Method and apparatus for estimating data traffic suppressed by radio problems

Abstract

A method and apparatus for analyzing data are provided. A method includes: classifying a plurality of data samples of a communication network into a plurality of groups corresponding to a plurality of data service scenarios, wherein the data samples of the plurality of data samples include one or more measurements of network performance in a cell and one or more measurements of service quality in a cell, and wherein the network performance includes data traffic; based on the data samples in the group corresponding to the data service scenario, establishing a data traffic volume variation model with service quality for the data service scenario; based on the variation model, identifying an inflection point of a variation trend of data traffic volume with service quality deterioration; and based on the inflection point, identifying a cell with data traffic suppression.

Description

Technical Field

[0001] Embodiments of the present disclosure relate generally to data analysis, and in particular to methods, apparatus, and computer-readable storage media for estimating data traffic inhibited by radio problems in a communication network from data samples related to network performance and quality of service. Background Art

[0002] The current demand for data services is growing rapidly, especially after the introduction of unlimited package rates. Sometimes, existing network resources cannot meet the data service needs of users. Then, the data service needs of users may be suppressed due to radio problems (such as poor coverage, poor quality, etc.). For example, when it takes a long time for a user to open a web page he / she wants to visit, then the user may reduce web browsing or even give up web browsing. In the present disclosure, this phenomenon in which data service demand is suppressed by radio problems is called data service suppression, and the cell in which this phenomenon occurs is called a cell with data service suppression.

[0003] How to achieve accurate investment, including network construction and network expansion, has become one of the business pain points of network operators. If data service suppression can be identified and estimated, this problem will be solved. Network operators can achieve accurate investment by considering the data service suppression of each cell.

[0004] However, it is difficult to identify and estimate data traffic suppression in communication networks. This is because it is affected by many factors, such as different social scenarios, different population sizes, different wireless environments, different time periods, different business types, different user behaviors, etc. Identifying and estimating data traffic suppression has become one of the pain points for network operators. Summary of the invention

[0005] The present disclosure will solve the above problems by proposing a solution for identifying and estimating data service suppression. Other features and advantages of the embodiments of the present disclosure will also be understood from the following description of specific embodiments when read in conjunction with the accompanying drawings, which illustrate the principles of the embodiments of the present disclosure by way of example.

[0006] According to various but not necessarily all embodiments, a method is provided, comprising: classifying a plurality of data samples of a communication network into a plurality of groups corresponding to a plurality of data service scenarios. The data samples in the plurality of data samples include one or more measurements of network performance in a cell and one or more measurements of service quality in the cell. The network performance includes data traffic. The method further comprises: establishing a data traffic variation model with service quality for the data service scenario based on the data samples in the group corresponding to the data service scenario in the plurality of data service scenarios; identifying an inflection point of the variation trend of the data traffic with deterioration of service quality based on the variation model; and identifying a cell with data service suppression based on the inflection point.

[0007] In some but not necessarily all examples, the method may also include collecting a plurality of data samples; and associating each data sample with a characteristic of a data business scenario.

[0008] In some but not necessarily all examples, the classification may be based on a clustering algorithm in which data samples of the same cell are classified into the same group.

[0009] In some but not necessarily all examples, network performance may also include utilization of physical resources. The method may also include: establishing a relationship model between the utilization of physical resources and the data traffic for the data traffic scenario based on data samples in a group corresponding to the data traffic scenario; and estimating a value of data traffic suppression for the identified cell based on the relationship model.

[0010] In some but not necessarily all examples, the value of data traffic suppression can be estimated by the following operations: calculating the theoretical data traffic volume of the identified cell with data traffic suppression based on a relationship model between the utilization of physical resources and the data traffic volume; and estimating the value of data traffic suppression based on the difference between the theoretical data traffic volume and the actual data traffic.

[0011] In some but not necessarily all examples, the method may also include: selecting data samples from a group of data samples corresponding to the data service according to predefined rules to construct a benchmark data set for the data service scenario; and establishing the relationship model based on the benchmark data set.

[0012] In some but not necessarily all examples, the method may further include: calculating a relationship model between the utilization of the physical resources and the data traffic using generalized linear fitting.

[0013] In some but not necessarily all examples, the variation model may be established by: selecting data samples from a group of data samples such that the selected data samples are uniformly distributed in a sample space of the group; and calculating the variation model based on the selected data samples.

[0014] In some but not necessarily all examples, a cell with data traffic suppression may be identified by: identifying a data sample having a quality of service measurement worse than a quality of service value corresponding to the inflection point in the variation model; and determining the cell of the identified data sample as a cell with data traffic suppression.

[0015] In some but not necessarily all examples, the network performance may include parameters indicated by key performance indicators (KPIs) of the network, and the quality of service may include parameters indicated by key quality indicators (KPIs) of the network.

[0016] According to various but not necessarily all embodiments, there is provided an apparatus comprising: at least one processor; and at least one memory including computer program code, the at least one memory and the computer program code being configured to, together with the at least one processor, perform the method.

[0017] According to various but not necessarily all embodiments, a device is provided, comprising: a module for classifying multiple data samples of a communication network into multiple groups corresponding to multiple data service scenarios, wherein data samples in the multiple data samples include one or more measurements of network performance in a cell and one or more measurements of service quality in the cell, and the network performance includes data traffic; a module for establishing a model for the change of data traffic with the service quality for the data service scenario based on data samples in a group corresponding to a data service scenario in the multiple data service scenarios; a module for identifying an inflection point of the change trend of the data traffic with deterioration of service quality based on the change model; and a module for identifying a cell with data traffic suppression based on the inflection point.

[0018] According to various but not necessarily all embodiments, there is provided a computer program comprising instructions for causing an apparatus to perform the method.

[0019] According to various, but not necessarily all, embodiments, there is provided a computer readable storage medium having stored thereon instructions which, when executed by at least one processor, cause the at least one processor to perform the method.

[0020] According to various, but not necessarily all, embodiments, examples are provided as claimed in the following claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Some example embodiments will now be described with reference to the accompanying drawings, in which:

[0022] Figure 1 A system is shown in which embodiments of the present disclosure may be implemented;

[0023] Figure 2 is a component diagram of a device capable of analyzing data service suppression according to an embodiment of the present disclosure;

[0024] Figure 3 is a flowchart describing a process for analyzing data traffic suppression according to an embodiment of the present disclosure;

[0025] Figure 4 Schematically shows a classification fitting example diagram according to an embodiment of the present disclosure;

[0026] Figure 5 is a flow chart describing a process for identifying a cell with data traffic suppression according to an embodiment of the present disclosure;

[0027] Figures 6A-6D The figure shows a simulation of the algorithm used to identify the inflection point;

[0028] Figure 7 is a flowchart depicting an exemplary process for classifying a data sample according to some embodiments of the present disclosure; and

[0029] Figure 8 A simplified block diagram of an apparatus according to an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0030] Embodiments of the present disclosure are described in detail below in conjunction with the accompanying drawings. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the present disclosure, rather than to impose any limitation on the scope of the present disclosure. Throughout the specification, references to features, advantages, or similar language do not mean that all features and advantages that can be achieved by the present disclosure should be any single embodiment of the present disclosure or in any single embodiment of the present disclosure. On the contrary, the language referring to features and advantages is understood to mean that the specific features, advantages, or characteristics described in conjunction with the embodiments are included in at least one embodiment of the present disclosure.

[0031] In addition, in one or more embodiments, the described features, advantages and characteristics of the present disclosure may be combined in any suitable manner. Those skilled in the relevant art will recognize that the present disclosure may be implemented without one or more specific features or advantages of a particular embodiment. In other cases, additional features and advantages that may not be present in all embodiments of the present disclosure may be recognized in certain embodiments.

[0032] As described above, data traffic suppression in a communication network is affected by too many factors. This makes it difficult to identify when and where data traffic suppression occurs in a communication network. This also makes it very difficult to accurately estimate the degree of data traffic suppression. The present disclosure provides a solution for identifying and estimating data traffic suppression by analyzing data samples of a communication network.

[0033] Figure 1 The system in which the embodiments of the present disclosure can be implemented is illustrated. In one embodiment, the communication network is a cellular network. As shown in the figure, the cellular network 110 can be divided into a plurality of coverage areas, hereinafter referred to as cells, such as cells 111, 112, and 113. A cell can be the coverage area of ​​a base station, or the coverage area of ​​any component (or set of components), such as BS 102a-102c, which is configured to provide wireless access to user devices in the network. For example, a base station can be a NodeB, an evolved base station (eNB), a macro cell, a femto cell, or a Wi-Fi access point, etc. A user device can be any component (or set of components) capable of establishing a wireless connection with a base station. A user device can be a mobile terminal, a repeater, or a device engaged in machine type communication used in the art.

[0034] The base station can establish an uplink and / or downlink connection with the user equipment to carry data services from the user equipment to the base station, and vice versa. For example, data services may include but are not limited to streaming media services, web browsing, file downloading, Voice over Internet Protocol (VoIP), Hypertext Transfer Protocol (HTTP) services, Instant Messaging (IM) services, File Transfer Protocol (FTP) services, point-to-point services, etc.

[0035] Examples of communication networks to which the present technology and system are applicable include wireless communication systems based on Code Division Multiple Access (CDMA) (e.g., CDMA20001x), High Rate Packet Data (HRPD), Long Term Evolution (LTE), Long Term Evolution-Advanced (LTE-A), 5th Generation (5G) cellular systems, Universal Terrestrial Radio Access Network (UTRAN), and Worldwide Interoperability for Microwave Access (WiMAX). It should be understood that the illustrated embodiments are non-limiting, and any number of various wireless devices and telecommunication systems may be used as will be readily appreciated by those skilled in the art.

[0036] It should be appreciated that the disclosed wireless communication network is a non-limiting example and may employ any number of different network components, services, and devices as are well known in the art.

[0037] In a communication network, in order to measure the communication performance of the network, various indicators can be used to monitor the network, such as key performance indicators (KPI) and key quality indicators (KQI). KPI can be used as an indication of network performance. KPI can be measured at the network layer or the layer below the network layer for a system that complies with OSI (Open System Interconnection) or at an equivalent layer for a non-OSI system. For example, KPI can indicate in real time how many users are in the network, the amount of data traffic in the network, the total usage of the network, the utilization of physical resources in the network, the degree of interference in the network, or other network performance. KPI may include, for example, reference signal received power (RSRP), channel quality information (CQI), signal to interference plus noise ratio (SINR), physical resource block (PRB) utilization (DL / UL), throughput, handover (HO) failure rate, etc.

[0038] KQI can be used as an indication of the quality of service that reflects the user's perception of the service. For OSI-compliant systems, KQI can be measured at a layer above the network layer, and for non-OSI systems, KQI can be measured at an equivalent layer. For example, KQI can indicate how well the network performs for web browsing services, video streaming services, instant messaging services, or other services, which directly affects the user's perception of the network service. KQI can include, for example, TCP (Transmission Control Protocol) delay, page open delay, first packet delay, average packet loss rate, etc.

[0039] In a communication network, KPIs and KQIs may be measured periodically. For example, measurements may be performed every hour, every day, or at any other predefined time period. Measurements may be performed per cell. Measurements of KPIs and KQIs may be reported from a user equipment to a base station, and / or generated at a base station.

[0040] These measurements can be collected by collector 101 and stored in database 103. Database 103 can be any form of storage or storage system. For example, collector 101 can be any component (or combination of components) configured to collect measurements of KPIs and KQIs of the network. Collector 101 can be co-located with a base station (e.g., BS102a). Optionally, collector 101 can be a separate and different device from the base station, in which case collector 101 can communicate with the base station of the network. In this regard, collector 101 can be implemented as a component of an operations support system (OSS) of the network and communicate with the base station of the network through the communication capabilities of the OSS. In another embodiment, the controller can also be a device communicatively linked to the OSS.

[0041] In some embodiments, the communication network may have engineering parameters that serve as an indication of the network configuration. For example, the engineering parameters may be network characteristics associated with each cell. The engineering parameters may indicate, for example, the frequency band and / or bandwidth allocated to the cell, the social scene located in the cell (e.g., a business center, a town center, a residential area, a college, a highway, a transportation hub, a rural area, etc.), an identifier of the cell, and the like. Some of these parameters may be fixed, while other parameters may be adjusted manually or automatically. The engineering parameters may also be stored in the database 103.

[0042] The data processing device 104 is coupled to the database 103 and can retrieve data from the database 103. From the database 103, the data processing device 104 can obtain measurements of KPIs and KQIs for multiple cells of the communication network and engineering parameters 110 as data samples for analyzing data traffic suppression according to the scheme proposed in the present disclosure. In other embodiments, the data processing device 104 can obtain the measurements of KPIs and KQIs and engineering parameters in an offline manner. In other embodiments, the data processing device 104 can collect measurements of KPIs and KQIs and engineering parameters directly from the network 110 (e.g., from the collector 101) or the base station (e.g., BS 102a-102c).

[0043] like Figure 1 As shown, database 103 may be a component external to network 110, coupled to network 110 via collector 101. In other embodiments, database 103 may be part of communication network 110.

[0044] like Figure 1 As shown, the data processing device 104 may be a device independent of the communication network 110. In other embodiments, the data processing device 104 may be an independent component in the communication network 110, or may be included as part of any network component. For example, the data processing device 104 may be part of a base station, a user equipment, or any other component.

[0045] Figure 2 2 is a component diagram of an apparatus capable of analyzing data traffic suppression according to an embodiment of the present disclosure. It is contemplated that the functions of these components may be combined in one or more components or performed by other components having equivalent functions. The apparatus may be a data processing device 200, which may be coupled to a data source 230. The data source 230 may be, for example, Figure 1 The database 103 shown. Or any component (or combination of components) that can provide data samples required for analyzing data traffic suppression.

[0046] The data processing device 200 may include a communication interface (not shown) for communicating with the data source 230. Optionally, the data processing device 200 may include a user interface (not shown) through which data samples and user settings or instructions may be manually input into the data processing device 200. Optionally, the data processing device 200 may include an external memory interface (not shown) through which data samples may be imported into the data processing device 200 offline.

[0047] The data processing device 200 may include a classification module 202, a fitting module 204, a suppression identification module 206, and a suppression estimation module 208. The classification module 202 is used to classify multiple data samples of the communication network into multiple groups corresponding to multiple data service scenarios. The data sample may include one or more measurements of network performance and one or more measurements of service quality. Classification may be performed based on one or more features of the data service scenario. Different groups reflect the characteristics of different data service scenarios. These features of the data service scenario can be used to distinguish data service scenarios. For example, the characteristics of the data service scenario may include, but are not limited to, social scenarios, number of users, radio environment, time period, and service type. For example, based on the experience of experts, the characteristics of the data service scenario for classifying the data sample may be predetermined. Additionally or alternatively, the characteristics of the data service scenario for classifying the data sample may be automatically determined by machine learning. In the present disclosure, a clustering algorithm is proposed to implement adaptive classification, which can classify the data sample more reasonably. The clustering algorithm will be discussed in detail later.

[0048] The fitting module 204 is configured to establish a relationship model between the utilization of physical resources and the data traffic for each data service scenario based on data samples classified into groups corresponding to each data service scenario. For data samples, the utilization of physical resources and the data traffic in the cell can be determined from the measurement of KPIs. For example, the utilization of physical resources in the downlink can be calculated from the downlink PRB utilization. Any suitable fitting model can be used to model the relationship. In order to increase robustness, generalized linear fitting can be used. The modeling relationship for a specific data service scenario can be used to estimate the theoretical data traffic corresponding to the given utilization of physical resources.

[0049] The suppression identification module 206 is configured to identify cells with data service suppression for each data service scenario. In this regard, the suppression identification module 206 can be configured to identify the inflection point of the changing trend of the data traffic volume as the service quality deteriorates for each data service scenario based on the data samples in the respective groups corresponding to each data service scenario. The inflection point indicates the extent to which the KQI degrades and the data traffic volume stops increasing and / or begins to decrease. Data service suppression can be reflected by the reduction of data traffic volume and the deterioration of KQI at the inflection point. It can be inferred that as the KQI deteriorates from the inflection point, data service suppression is very likely to occur. Therefore, cells with data service suppression can be identified based on the inflection point.

[0050] The suppression estimation module 208 is configured to estimate the degree of data traffic suppression of the cell identified by the suppression identification module 206. In this regard, the theoretical data traffic of the identified cell with data traffic suppression can be calculated based on the physical resource utilization of the network in the state where the identified data traffic suppression occurs, according to the relationship model between the physical resource utilization and the data traffic calculated in the fitting module 204. The degree of data traffic suppression can be estimated as a value of data traffic suppression, which can be calculated based on a comparison between the theoretical data traffic and the data traffic actually measured under specific conditions of utilization of physical resources in the cell.

[0051] Figure 3 is a diagram depicting a process for analyzing data traffic suppression according to an embodiment of the present disclosure. The data processing devices 104, 200 described above may implement the process. In addition, the embodiments described are not limited to being implemented by the data processing device 200. For example, any network component (e.g. Figure 1 Those depicted in ) can all be configured to implement the disclosed processes.

[0052] At 310, a plurality of data samples of the communication network may be classified into a plurality of groups corresponding to data traffic scenarios. The data samples include one or more measurements of network performance in a cell and one or more measurements of quality of service. For example, network performance may include data traffic and utilization of physical resources. Quality of service may include transmission control protocol (TCP) delay. In one example, data samples may be sampled per cell. In addition, data samples may be sampled periodically, such as every hour or every day.

[0053] Figure 4A schematic diagram of such classification in a shaded polygonal box is schematically illustrated. As shown in the figure, data samples can be classified into groups corresponding to multiple data business scenarios according to different social scenarios, different frequency bands and bandwidths, busy or idle times, different user scales, and different service types. For example, data samples associated with social scenarios of business centers, 1800MHz frequency bands, 15M bandwidth, busy times, user scales of 0 to 50 users, and large package service types can be classified into the same group. It should be noted that groups can be classified according to any suitable network characteristics or standards. Network features can be associated with data samples in advance (for example, even in the data source 230). In some embodiments, the criteria for classification can be predefined, for example, based on the experience of experts. In other embodiments, data samples can be adaptively classified, for example by utilizing a clustering algorithm. Reference will be made later. Figure 7 Describe the details of the classification.

[0054] At 320, a cell with data service suppression can be identified based on the data samples in the corresponding group for each data service scenario. In this regard, data service suppression for a specific data service scenario is identified based on the inflection point of the changing trend of data service volume as KQI deteriorates in the specific data service scenario. It has been found that the root cause of suppressing user data service demand is that user perception is affected. Generally, data service volume increases as KQI decreases. This is because as data services in the cell increase, network resources may become increasingly insufficient. Therefore, KQI will decrease. When KQI decreases to a threshold, the data service volume in the cell may not increase as KQI decreases. It can be determined that in this case, the KQI is too poor to allow users to continue to use data services. In other words, the user's data service demand is suppressed. The threshold of KQI is an inflection point, beyond which the data service volume will not continue to increase as KQI decreases.

[0055] Figure 5 An exemplary process for identifying a cell with data traffic suppression based on an inflection point for a specific data traffic scenario is illustrated. Figure 2 The suppression identification module 206 shown is implemented. As described above, the main reason for users to refuse to use the network is the reduction of user perception. Therefore, the main purpose of data service suppression is to clarify the relationship between KQI parameters and user data services. At 510, for a specific data service scenario, a data traffic variation model with KQI parameters can be established based on data samples in a group corresponding to the specific data service scenario. KQI parameters can be selected as parameters indicating user perception associated with radio problems in the communication network. For example, a parameter indicating TCP delay in the service can be selected as a KQI parameter for modeling the variation model.

[0056] At 520, the inflection point of the change trend of data traffic as the quality of service decreases can be identified based on the change model. In one embodiment, the threshold of such a KQI parameter can be determined as an inflection point, and the data traffic will not continue to increase as the KQI decreases beyond this threshold. In some embodiments, the change model can be calculated by utilizing quadratic fitting. In this case, the peak of the fitting curve can be determined as the inflection point. It should be understood that the embodiment is not limited to quadratic fitting, but any other fitting model suitable for identifying the change pattern can be used. Therefore, the inflection point is not limited to the peak point of the fitting curve. For example, the inflection point can be an interval of KQI.

[0057] Figures 6A-6D A simulation of an algorithm for identifying inflection points is illustrated. Fig. 6A The distribution of data samples in the sample space is shown. For example, the horizontal axis x indicates a KQI parameter, such as TCP download delay, and the vertical axis y indicates the total number of downlink bytes (MB). Traditionally, in order to find out the change in downlink data traffic due to an increase in TCP delay, fitting can be directly applied to these samples. Figure 6B The quadratic fit based on these samples is shown. Although the inflection point of the data service can also be found (such as Figure 6B However, since there are few users when TCP latency is high, the inflection point is greatly reduced (ie, biased towards low latency, the left side of the x-axis). It cannot truly reflect that users give up using the network because of increased TCP latency.

[0058] The present disclosure proposes a new algorithm for establishing a change model. According to the algorithm, data samples are further selected so that the selected data samples can be evenly distributed in the sample space. The change model is calculated based on the selected data samples. In this regard, the number of data samples in the case of high TCP delay is not less than the number of data samples in the case of low TCP delay. The inflection point deviation caused by the uneven distribution of data samples can be alleviated. The inflection point identified from the change module may be more reliable.

[0059] Figure 6C The diagram shows an example for making the distribution of data samples in the sample space uniform. The two-dimensional sample space can be divided into M×N grids. Figure 6C In the example shown in , M is set to 13 and N is set to 1. Then, multiple data samples are randomly sampled or selected in each grid. The number of data samples sampled from the grid (indexed by i, i=1, ..., M×N) is denoted by p i , and p i =Min(N samples , N i total count ), where N samplesis a predetermined number of data samples selected from each grid, N i total count is the number of data samples in a grid i. Therefore, the samples will be evenly distributed throughout the two-dimensional sample space. This means that each grid is likely to have equal data sample values.

[0060] Fig.6D The quadratic fit based on these uniformly distributed samples is shown in the figure. Figure 6B In comparison, the inflection point of data service changes (such as Fig.6D The KQI is slightly lower than the traditional algorithm (indicated by the asterisk in the figure). It more accurately reflects the reduced network usage due to poor KQI perception.

[0061] return Figure 5 Based on the inflection point, a cell with data service suppression can be identified at 530. As described above, the inflection point indicates a threshold of the KQI parameter, and when the KQI parameter degrades to exceed the threshold, the user's data service demand will generally be suppressed. Therefore, a data sample whose measurement of the KQI parameter is worse than the threshold of the KQI parameter corresponding to the inflection point in the change model can be identified as a data sample of a data service suppression cell. For example, Figure 6B and 6D The data samples in the shaded area can be identified as data samples of cells with data service suppression. In some embodiments, the identification can be performed in combination with the radio environment indicated by other measurements of the data samples. For example, data samples with a high-quality radio environment can be excluded. The radio environment in which the sampled data samples are located can be determined based on the relevant KPIs of the samples, such as SINR, CQI and other related indicators. It should be understood that since these data samples are sampled in a high-quality radio environment, the KQI degradation of these data samples may not be caused by radio problems.

[0062] Return now Figure 3 , the process may be further performed to estimate the degree of data traffic suppression of each cell with data traffic suppression. In this regard, the process may also include establishing a relationship model between the utilization of physical resources and the amount of data traffic for the data traffic scenario based on the data samples in the group corresponding to the data traffic scenario, as shown in 330. In order to increase the robustness of the relationship module, generalized linear fitting may be used here. Figure 4 The simulation of the linear fitting algorithm for establishing the relationship model between downlink PRB utilization and the total amount of data traffic based on the data samples for each data traffic scenario is shown. As shown in the figure, for different data traffic scenarios, the calculated fitting model may be different.

[0063] In some embodiments, for a specific data service scenario, data samples with a high-quality radio environment may be selected to construct a benchmark data set. In other embodiments, data samples with reduced radio environment quality may be filtered out so that the remaining data samples construct a benchmark data set. In addition, it may be determined that data samples with data service suppression may be excluded from the benchmark data set. In other words, data samples in the benchmark data set will not be identified as data samples with data service suppression. A relationship model may be established based on the benchmark data set. For example, data samples with a high-quality radio environment and / or data samples with reduced radio environment quality may be identified according to predefined rules. It is assumed that these data samples in the benchmark data set can reflect the ideal or desired relationship between the utilization of physical resources and the data traffic, because the radio environment has a small impact on the data traffic. The quality of the radio environment may be indicated by relevant KPI parameters, such as SINR, CQI and other parameters related to the radio environment. The predefined rules may be established based on the experience of experts, machine learning or other applicable mechanisms. It should be noted that the predefined rules for different service scenarios may be different.

[0064] At 340, for each cell with data service suppression identified at 320, the degree of data service suppression can be estimated based on the relationship model between the utilization of physical resources and the data service volume established at 330. For example, if it is determined that the data service of a data sample in a group of a specific data service scenario is suppressed at an inflection point. This indicates that the cell of the data sample is a cell with data service suppression in the data service scenario. The utilization of physical resources by the cell can be calculated by KPI measurements, such as downlink / uplink PRB utilization. Then, based on the relationship model between the utilization of physical resources and the data service volume for the data service scenario, the theoretical data service volume of the cell with data service suppression. At the same time, the actual data service of the cell with data service suppression in the data service scenario can be determined. Based on the difference between the theoretical data service and the actual data service, the degree to which the needs of users in the cell are suppressed in the data service scenario can be estimated.

[0065] If cells with data traffic suppression can be identified, and the degree or amount of data traffic suppression in the cells can be estimated, the results of the identification and estimation can guide network construction and network expansion under limited resources. Resource investment is more suitable for the actual data traffic demand of users. For example, the value of data traffic suppression can maximize the use of resources, ensure user perception, and improve the return on investment (ROI).

[0066] Figure 7 It is a description of the previous Figure 3A flowchart of a detailed example of the process 300 is shown in FIG. Figure 2 The classification module 202 shown is implemented as Figure 2 As described. At 710, multiple data samples are collected. The data samples can be obtained from a data source 230. Optionally, the data samples can be filtered to remove outliers in the data samples, as shown in 720. Optionally, each data sample can be associated with at least one feature of a data service scenario, as shown in 730. For example, the measurement of KPIs and KQIs can be associated with identifiers indicating when the KPIs and KQIs were measured and in which cell the KPIs and KQIs were measured. At 740, the data samples can be clustered into multiple groups corresponding to multiple data service scenarios, for example, based on the characteristics of the data service scenarios. For example, the characteristics of the data service scenarios include, but are not limited to, average data traffic, number of users, and time period.

[0067] According to an embodiment of the present disclosure, data service scenarios can be automatically divided by using a clustering algorithm based on data samples. Traditional scenario division algorithms are divided into three categories:

[0068] Divide scenarios based on base station locations (collectively referred to as geographic scenarios);

[0069] Divide scenarios according to the coverage type of base stations and service objects (collectively referred to as service types);

[0070] Divide scenarios based on base station antenna angle, power, station height and other information (collectively referred to as wireless hardware configuration).

[0071] Since the number of wireless base stations in urban areas is already very large, communication services have changed from original voice communication to diversified data services (images, videos, audio, web browsing, etc.), and traditional scene division algorithms cannot truly reflect data service scenarios. In order to accurately divide data service scenarios, this paper proposes a new clustering algorithm, which gives a more accurate scene division by applying and modifying machine learning algorithms (such as K-means clustering algorithm).

[0072] In addition, it is not feasible to directly apply the K-means clustering algorithm to the division of data service scenarios because it assumes that the data is independent regardless of the data context in wireless communication. For example, the normal K-means clustering algorithm does not take into account that many measurements of KPI indicators will change periodically. For example, the data traffic in the network may be high during the day and low at midnight. The data traffic around the business center may increase sharply every weekend. In addition, this periodic change will be reflected by the data samples sampled in each cell. At this point, the periodicity in the data samples should be considered based on the cell identifier. In other words, data samples with the same cell identifier should be classified into the same category according to the time period. Therefore, the clustering algorithm in the field of wireless communication should be an unsupervised clustering with certain constraints.

[0073] In the scene division proposed in the present invention, factors that need to be considered may include but are not limited to: frequency band, bandwidth, social scene, indoor and outdoor, number of people, business type, busy and idle time period, etc. The improved clustering algorithm is based on the K-means algorithm in the clustering algorithm. The concept of the normal K-means algorithm is as follows:

[0074] Step 1: First, randomly select k sample points as the initial starting point, where k is the number of groups to be formed by clustering, and k is predefined.

[0075] Step 2: Calculate the distance from all sample points to the initial starting point, and collect the nearest set of points into the same group.

[0076] Step 3: Recalculate the center of the points in the same group, i.e. the centroid, by using the Euclidean distance.

[0077] Step 4: Recalculate the distances from all points to the centroids of the various groups newly calculated in step 3. These centroids are the new starting points in this regard.

[0078] Step 5: Go back to Step 1 and Step 2 to reclassify the classification and centroid.

[0079] The above steps 1-5 can be repeated until the termination condition is met. For example, the termination condition of the K-means algorithm can be expressed as:

[0080]

[0081] Among them, S i is the set of sample points in the group, j is S i The index of μ i YesS i The centroid (e.g., mean) of S i where i is the index of x. Each sample point x can be a multidimensional vector. Each component of its dimension can correspond to a factor that needs to be considered in the classification as described above.

[0082] Using the K-means algorithm, the groups tend to be spherical. This is because the distances calculated in the algorithm are defined by Euclidean distances. The clusters are selected by initial starting points. These starting points can be randomly selected or determined by some specific algorithm. Since the starting point affects the clustering accuracy, the clustering algorithm can be run multiple times to select the best set of starting points. The time complexity of the algorithm is O(n dk+1 logn), so it runs very fast. In addition, this algorithm is a heuristic algorithm. It is not guaranteed to converge to the global optimal solution, but in most cases, it can converge to a good local solution.

[0083] As mentioned above, the normal K-means clustering algorithm assumes that the data is independent and has nothing to do with the data context. However, wireless communication data is actually closely related to cells. In the present disclosure, an improved algorithm is proposed to divide all data of the same cell into the same group as much as possible.

[0084] In one embodiment, a new distance penalty coefficient p may be added to the improved K-means algorithm. The coefficient p is used to weight the distance between two sample points belonging to the same cell. For example, if the two sample points are samples of the same cell, the penalty coefficient p is set to 1; otherwise, the penalty coefficient p is set to 0.

[0085] The distance between sample points can still be calculated using Euclidean distance. All sample points can be standardized. The arithmetic description of the improved algorithm can be as follows:

[0086] The sample space is Q = X (m) , where X (m) is the sample point x m A vector set of x m is a multidimensional vector, m is x in X (m) The index in ;

[0087] Standardize all sample points so that their mean μ = 0 and variance E = 1;

[0088] Distance definition

[0089]

[0090] in,

[0091] The sample points may then be clustered using a K-means algorithm, for example, with a termination condition of equation (1). One modification is to calculate the distance as the sum of the Euclidean distance and a penalty coefficient. It will be appreciated that the distance may be weighted in any suitable manner to facilitate classifying sample points in the same cell into the same group. In this regard, for example, a cell identifier indicating which cell is associated with the sample point may be a factor of the vector of the sample point.

[0092] Reference now Figure 8 , which shows a simplified block diagram of various electronic devices that are suitable for practicing the exemplary embodiments of this invention. Figure 8 In the embodiment, the data processing device 800 (eg, the data processing devices 104 and 200) is adapted to analyze data traffic suppression based on data samples of KPIs and KQIs of the communication network. The data analysis may be performed according to the exemplary embodiments of the present disclosure as described above.

[0093] The data processing device 800 includes a data processor (DP) 801 and a memory (MEM) 803 storing a program (PROG) 805. Optionally, the data processing device 800 may also include a suitable transceiver 807 for communicating with data sources, network components of communication networks, and other communication devices via one or more communication networks. It is assumed that PROG 805 includes program instructions, which, when executed by DP 801, enable the data processing device to operate according to the exemplary embodiments of the present disclosure as described above. That is, the exemplary embodiments of the present disclosure may be implemented at least in part by computer software executable by DP 801, or by hardware, or by a combination of software and hardware. The basic structure and operation of the data processing device 800 are known to those skilled in the art.

[0094] MEM 803 may be of any type suitable for the local technical environment and may be implemented using any suitable data storage technology, such as semiconductor-based storage devices, flash memory, magnetic storage devices and systems, optical storage devices and systems, fixed memory, and removable memory. As non-limiting examples, DP 801 may be of any type suitable for the local technical environment and may include one or more of a general-purpose computer, a special-purpose computer, a microprocessor, a digital signal processor (DSP), and a processor based on a multi-core processor architecture.

[0095] In general, various exemplary embodiments may be implemented in hardware or dedicated circuits, software, logic, or any combination thereof. For example, some aspects may be implemented in hardware, while other aspects may be implemented in firmware or software that may be executed by a controller, microprocessor, or other computing device, but the invention is not limited thereto. Although various aspects of the exemplary embodiments of the invention may be illustrated and described as block diagrams, flow charts, or using some other graphical representations, it is well understood that, as non-limiting examples, these block diagrams, devices, systems, techniques, or methods described herein may be implemented in hardware, software, firmware, dedicated circuits or logic, general hardware or controllers or other computing devices, or some combination thereof.

[0096] It should therefore be understood that at least some aspects of the exemplary embodiments of the present invention may be implemented in various components such as integrated circuit chips and modules. It should therefore be understood that the exemplary embodiments of the present invention may be implemented in a device embodied as an integrated circuit, wherein the integrated circuit may include circuits (and possible firmware) for embodying at least one or more of a data processor, a digital signal processor, a baseband circuit, and a radio frequency circuit, which can be configured to operate in accordance with the exemplary embodiments of the present invention.

[0097] It should be understood that at least some aspects of the exemplary embodiments of the present invention may be embodied in computer executable instructions executed by one or more computers or other devices, such as in one or more program modules. Typically, a program module includes routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types when executed by a processor in a computer or other device. Computer executable instructions may be stored on a computer readable medium, such as a hard disk, an optical disk, a removable storage medium, a solid-state memory, a random access memory, etc. As will be appreciated by those skilled in the art, in various embodiments, the functions of the program modules may be combined or distributed as desired. In addition, the functions may be embodied in whole or in part in firmware or hardware equivalents, such as integrated circuits, field programmable gate arrays (FPGAs), etc.

[0098] The present invention includes any novel feature or combination of features explicitly disclosed herein, or any generalization thereof. When read in conjunction with the accompanying drawings, various modifications and amendments to the aforementioned exemplary embodiments of the present invention will become apparent to those skilled in the relevant art in view of the foregoing description. However, any and all modifications will still fall within the scope of the non-limiting and exemplary embodiments of the present invention.

Claims

1. A method for identifying data traffic suppression in a communication network, comprising: Classifying a plurality of data samples of the communication network into a plurality of groups corresponding to a plurality of data traffic scenarios, wherein data samples in the plurality of data samples include one or more measurements of network performance in a cell and one or more measurements of quality of service in the cell, and wherein the network performance includes data traffic; Based on data samples in a group corresponding to a data service scenario in the plurality of data service scenarios, establishing a data service volume variation model along with the service quality for the data service scenario; Based on the change model, identifying an inflection point of a change trend of the data traffic as the service quality deteriorates; as well as A cell with data traffic suppression is identified based on the inflection point.

2. The method of claim 1, further comprising: Collect multiple data samples; as well as Associate each data sample with the characteristics of the data business scenario.

3. The method according to claim 1 or 2, wherein the classification is performed based on the characteristics of the data service scenario. 4 . The method according to claim 1 , wherein the classification is based on a clustering algorithm in which data samples of the same cell are classified into the same group.

5. The method according to claim 1, characterized in that The network performance also includes utilization of physical resources, and the method further includes: Based on the data samples in the group corresponding to the data service scenario, establishing a relationship model between the utilization of the physical resources and the data traffic for the data service scenario; and A value for data traffic suppression for the identified cell is estimated based on the relationship model.

6. The method of claim 5, wherein estimating the value of data traffic suppression comprises: Calculating theoretical data traffic volume of the identified cell with data traffic suppression according to a relationship model between the utilization of the physical resources and the data traffic volume; as well as The value of the data traffic suppression is estimated based on a difference between the theoretical data traffic and the actual data traffic.

7. The method of claim 5, further comprising: Selecting data samples from a group of data samples corresponding to the data service scenario according to a predetermined rule to construct a benchmark data set for the data service scenario; as well as The relationship model is established based on the benchmark data set.

8. The method according to claim 5, further comprising: A generalized linear fit is used to calculate a relationship model between the utilization of the physical resources and the data traffic.

9. The method of claim 1, wherein establishing the change model comprises: Selecting data samples from the data samples of the group such that the selected data samples are evenly distributed in the sample space of the group; as well as The variation model is calculated based on the selected data samples.

10. The method of claim 1, wherein identifying a cell with data traffic suppression comprises: identifying data samples having a quality of service measurement that is worse than a quality of service value corresponding to the inflection point in the variation model; as well as The cell of the identified data sample is determined as a cell with data traffic suppression.

11. The method according to claim 1, wherein: The network performance includes parameters indicated by a key performance indicator KPI of the network, and the service quality includes parameters indicated by a key quality indicator KQI of the network.

12. An apparatus for identifying data traffic suppression in a communication network, comprising: at least one processor; and at least one memory including computer program code, the memory and the computer program code being configured to, with the processor, cause the apparatus to at least perform the following operations: classifying a plurality of data samples of the communication network into a plurality of groups corresponding to a plurality of data traffic scenarios, wherein the data samples of the plurality of data samples include one or more measurements of network performance in a cell and one or more measurements of quality of service in the cell, and wherein the network performance includes data traffic; Based on data samples in a group corresponding to a data service scenario in the plurality of data service scenarios, establishing a data service volume variation model along with the service quality for the data service scenario; Based on the change model, identifying an inflection point of a change trend of the data traffic as the quality of service deteriorates; as well as A cell with data traffic suppression is identified based on the inflection point. 13 . The apparatus according to claim 12 , wherein the classification is based on a clustering algorithm in which data samples of the same cell are classified into the same group.

14. The apparatus of claim 12, wherein the network performance further comprises utilization of physical resources, and The memory and the computer program code are configured to, together with the processor, cause the apparatus to perform at least the following operations: Based on the data samples in the group corresponding to the data service scenario, establishing a relationship model between the utilization of the physical resources and the data traffic for the data service scenario; and A value of data traffic suppression for the identified cell is estimated based on the relationship model.

15. The apparatus of claim 14, wherein the memory and the computer program code are configured to, with the processor, cause the apparatus to estimate a value of data traffic suppression by: Calculating theoretical data traffic volume of the identified cell with data traffic suppression according to the relationship model between the utilization of the physical resources and the data traffic volume; and The value of the data traffic suppression is estimated based on a difference between the theoretical data traffic and the actual data traffic.

16. The apparatus of claim 14, wherein the memory and the computer program code are configured to, together with the processor, cause the apparatus to at least further perform the following operations: selecting data samples from a group of data samples corresponding to the data service scenario according to a predetermined rule to construct a reference data set for the data service scenario; and The relationship model is established based on the benchmark data set.

17. The device according to claim 14, wherein: The memory and the computer program code are configured to, together with the processor, cause the apparatus to at least further perform the following operations: The relationship model between the utilization of the physical resources and the data traffic is calculated using generalized linear fitting.

18. The device according to claim 12, wherein: The memory and the computer program code are configured to, together with the processor, cause the apparatus to establish the change model by: selecting data samples from the data samples of the group such that the selected data samples are uniformly distributed in the sample space of the group; and The variation model is calculated based on the selected data samples.

19. The device according to claim 12, wherein: The memory and the computer program code are configured to, together with the processor, cause the apparatus to identify a cell with data traffic suppression by: identifying data samples having a quality of service measurement that is worse than a quality of service value corresponding to the inflection point in the variation model; as well as The cell of the identified data sample is determined as a cell with data traffic suppression.

20. The device according to claim 12, wherein: The network performance includes parameters indicated by a key performance indicator KPI of the network, and the service quality includes parameters indicated by a key quality indicator KQI of the network.

21. An apparatus for identifying data traffic suppression in a communication network, comprising: means for classifying a plurality of data samples of the communication network into a plurality of groups corresponding to a plurality of data traffic scenarios, wherein the data samples of the plurality of data samples include one or more measurements of network performance in a cell and one or more measurements of quality of service in the cell, and wherein the network performance includes data traffic; A module for establishing a data traffic volume variation model with the service quality for the data traffic scenario based on data samples in a group corresponding to the data traffic scenario in the plurality of data traffic scenarios; A module for identifying, based on the change model, an inflection point of a change trend of the data traffic as the quality of service deteriorates; as well as Means for identifying cells having data traffic suppression based on the inflection point.

22. The apparatus according to claim 21, further comprising: A module for establishing a relationship model between utilization of physical resources and the data service for the data service scenario based on data samples in a group corresponding to the data service scenario; as well as Means for estimating a value for data traffic suppression of the identified cell based on the relationship model.

23. The apparatus of claim 22, wherein the means for estimating a value for data traffic suppression comprises: means for calculating theoretical data traffic of the identified cell with data traffic suppression based on a relationship model between the utilization of physical resources and the data traffic volume; as well as Means for estimating a value of the data traffic suppression based on a difference between the theoretical data traffic and the actual data traffic.

24. The apparatus according to claim 21, wherein the module for establishing the change model comprises: A module for selecting data samples from the data samples of the group so that the selected data samples are evenly distributed in the sample space of the group; as well as Means for calculating the variation model based on the selected data samples.

25. The apparatus of claim 21, wherein the means for identifying cells with data traffic suppression comprises: means for identifying data samples having quality of service measurements that are worse than quality of service values ​​corresponding to inflection points in said variation model; as well as Means for determining a cell of the identified data sample as a cell with data traffic suppression.

26. A computer-readable storage medium having instructions stored thereon, which, when executed by at least one processor, cause the at least one processor to: classifying a plurality of data samples of a communication network into a plurality of groups corresponding to a plurality of data traffic scenarios, wherein data samples of the plurality of data samples include one or more measurements of network performance in a cell and one or more measurements of quality of service in the cell, and wherein the network performance includes data traffic; Based on data samples in a group corresponding to a data service scenario in the plurality of data service scenarios, establishing a data service volume variation model along with the service quality for the data service scenario; Based on the change model, identifying an inflection point of a change trend of the data traffic as the quality of service deteriorates; as well as A cell with data traffic suppression is identified based on the inflection point.

27. The computer-readable storage medium of claim 26, wherein when the instructions are executed by at least one processor, the instructions cause the at least one processor to further: Based on the data samples in the group corresponding to the data service scenario, establishing a relationship model between the utilization of physical resources and the data traffic for the data service scenario; and Based on the relationship model, a value of data traffic suppression for the identified cell is estimated.

28. The computer-readable storage medium of claim 27, wherein when the instructions are executed by at least one processor, the instructions cause the at least one processor to further: Calculating theoretical data traffic volume of the identified cell with data traffic suppression according to the relationship model between the utilization of the physical resources and the data traffic volume; and The value of the data traffic suppression is estimated based on the difference between the theoretical data traffic and the actual data traffic.

29. The computer-readable storage medium of claim 26, wherein when executed by at least one processor, the instructions cause the at least one processor to further: selecting data samples from the data samples of the group such that the selected data samples are uniformly distributed in the sample space of the group; and The variation model is calculated based on the selected data samples.

30. The computer-readable storage medium of claim 26, wherein when the instructions are executed by at least one processor, the instructions cause the at least one processor to further: identifying data samples having a quality of service measurement worse than a quality of service value corresponding to the inflection point in the variation model; and The cell of the identified data sample is determined as a cell with data traffic suppression.

Images