based on the data characteristics corresponding to the data of the valid sample points; When the target cell is the first cell, cluster the data of the valid sample points in the target cell to obtain the data of the clustered sample points; Based on the data of the clustered sample points, pre-determine the target cell to obtain a first determination result; When the target cell is the second cell, pre-determine the target cell according to the PRB utilization rate in the data of the valid sample points, where the distribution trends of the sample points in the first cell and the second cell are different.

[0067] Among them, the sample points of the first cell (special cell) show characteristics such as a small number of large-traffic users occupying large bandwidths, low correlation between indicators such as the number of users and traffic, inconsistent distribution trends of different sample points, and non-stationarity of traffic in the time domain.

[0068] When the target cell is the first cell, the first cell contains ordinary sample points and special sample points, and the sample points need to be clustered separately. After clustering, the ordinary clusters are judged for traffic depression according to ordinary cells, and only relevant statistics are output for the special clusters.

[0069] As Figure 3 shown, the blue data points marked as 0 are ordinary sample points, the cyan sample points marked as 1 are the sample points classified into the ordinary sample point cluster after cluster analysis, and the gray sample points marked as -1 are the sample points of the special sample point cluster.

[0070] Specifically, the clustering method is to use the binary hierarchical clustering method, perform clustering after normalizing data in multiple dimensions, and use the dip test (a statistical method for detecting whether the data distribution is a unimodal distribution) to control whether to continue hierarchical clustering until all classification dip tests show a single cluster.

[0071] Figure 4 Shows a schematic interface diagram of using the dip test.

[0072] In some embodiments of the present application, the specific steps for pre-determining the target cell according to the PRB utilization rate in the data of the effective sample points are as follows: Obtain the PRB utilization rate per second of all effective sample points within a preset period; Determine the effective sample points with the proportion of the duration in which the PRB utilization rate per second within the preset period is not less than a first preset ratio and accounts for more than a second preset ratio of the preset period as severely suppressed sample points; Determine the effective sample points with the proportion of the duration in which the PRB utilization rate per second within the preset period is not less than a first preset ratio and accounts for less than a third preset ratio of the preset period as non-suppressed sample points; Determine the other effective sample points among all effective sample points as target sample points; Count the severely suppressed sample points and the non-suppressed sample points in the target cell to obtain a statistical result, and determine the first determination result of the target cell based on the statistical result.

[0073] Specifically, taking the first preset ratio as 90% as an example, the second-level resource utilization rate distribution index is used to pre-determine the suppression of the cell. Classify all sample points of each cell (taking a 15-minute statistic as a sample point as an example) according to the utilization rate distribution, and then pre-determine the cell according to the classification ratio. For ordinary cells, all effective data is used for cell pre-classification determination. For special cells, only the samples determined as ordinary sample point clusters are used for determination. Output severely suppressed cells, mildly suppressed cells, and non-suppressed cells.

[0074] Based on the downlink PRB utilization rate, determine the suppression situation of the sample points.

[0075] 1) Severely suppressed sample points: Sample points with a second-level PRB utilization rate greater than or equal to 90% for more than the second preset ratio of seconds;

[0076] 2) Non-suppressed sample points: Sample points with a second-level PRB utilization rate greater than or equal to 90% for less than the third preset ratio of seconds;

[0077] 3) Possibly suppressed sample points (target sample points): The remaining sample points.

[0078] As Figure 5 shown, the blue sample points are non-suppressed sample points, and the purple sample points are severely suppressed sample points.

[0079] In some embodiments of the present application, the specific steps for determining the first determination result of the target cell based on the statistical result are as follows: Divide the valid sample points into multiple bins based on the average number of active users corresponding to each valid sample point, where the average number of active users corresponding to the valid sample points within each bin is the same; Obtain the proportion of severely suppressed sample points in all valid sample points within each bin, and determine the bin with the proportion of severely suppressed sample points in all valid sample points within the bin greater than the first threshold as the first bin, determine the bin with the proportion of non-suppressed sample points in all valid sample points within the bin greater than the second threshold as the second bin, and determine the bin with the proportion of severely suppressed sample points in all valid sample points within the bin less than the first threshold as the third bin; In the case where at least one of the first bins is included in the multiple bins, determine the first determination result as the target cell being a severely suppressed cell; In the case where all of the multiple bins are the second bins, determine the first determination result as the target cell being a non-suppressed cell; In the case where the first bin is not included in the multiple bins and at least one of the third bins is included, determine the target cell as a mildly suppressed cell.

[0080] For example: Statistically analyze the proportion of severely suppressed sample points and non-suppressed sample points in the cell, and use binning to eliminate the influence of uneven user number distribution.

[0081] 1) Severely suppressed cell: Use binning of the average number of users. In at least one bin, the proportion of severely suppressed sample points is higher than the first threshold, for example: 20%.

[0082] 2) Non-suppressed cell: In each bin, the proportion of non-suppressed sample points is higher than the second threshold, for example: 50%.

[0083] 3) Possibly suppressed cell: Other cells.

[0084] In some embodiments of the present application, the specific steps for fitting the traffic data of the target cell using a piecewise linear regression function to obtain the fitting result are as follows: In the case where the first determination result of the target cell indicates that the target cell is a severely suppressed cell, determine the piecewise linear regression function as a three-segment regression function; In the case where the first determination result of the target cell indicates that the target cell is a mildly suppressed cell, determine the piecewise linear regression function as a two-segment regression function; In the case where the first determination result of the target cell indicates that the target cell is a non-suppressed cell, determine the piecewise linear regression function as a one-segment regression function; Respectively fit the traffic data of the target cell based on different piecewise linear regression functions to obtain the fitting result.

[0085] In an embodiment of the present application, by introducing a piecewise linear regression algorithm, a step-by-step traversal fitting method for cells is implemented. According to the pre-determination result, fitting modeling is performed on each cell, and the pre-classification result is corrected to output the fitting curves of severely depressed, mildly depressed, and non-depressed cells, as well as the depressed traffic volume and depression rate during the full cycle and busy hours.

[0086] Traffic depression determination features, where the sample points with low PRB occupancy rate approximately grow freely and have an approximately linear increasing relationship with the number of users. The sample points with high PRB occupancy rate are severely depressed points and have an approximately linear relationship with the number of users. The fitting function is: straight line - curve - straight line, as Figure 5 shown.

[0087] It can be understood that the segmentation points for two-segment fitting and three-segment fitting need to be traversed for fitting. By stepwise setting the initial value and search range of the segmentation point, the segmentation point is traversed and searched. As Figure 6 shown, the fitting curve includes three sections: a non-depressed section, a mildly depressed section, and a severely depressed section. In the non-depressed section interval, the traffic volume increases as the number of users increases (free increase).

[0088] The operation process of the piecewise linear regression algorithm is as follows:

[0089] Parameter definition:

[0090] Define k0, k, and k1 as the derivatives in the non-depressed stage, mildly depressed section, and severely depressed section respectively.

[0091] Define x0 and xMC as the starting point and ending point of the mildly depressed section respectively.

[0092] Train the piecewise linear regression model to obtain the fitting curve function:

[0093]

[0094] In the formula, N represents the number of sample points, y i represents the service traffic volume of the i-th sample point, x i represents the number of users of the i-th sample point, and a and b are the intercept and slope of the fitting curve function respectively.

[0095] Derive to obtain the parameters:

[0096]

[0097] After sorting, the system of equations is obtained:

[0098]

[0099] Solving the above system of equations can obtain the best estimated values of the parameters a and b in the linear function y = a + kx.

[0100]

[0101] In the formula, represents the best estimate of a, represents the best estimate of b.

[0102] For the slope value k of the linear fitting function, starting from the abscissa value where the value of X is 0, increasing by 1 each time until the value of X ends, the slope values from k0 to k1 are obtained. When the calculated slope k is less than k0, the point is obtained as x0, and at this time the derivative is k0. When the calculated slope k is equal to 0, the point is obtained as xMC. The threshold k0 is an empirical value obtained by averaging the curves of traffic-suppressed cells in the existing network and will be iteratively updated later.

[0103]

[0104] In some embodiments of the present application, the specific process of using the fitting result to correct the first determination result to obtain the second determination result is as follows: when the target cell is the severely suppressed cell, the traffic data of the target cell is divided into three sections by using the initial segmentation point. The three sections include: the first section, the second section, and the third section; when the slope of the fitting curve corresponding to the traffic data in the third section is greater than zero, the target cell is corrected to the mildly suppressed cell; when the target cell is the mildly suppressed cell, it is obtained whether the slope of the fitting curve corresponding to the traffic data in the second section is continuously decreasing. When the slope of the fitting curve corresponding to the traffic data in the second section does not decrease, the target cell is corrected to the non-suppressed cell.

[0105] For example: For a cell pre-determined as a severely suppressed cell, a three-segment curve model is used for fitting. It is judged by whether the slope value k is less than 0. For a cell whose third segment of the fitted curve does not conform to severe suppression (the third segment is still rising), it is corrected to a mildly suppressed cell and a two-segment curve fitting is performed; for a cell pre-determined as a mildly suppressed cell or a cell whose suppression determination is corrected to a mildly suppressed cell, a two-segment curve fitting is performed. It is judged by whether the slope value k is continuously decreasing. For a cell that does not conform to the characteristics of mild suppression (the maximum suppression rate is too small), it is corrected to a non-suppressed cell and a one-segment curve fitting is performed; for a cell pre-determined as a non-suppressed cell or a cell corrected to a non-suppressed cell, a one-segment curve (straight line) fitting is performed, and at this time the slope is k0.

[0106] The value of traffic depression can be determined in the following way: Obtain the slope of the fitting curve corresponding to the traffic data of the first section, and determine a linear fitting function based on the slope of the fitting curve corresponding to the traffic data of the first section; Obtain the average number of activated users corresponding to the traffic data, and determine the required traffic based on the average number of activated users corresponding to the traffic data and the linear fitting function; Determine the difference between the required traffic and the actual traffic as the value of traffic depression.

[0107] Specifically, a linear function is obtained through the slope k0 value, and then the average number of activated UE users is substituted to estimate the required traffic. The final value of traffic depression is calculated through the required traffic and the actual traffic value. Finally, the fitting curves and corresponding statistics of severely depressed, mildly depressed, and non-depressed cells are output.

[0108] Figure 7 Another fitting curve is shown, as Figure 7 shown, the depression amount can be the difference between the required traffic and the actual traffic corresponding to the sample points.

[0109] Figure 8 Another evaluation method for traffic depression is also shown, as Figure 8 shown, including:

[0110] Step 1: Data collection. Through the professional network management of the existing mobile network, extract the original granularity indicators at the mobile cellular cell level of the existing network. The indicator requirements include hourly PRB utilization rate, second-level PRB utilization rate, uplink and downlink traffic, and average number of activated UE users.

[0111] Step 2: Data cleaning. First, screen out the sample points with the average number of activated UE users and uplink and downlink traffic being 0 and fill the missing values with zero. Then, by introducing the Isolation Forest algorithm, remove the outlier sample points with low density and far from the valid data in the massive original sampling point data to reduce the interference of abnormal data.

[0112] Step 3: Judgment of data depression characteristics. The user behaviors of some cells may vary greatly at different time periods, resulting in different characteristics of the sample points, or there may be some special sample points that do not conform to the traffic depression model. Such cells need to be identified and processed separately and cannot be directly judged for traffic depression. Screen out ordinary cells and special cells.

[0113] Step 4: Processing of special cells. Special cells may contain ordinary sample points and special sample points. The sample points need to be clustered separately. After clustering, the ordinary clusters are judged for traffic depression according to ordinary cells, and only the relevant statistics are output for the special clusters.

[0114] Step 5: Preliminary determination of cell depression. The depression determination introduces a second-level resource utilization rate distribution index to preliminarily determine cell depression. All sample points of each cell (one sample point is counted every 15 minutes) are classified according to the utilization rate distribution, and then the cell is preliminarily determined according to the classification ratio. For ordinary cells, all valid data are used for cell pre-classification determination. For special cells, only the samples determined as ordinary sample point clusters are used for determination. Output severely depressed cells, mildly depressed cells, and non-depressed cells.

[0115] Step 6: Cell depression determination. By introducing a piecewise linear regression algorithm, the method realizes a step-by-step traversal fitting method for each cell, fits and models each cell according to the preliminary determination result, and corrects the pre-classification result, and outputs the fitting curves of severely depressed, mildly depressed, and non-depressed cells, as well as the depression traffic and depression rate during the full cycle and busy hours.

[0116] Step 7: Application of depression data. Through the potential demand traffic, evaluate the cell load and estimate the increased traffic and revenue after capacity expansion according to the evaluation result of the depression traffic.

[0117] Figure 9 is an evaluation device for traffic depression according to an embodiment of the present application, as Figure 9 shown. The device includes:

[0118] A collection module 90, configured to collect data of sample points of a target cell, where the data of the sample points includes at least one of the following: physical resource block (PRB) utilization rate, uplink and downlink traffic, and average number of active users;

[0119] A cleaning module 92, configured to clean abnormal sample points in the data of the sample points by using at least a pre-trained isolation forest model to obtain data of valid sample points;

[0120] A determination module 94, configured to preliminarily determine the target cell according to the data of the valid sample points to obtain a first determination result, where the first determination result is used to represent the preliminary determination result of the traffic depression state of the target cell;

[0121] A correction module 96, configured to fit the traffic data of the target cell by using a piecewise linear regression function to obtain a fitting result, and correct the first determination result by using the fitting result to obtain a second determination result, where the second determination result is used to represent the final determination result of the traffic depression state of the target cell.

[0122] Through the above-mentioned traffic depression evaluation device, data of sample points in the target cell are collected. The data of the sample points includes at least one of the following: physical resource block (PRB) utilization rate, uplink and downlink traffic, and average number of active users. At least the pre-trained isolation forest model is used to clean the abnormal sample points in the data of the sample points to obtain the data of valid sample points. The target cell is pre-determined according to the data of the valid sample points to obtain a first determination result, and the first determination result is used to represent the pre-determined result of the traffic depression state of the target cell. A piecewise linear regression function is used to fit the traffic data of the target cell to obtain a fitting result, and the first determination result is corrected by using the fitting result to obtain a second determination result. The second determination result is used to represent the final determination result of the traffic depression state of the target cell, thereby achieving the purpose of cleaning the traffic data of the target cell by using a machine learning method and making two determinations based on the data of the cleaned valid sample points, thus realizing the technical effect of accurately evaluating according to the actual traffic data of different cells, and further solving the technical problem that the evaluation method in the related technology is relatively single and has low adaptability to different cells.

[0123] The cleaning module 92 includes: a cleaning sub-module, which is used to at least use the pre-trained isolation forest model to clean the abnormal sample points in the data of the sample points to obtain the data of valid sample points, including: screening out the sample points with invalid values of the average number of active users and the uplink and downlink traffic from the data of the sample points and determining them as sample points to be processed; zero-padding the average number of active users and the uplink and downlink traffic in the sample points to be processed to obtain the data of the initial sample points; obtaining a training data set, which contains the data of multiple sample points in the target cell; generating a plurality of isolation trees based on the training data set, and forming the plurality of isolation trees into an isolation forest. The isolation tree is used to represent the distribution of each sample point in the sample point subset in the target feature dimension. The sample point subset is a set of sample points randomly obtained from the training data set, and the target feature dimension is one of the physical resource block (PRB) utilization rate, uplink and downlink traffic, and average number of active users; screening out abnormal sample points from the data of the initial sample points based on the isolation forest, and removing the abnormal sample points from the initial sample points to obtain the data of the valid sample points.

[0124] The cleaning sub-module includes a screening unit that screens out abnormal sample points from the data of the initial sample points based on the isolation forest, including: sequentially traversing each sample point in the data of the initial sample points through each isolation tree in the isolation forest, and obtaining the average value of the path lengths of each sample point in each isolation tree in the isolation forest; obtaining the average path length of each isolation tree; determining the score of each sample point according to the average value of the path lengths of each sample point in each isolation tree in the isolation forest and the average path length of each isolation tree; and determining the sample points with scores higher than a preset threshold in the data of the initial sample points as the abnormal sample points.

[0125] The traffic depression evaluation device further includes: a clustering sub-module, configured to obtain the data features of each sample point in the data of the valid sample points; determine the type of the target cell based on the data features corresponding to the data of the valid sample points; in the case where the target cell is the first cell, cluster the data of the valid sample points in the target cell to obtain the data of the clustered sample points; pre-determine the target cell based on the data of the clustered sample points to obtain a first determination result; in the case where the target cell is the second cell, pre-determine the target cell according to the PRB utilization rate in the data of the valid sample points, where the distribution trends of the sample points in the first cell and the second cell are different.

[0126] The clustering sub-module includes: a pre-judgment unit, configured to pre-determine the target cell according to the PRB utilization rate in the data of the valid sample points, including: obtaining the PRB utilization rate per second of all valid sample points within a preset period; determining the sample points with the proportion of the duration with the PRB utilization rate per second not less than a first preset ratio in the preset period greater than a second preset ratio as severely depressed sample points; determining the sample points with the proportion of the duration with the PRB utilization rate per second not less than a first preset ratio in the preset period less than a third preset ratio as non-depressed sample points; determining the other valid sample points among all valid sample points as target sample points; counting the severely depressed sample points and the non-depressed sample points in the target cell to obtain a statistical result, and determining the first determination result of the target cell based on the statistical result.

[0127] The prediction unit includes: a prediction subunit, configured to determine the first determination result of the target cell based on the statistical result, including: dividing the valid sample points into multiple bins based on the average number of active users corresponding to each valid sample point, where the average number of active users corresponding to the valid sample points within each bin is the same; obtaining the proportion of severely depressed sample points in all valid sample points within each bin, and determining the bin with the proportion of severely depressed sample points in all valid sample points within the bin greater than the first threshold as the first bin, determining the bin with the proportion of non-depressed sample points in all valid sample points within the bin greater than the second threshold as the second bin, and determining the bin with the proportion of severely depressed sample points in all valid sample points within the bin less than the first threshold as the third bin; in the case where at least one of the first bins is included in the multiple bins, determining the first determination result as the target cell being a severely depressed cell; in the case where all the multiple bins are the second bins, determining the first determination result as the target cell being a non-depressed cell; in the case where the first bin is not included in the multiple bins and at least one of the third bins is included, determining the target cell as a mildly depressed cell.

[0128] The correction module 96 includes: a fitting sub-module, configured to fit the traffic data of the target cell using a piecewise linear regression function to obtain a fitting result, including: in the case where the first determination result of the target cell indicates that the target cell is a severely depressed cell, determining the piecewise linear regression function as a three-segment regression function; in the case where the first determination result of the target cell indicates that the target cell is a mildly depressed cell, determining the piecewise linear regression function as a two-segment regression function; in the case where the first determination result of the target cell indicates that the target cell is a non-depressed cell, determining the piecewise linear regression function as a one-segment regression function; respectively fitting the traffic data of the target cell based on different piecewise linear regression functions to obtain a fitting result.

[0129] The fitting sub-module includes: a correction unit, configured to correct the first determination result using the fitting result to obtain a second determination result, including: in the case where the target cell is the severely depressed cell, dividing the traffic data of the target cell into three sections using an initial segmentation point; in the case where the slope of the fitting curve corresponding to the traffic data in the third section is greater than zero, correcting the target cell to the mildly depressed cell; in the case where the target cell is the mildly depressed cell, obtaining whether the slope of the fitting curve corresponding to the traffic data in the second section is continuously decreasing, and in the case where the slope of the fitting curve corresponding to the traffic data in the second section does not decrease, correcting the target cell to the non-depressed cell.

[0130] The correction unit further includes: an output subunit, configured to obtain the slope of the fitting curve corresponding to the traffic data of the first section, and determine a linear fitting function based on the slope of the fitting curve corresponding to the traffic data of the first section; obtain the average number of activated users corresponding to the traffic data, and determine the required traffic based on the average number of activated users corresponding to the traffic data and the linear fitting function; and determine the difference between the required traffic and the actual traffic as the traffic depression value.

[0131] It should be noted that Figure 9 the traffic depression evaluation device shown is used to execute Figure 2 the traffic depression evaluation method shown. Therefore, the relevant explanations in the above traffic depression evaluation method also apply to this traffic depression evaluation device, and will not be elaborated here.

[0132] An embodiment of the present application further provides a computer device, including: a memory and a processor, where the memory is used to store program instructions; the processor is connected to the memory and is configured to execute the above traffic depression evaluation method.

[0133] The method executed by the above computer device includes collecting the standing wave ratio data of multiple passive distributed antenna systems; determining the fluctuation of the standing wave ratios of the multiple passive distributed antenna systems according to the standing wave ratio data of the multiple passive distributed antenna systems, and determining a first abnormal passive distributed antenna system from the standing wave ratio data of the multiple passive distributed antenna systems according to the fluctuation of the standing wave ratios of the multiple passive distributed antenna systems; determining a training data set according to the user terminal data covered by the first abnormal passive distributed antenna system; and training an identification model based on the training data set, where the identification model is used to identify the user terminal data covered by the to-be-monitored distributed antenna system to determine whether the to-be-monitored distributed antenna system is abnormal, thereby achieving the purpose of determining the abnormal distributed antenna system data according to the fluctuation of the standing wave ratio and training an abnormal distributed antenna system identification model using the determined abnormal distributed antenna system data, thus realizing the technical effect of improving the accuracy of passive distributed antenna system fault prediction, and further solving the technical problem of low accuracy of passive distributed antenna system fault prediction in the related art.

[0134] An embodiment of the present application further provides a non-volatile storage medium, which includes a stored computer program. The device where the non-volatile storage medium is located executes the above traffic depression evaluation method by running the computer program.

[0135] The method stored in the above non-volatile storage medium includes collecting the standing wave ratio data of multiple passive distributed antenna systems; determining the fluctuation condition of the standing wave ratio of the multiple passive distributed antenna systems according to the standing wave ratio data of the multiple passive distributed antenna systems, and determining the first abnormal passive distributed antenna system from the standing wave ratio data of the multiple passive distributed antenna systems according to the fluctuation condition of the standing wave ratio of the multiple passive distributed antenna systems; determining a training data set according to the user terminal data covered by the first abnormal passive distributed antenna system; training an identification model based on the training data set, where the identification model is used to identify the user terminal data covered by the to-be-monitored distributed antenna system to determine whether the to-be-monitored distributed antenna system is abnormal, thereby achieving the purpose of determining the abnormal distributed antenna system data according to the fluctuation condition of the standing wave ratio and training an abnormal distributed antenna system identification model by using the determined abnormal distributed antenna system data, thereby realizing the technical effect of improving the fault prediction accuracy of the passive distributed antenna system, and further solving the technical problem of the low fault prediction accuracy of the passive distributed antenna system in the related art.

[0136] The embodiment of the present application also provides a computer program product, including computer instructions, which implement the steps of the traffic depression evaluation method in the present application when executed by a processor.

[0137] The serial numbers of the above embodiments of the present application are only for description and do not represent the advantages or disadvantages of the embodiments.

[0138] In the above embodiments of the present application, the descriptions of the respective embodiments have their own emphases. For parts not detailed in a certain embodiment, reference may be made to the relevant descriptions of other embodiments.

[0139] In several embodiments provided by the present application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only illustrative. For example, the division of the units can be a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point, the displayed or discussed mutual coupling or direct coupling or communication connection can be through some interfaces, and the indirect coupling or communication connection of units or modules can be in an electrical or other form.

[0140] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place, or can be distributed to multiple units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0141] In addition, in each embodiment of the present application, the functional units can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0142] If the above-mentioned integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The foregoing storage medium includes: various media such as USB flash drives, read-only memories (ROMs), random access memories (RAMs), mobile hard disks, magnetic disks, or optical discs that can store program codes.

[0143] The above are only the preferred embodiments of the present application. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present application, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present application.

Claims

1. A method for evaluating flow suppression, characterized in that: include: Collecting data of sample points of the target cell, wherein the data of the sample points includes at least one of the following: physical resource block (PRB) utilization, uplink and downlink traffic, and average number of activated users; At least one pre-trained isolation forest model is used to clean abnormal sample points in the data of the sample points to obtain data of valid sample points; Pre-determine the target cell according to the data of the valid sample point to obtain a first determination result, where the first determination result is used to indicate a pre-determination result of a traffic suppression state of the target cell; A piecewise linear regression function is used to fit the traffic data of the target cell to obtain a fitting result, and the fitting result is used to correct the first judgment result to obtain a second judgment result, which is used to represent the final judgment result of the traffic suppression state of the target cell.

2. The method according to claim 1, characterized in that At least a pre-trained isolation forest model is used to clean abnormal sample points in the data of the sample points to obtain data of valid sample points, including: Filter out the sample points where the average number of activated users and the uplink and downlink traffic are invalid values ​​from the data of the sample points, and determine them as the sample points to be processed; Pad the average number of activated users and the uplink and downlink traffic in the sample point to be processed with zeros to obtain data of the initial sample point; Acquire a training data set, where the training data set includes data of multiple sample points in the target cell; Based on the training data set, a plurality of isolated trees are generated, and the plurality of isolated data are combined into an isolated forest, wherein the isolated tree is used to represent the distribution of each sample point in a sample point subset on a target feature dimension, wherein the sample point subset is a set of sample points randomly obtained from the training data set, and the target feature dimension is one of the physical resource block PRB utilization, uplink and downlink traffic, and an average number of activated users; Based on the isolation forest, abnormal sample points are screened out from the data of the initial sample points, and the abnormal sample points are removed from the initial sample points to obtain the data of the valid sample points.

3. The method according to claim 2, characterized in that Screening out abnormal sample points from the data of the initial sample points based on the isolation forest includes: Each sample point in the data of the initial sample point is sequentially traversed through each isolated tree in the isolation forest to obtain an average value of the path length of each sample point in each isolated tree in the isolation forest; Get the average path length of each isolated tree; Determine the score of each sample point according to the average value of the path length of each isolated tree in the isolation forest and the average path length of each isolated tree; The sample points in the data of the initial sample points whose scores are higher than a preset threshold are determined as the abnormal sample points.

4. The method according to claim 1, characterized in that: The method further comprises: Acquire data features of each sample point in the data of the valid sample points; Determine the type of the target cell based on data features corresponding to the data of the valid sample points; In a case where the target cell is the first cell, clustering data of valid sample points in the target cell to obtain data of clustered sample points; Pre-determine the target cell based on the data of the clustered sample points to obtain a first determination result; In the case that the target cell is the second cell, the target cell is pre-determined according to the PRB utilization rate in the data of the valid sample points, wherein the distribution trends of the sample points in the first cell and the second cell are different.

5. The method according to claim 4, characterized in that Pre-determining the target cell according to the PRB utilization rate in the data of the valid sample point includes: Obtain the PRB utilization rate per second of all valid sample points within a preset period; Determine the valid sample points whose proportion of the time duration during which the PRB utilization rate per second is not less than the first preset ratio in the preset period is greater than the second preset ratio as the severely suppressed sample points; Determine valid sample points in which the PRB utilization rate per second within the preset period is not less than the first preset ratio and the ratio of the duration within the preset period to the preset period is less than the third preset ratio as non-suppressed sample points; Determine other valid sample points among all valid sample points as target sample points; The heavily suppressed sample points and the non-suppressed sample points in the target cell are counted to obtain statistical results, and the first determination result of the target cell is determined based on the statistical results.

6. The method according to claim 5, characterized in that Determining the first determination result of the target cell based on the statistical result includes: Dividing the valid sample points into a plurality of boxes based on the average number of activated users corresponding to each valid sample point, wherein the average number of activated users corresponding to the valid sample points in each box is the same; Obtain the ratio of the heavily suppressed sample points in all the valid sample points in the sealed boxes, and determine the sealed box in which the ratio of the heavily suppressed sample points in all the valid sample points in the sealed box is greater than a first threshold value as the first sealed box, determine the sealed box in which the ratio of the non-suppressed sample points in all the valid sample points in the sealed box is greater than a second threshold value as the second sealed box, and determine the sealed box in which the ratio of the heavily suppressed sample points in all the valid sample points in the sealed box is less than the first threshold value as the third sealed box; In a case where the plurality of boxes contain at least one of the first boxes, determining that the first determination result is that the target cell is a severely depressed cell; In a case where the plurality of sealed boxes are all the second sealed boxes, determining that the first determination result is that the target cell is a non-suppressed cell; When the plurality of boxes do not include the first box and include at least one third box, the target cell is determined to be a lightly depressed cell.

7. The method according to claim 1, characterized in that The flow data of the target cell is fitted using a piecewise linear regression function to obtain a fitting result, including: When the first determination result of the target cell indicates that the target cell is a severely depressed cell, determining the piecewise linear regression function to be a three-segment regression function; When the first determination result of the target cell indicates that the target cell is a lightly depressed cell, determining that the piecewise linear regression function is a two-segment regression function; When the first determination result of the target cell indicates that the target cell is a non-suppressed cell, determining that the piecewise linear regression function is a one-segment regression function; The flow data of the target cell is fitted based on different piecewise linear regression functions to obtain fitting results.

8. The method according to claim 7, characterized in that The first determination result is corrected by using the fitting result to obtain a second determination result, including: In the case where the target cell is the severely suppressed cell, the flow data of the target cell is divided into three sections using an initial segmentation point, the three sections including: a first section, a second section and a third section; When the slope of the fitting curve corresponding to the traffic data of the third section is greater than zero, the target cell is corrected to the slightly depressed cell; When the target cell is the slightly suppressed cell, it is determined whether the slope of the fitting curve corresponding to the traffic data of the second section is decreasing. When the slope of the fitting curve corresponding to the traffic data of the second section is not decreasing, the target cell is corrected to the non-suppressed cell.

9. The method according to claim 8, characterized in that The method further comprises: Obtaining a slope of a fitting curve corresponding to the flow data of the first section, and determining a linear fitting function based on the slope of the fitting curve corresponding to the flow data of the first section; Obtaining an average number of activated users corresponding to the traffic data, and determining a demand traffic based on the average number of activated users corresponding to the traffic data and the linear fitting function; The difference between the required flow rate and the actual flow rate is determined as the flow suppression value.

10. A flow suppression evaluation device, characterized in that: include: A collection module, used to collect data of sample points of the target cell, wherein the data of the sample points includes at least one of the following: physical resource block PRB utilization, uplink and downlink traffic, and average number of activated users; A cleaning module, used to clean abnormal sample points in the data of the sample points by at least using a pre-trained isolation forest model to obtain data of valid sample points; A determination module, configured to perform a pre-determination on the target cell according to the data of the valid sample point to obtain a first determination result, wherein the first determination result is used to indicate a pre-determination result of a traffic suppression state of the target cell; The correction module is used to fit the traffic data of the target cell using a piecewise linear regression function to obtain a fitting result, and use the fitting result to correct the first judgment result to obtain a second judgment result, wherein the second judgment result is used to represent the final judgment result of the traffic suppression state of the target cell.

11. A computer device, characterized in that: include: A memory and a processor, wherein the memory is used to store program instructions; The processor is connected to the memory, and is used to execute the flow suppression evaluation method described in any one of claims 1 to 9.

12. A computer program product comprising computer instructions, characterized in that: When the computer instructions are executed by a processor, the flow suppression evaluation method according to any one of claims 1 to 9 is implemented.

Images