Method, device and equipment for locating poor quality of home broadband services

Abstract

The present invention proposes a method, device and equipment for demarcating and locating poor quality of home broadband services, which relates to the field of communication technology. The method includes: obtaining network link data of home broadband services; aggregating application network key performance indicator data based on a network topology diagram to obtain aggregated data, and extracting poor quality data; matching the data into time series data; determining the support and confidence of each data in the time series data, filtering out frequent item data and establishing a frequent pattern tree based on the frequent item data; combining the associated frequent item data with the tree root, and obtaining frequent time series patterns based on the confidence; determining the data to be searched based on the poor quality data, and obtaining the root cause of the poor quality problem and identifying poor quality users based on the data to be searched and the frequent time series patterns. The present invention improves the accuracy and efficiency of demarcation and positioning and reduces the cost of demarcation and positioning, giving users a better home broadband service experience.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a method, device and equipment for locating poor quality of home broadband services. Background Art

[0002] The network link for home broadband services typically consists of terminal devices, routers, switches, optical network units (ONUs), optical distribution networks (ODNs), passive optical networks (PONs), optical line terminals (OLTs), bearer networks, and content source networks. The network link for home broadband services involves numerous interactive devices. Once poor quality issues such as low access, frequent disconnections, and low data rates are detected in user-side applications, operations and maintenance personnel must inspect the relevant devices from the bottom (layer) up according to the network topology of the network link, demarcating and locating the poor quality issue to ultimately find the root cause.

[0003] However, with the increasing number of home broadband users, this bottom-up approach to identifying and locating poor quality services consumes considerable time and labor. It can also suffer from poor accuracy, leading to a poor user experience. Therefore, a more cost-effective method for identifying and locating poor quality services in home broadband services is needed, while maintaining accuracy. Summary of the Invention

[0004] In view of this, the present invention proposes a method, device and equipment for locating poor quality of home broadband services to solve the problem that the current demarcation and positioning of poor quality of home broadband services consumes a lot of time and manpower costs and cannot guarantee accuracy.

[0005] The technical solution of the present invention is achieved as follows:

[0006] According to a first aspect, an embodiment of the present invention provides a method for locating poor quality of a home broadband service, the method comprising:

[0007] Acquire network link data for home broadband services; the network link data includes application network key performance indicator data, home network data, and performance data;

[0008] aggregating the application network key performance indicator data according to a network topology diagram of the home broadband service to obtain aggregated data, and extracting quality difference data from the aggregated data; the quality difference data is used to reflect network failures and to delimit the network layer where the network failure occurs in the network topology diagram;

[0009] Matching the quality difference data, the home network data and the performance data into time series data in chronological order; the time series data includes at least one item of data, and each item of data has a corresponding occurrence time;

[0010] Determining the support and confidence of each item in the time series data according to a preset time span, filtering out frequent item data from the time series data according to the support, and establishing a frequent pattern tree with the frequent item data as a tree root; the frequent pattern tree includes at least one of the frequent item data;

[0011] Determining a time period in which each of the frequent item data in each of the frequent pattern trees appears, associating each of the frequent item data with the corresponding time period, combining all associated frequent item data in each of the frequent pattern trees with the tree root, and obtaining a frequent time series pattern based on the confidence level; each of the frequent time series patterns corresponds to a quality difference association rule;

[0012] Determining the performance data and / or the home network data to be searched based on the poor quality data, and locating the poor quality problem based on the performance data and / or the home network data and the frequent timing pattern to obtain a root cause of the poor quality problem;

[0013] Based on the root cause, poor quality users are identified.

[0014] In combination with the first aspect, in a first implementation of the first aspect, determining the support and confidence of each item of data in the time series data according to a preset time span, screening frequent item data from the time series data according to the support, and establishing a frequent pattern tree with the frequent item data as the root specifically includes:

[0015] Divide the time series data into a number of preset time windows of equal length, and treat each event consisting of the time series data in the preset time window as a transaction;

[0016] Traversing the time series data, determining the occurrence probability of each data item in the time series data among all things according to the preset time span, and obtaining the support degree of each data item;

[0017] The item data whose support is not less than the preset support is regarded as frequent item data;

[0018] A frequent pattern tree is established with the frequent item data as a tree root, to obtain frequent pattern trees having the same number as and corresponding to the frequent item data.

[0019] In combination with the first implementation of the first aspect, in the second implementation of the first aspect, establishing a frequent pattern tree with the frequent item data as the root to obtain the same number of frequent pattern trees as the frequent item data and corresponding to the frequent pattern trees specifically includes:

[0020] Taking each of the frequent item data as a tree root;

[0021] Obtain the frequent item data that is before the frequent item data that is the current tree root in time order;

[0022] The frequent item data are placed into the branches of the tree in the order of their occurrence time to generate a frequent pattern tree of the current root.

[0023] In combination with the second implementation of the first aspect, in the third implementation of the first aspect, acquiring the frequent item data that precedes the frequent item data that serves as the current tree root in time sequence specifically includes:

[0024] Scan the root path of the frequent item data as the current tree root and mark the scanned root path;

[0025] Scan the root paths of the frequent item data of the current tree root until all root paths are marked, and obtain the item data that is located before the frequent item data of the current tree root in time sequence;

[0026] Obtain frequent item data whose support is not less than a preset support in the item data.

[0027] In combination with the first aspect, in a fourth implementation of the first aspect, determining the time period in which each of the frequent item data in each of the frequent pattern trees appears, associating each of the frequent item data with the corresponding time period, combining all associated frequent item data in each of the frequent pattern trees with the tree root, and obtaining a frequent temporal pattern based on the confidence level, specifically includes:

[0028] Determine the time period in which each frequent item data in each frequent pattern tree appears, and associate each frequent item data with the corresponding time period;

[0029] Obtain the frequent item data of the non-empty true subset of the tree root in each frequent pattern tree, and use the frequent items as the initial association rules, determine the confidence of the initial association rules, and use the initial association rules whose confidence exceeds the preset confidence as frequent time series patterns;

[0030] According to the network topology diagram of the home broadband service, it is verified whether each data item in the frequent time series pattern satisfies the network topology relationship, and the data items that do not satisfy the network topology relationship are deleted from the frequent time series pattern.

[0031] In combination with the first aspect, in a fifth implementation of the first aspect, aggregating the application network key performance indicator data according to the network topology diagram of the home broadband service to obtain aggregated data, and extracting quality difference data from the aggregated data specifically includes:

[0032] According to the network topology diagram, and according to the preset window time period, the application network key performance indicator data of each network layer is aggregated to obtain the aggregated data within several preset window time periods corresponding to each network layer;

[0033] The aggregated data of each network layer is detected for traffic anomalies from the lower layer to the upper layer of the network topology diagram, and the poor quality data in the aggregated data of the network layer with poor quality problems is extracted.

[0034] In combination with the first aspect, in a sixth implementation of the first aspect, the support degree is calculated as follows:

[0035]

[0036] Among them, |A∪B| represents the number of time periods in the time series data that contain item sets A and B and maintain the order of item sets A and B in the time series database, and the time series database is used to store and record all time series data; |D| represents the number of time periods in the time series data with a time span of the preset time span E.

[0037] In combination with the first aspect, in a seventh implementation of the first aspect, the confidence calculation formula is:

[0038]

[0039] Among them, |A∪B| represents the number of time periods in the time series data that contain item sets A and B and maintain the order of item sets A and B in the time series database. The time series database is used to store and record all time series data; |A| represents the number of time periods in the time series data that contain item set A.

[0040] According to a second aspect, an embodiment of the present invention further provides a device for locating and demarcating poor quality of a home broadband service, the device comprising:

[0041] A data acquisition module is used to acquire network link data of the home broadband service; the network link data includes application network key performance indicator data, home network data and performance data;

[0042] a quality difference delimiting module for aggregating the application network key performance indicator data based on the network topology diagram of the home broadband service to obtain aggregated data, and extracting quality difference data from the aggregated data; the quality difference data is used to reflect network failures and delimit the network layer where the network failure occurs in the network topology diagram;

[0043] a data matching module, configured to match the quality difference data, the home network data, and the performance data into time series data in chronological order; the time series data including at least one item of data, and each item of data having a corresponding occurrence time;

[0044] a tree building module, configured to determine the support and confidence of each data item in the time series data according to a preset time span, filter out frequent item data from the time series data according to the support, and build a frequent pattern tree with the frequent item data as the tree root; the frequent pattern tree includes at least one frequent item data;

[0045] a rule association module, configured to determine the time period in which each of the frequent item data in each of the frequent pattern trees appears, associate each of the frequent item data with the corresponding time period, combine all associated frequent item data in each of the frequent pattern trees with the tree root, and obtain a frequent time series pattern based on the confidence level; each of the frequent time series patterns corresponds to a quality difference association rule;

[0046] a poor quality location module, configured to determine the performance data and / or home network data to be searched based on the poor quality data, and locate the poor quality problem based on the performance data and / or home network data and the frequent timing pattern to determine the root cause of the poor quality problem;

[0047] The user identification module is used to identify poor quality users based on the root cause.

[0048] According to the third aspect, an embodiment of the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and runnable on the processor, wherein when the processor executes the program, the steps of the method for locating and delimiting poor quality of home broadband services as described in any one of the above are implemented.

[0049] According to a fourth aspect, an embodiment of the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the above-described methods for locating and delimiting poor quality of home broadband services.

[0050] According to a fifth aspect, an embodiment of the present invention further provides a computer program product, comprising a computer program, which, when executed by a processor, implements the steps of the method for demarcating and locating poor quality of home broadband services as described in any one of the above items.

[0051] The method, device, and apparatus for determining and locating poor quality of home broadband services of the present invention have the following beneficial effects compared to the prior art:

[0052] By aggregating key performance indicator data of the application network according to the network topology of the home broadband service and extracting poor quality data from the aggregated data, since network faults in the home broadband service propagate from the upper layers of the network topology to the lower layers, this method can quickly and accurately demarcate the network layer where the poor quality problem occurs. Subsequently, by chronologically aligning the poor quality data, home network data, and performance data into time series data, a frequent pattern tree is constructed based on the time series data, and frequent time series patterns are extracted from the frequent pattern tree. The resulting frequent time series patterns contain both the temporal logic between network data and the topological relationships between network devices. This setup facilitates fast, efficient, and low-cost location determination based on the poor quality demarcation results, thereby improving the accuracy and efficiency of demarcation and positioning and reducing its cost. This approach is suitable for large-scale end users and can be used for operator network quality testing and rapid fault location. This helps operator broadband operation and maintenance personnel proactively identify poor quality users and carry out targeted problem resolution and repair, reducing user feedback and complaint rates and providing users with a better home broadband service experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0054] Figure 1 A network topology diagram of a home network service network link in the prior art;

[0055] Figure 2 Schematic diagram of the process of the method for locating poor quality of home broadband services of the present invention;

[0056] Figure 3 A service schematic diagram for identifying poor-quality users in the method for demarcating and locating poor-quality home broadband services of the present invention;

[0057] Figure 4 This is a schematic structural diagram of a device for locating and determining poor quality of home broadband services according to the present invention;

[0058] Figure 5 A schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0059] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0060] See also Figure 1 The network link for home broadband services typically consists of terminal devices, routers, switches, ONUs, ODNs, PONs, OLTs, bearer networks, and content source networks. Based on the upper, middle, and lower boundaries of the network topology, these network devices can be further divided into several network layers: the bearer network side, the OLT side, the PON side, and the home side. Because home broadband service networks contain numerous interactive devices, if poor quality issues such as low access, frequent disconnections, and low data rates are detected in user-side applications, operations and maintenance personnel must inspect the relevant devices from the bottom (layer) upwards according to the network topology of the network link, demarcate and locate the poor quality issue, and ultimately find the root cause.

[0061] However, with the increasing number of home broadband users, this bottom-up approach to locating poor quality issues consumes considerable time and manpower. It can also suffer from poor accuracy, leading to a reduced user experience. Therefore, a more cost-effective method for locating poor quality home broadband services is needed, while maintaining accuracy.

[0062] The method for delimiting and locating poor quality of home broadband services provided in this specification can be applied to electronic devices with the ability to process home broadband service communication data, aiming to delimit and locate the problem of poor quality of home broadband services more quickly and accurately, and to enhance the user experience. The electronic device may include a notebook, a desktop computer, a smart phone, a smart wearable device (virtual reality glasses, smart watches, etc.), a tablet computer, etc. Of course, the method for delimiting and locating poor quality of home broadband services provided in this specification can also be applied to applications running in the above-mentioned electronic devices. For example, the control method for autonomous driving can be applied to a browser with the ability to process home broadband service communication data, or it can be applied to software with the ability to process home broadband service communication data.

[0063] See also Figure 2 , Figure 2A flow chart of a method for locating poor quality of home broadband services according to an embodiment of the present invention is shown. The method may include the following steps:

[0064] S101. Acquire real-time network link data of home broadband services. The network link data includes application network key performance indicator (KPI) data, home network data, and performance data.

[0065] In this embodiment, real-time data collection can be performed through various data collection devices in the network link to obtain the above-mentioned application KPI data, home network data and performance data.

[0066] S102. Based on the network topology diagram of the home broadband service, the application KPI data in the network link data is aggregated to obtain aggregated data, and quality difference data is extracted from the aggregated data, wherein the quality difference data is used to reflect network failures and to delimit the network layer where the network failure occurs in the network topology diagram, that is, a quality difference delimitation result can be obtained based on the quality difference data.

[0067] The propagation path of network faults in home broadband services is from the upper layer (layer) to the lower layer (layer) of the network topology diagram, i.e., upper → middle → lower. This means that a fault in a device in the upper layer of the network topology diagram can cause abnormal data flow in devices in the lower layer, ultimately affecting the quality of the home broadband service. The reverse is not true. Therefore, based on this network fault propagation path, once poor quality data is obtained, quality demarcation can be further performed based on whether poor quality data is detected at each network layer.

[0068] Therefore, by aggregating application KPI data according to the network topology diagram, that is, taking the data mean, and then extracting traffic anomaly data, also known as poor quality data, from the aggregated data, quality delimitation can be performed based on the poor quality data. Specifically, the boundary of the poor quality delimitation is the last network layer where the abnormal result occurs. For example, if the data on the OLT side is normal, but the data on the PON side is abnormal, then the poor quality delimitation result is: OLT side; if the poor quality delimitation result is the bearer network side, it is necessary to count the number of applications with abnormal aggregated data. Other delimitation reasons do not need to be processed, and finally the corresponding poor quality delimitation result for each application poor quality record is output.

[0069] S103: Match the poor quality data, home network data, and performance data into time series data in chronological order. In this step, the poor quality data, performance data, and home network data are matched in chronological order using a preset time window, and the ONU name is used as the unique identifier of the matched data.

[0070] Considering that the acquisition cycles corresponding to the multiple data acquisition devices are different, in this embodiment, the data acquisition device with the largest acquisition cycle is used as the corresponding preset time window (time period).

[0071] The data fields corresponding to the quality difference data, home network data, and performance data all include the data generation time and the corresponding network device. Examples of these three types of data are shown in Table 1:

[0072] Table 1 Data examples of some poor quality data, home network data and performance data

[0073]

[0074]

[0075]

[0076] First, based on the quality difference data, the network layer that causes the quality difference problem can be demarcated. Then, the performance data corresponding to the network layer indicated by the quality difference demarcation result during the time period when the quality difference problem occurs and the home network data are extracted. Then, these three types of data are combined into the initial item set h i ={s1,s2,……,s m}, where s k (k=1,2,……,m) is an item in the initial item set, which contains at least one item of data. Then, time attributes are added to the initial item set in chronological order, that is, time constraints are added to the constraints, to form a time series data T={ ,<h2,t2> ,……, <h <h1> m ,t m >}, where t i Indicates h i The time of occurrence and t i <t i+1 , <h k ,t k >(k=1,2,……,m) is an item in the time series data, and the time series data also contains at least one item of data.

[0077] S104. Determine the support of each data item in the time series data according to the preset time span, filter out frequent item data from the time series data according to the support, and establish a frequent pattern tree with the frequent item data as the tree root, wherein the frequent pattern tree established based on the frequent item data will contain at least one frequent item data.

[0078] Since the time constraint is added to the time series data, in this embodiment, the corresponding support and confidence are re-determined to determine the association rules within the time span not exceeding the preset time span E.

[0079] Specifically, the support is calculated as follows:

[0080]

[0081] In formula (1), |A∪B| represents the number of time periods in the time series data T that contain item sets A and B and maintain the order of item sets A and B in the time series database. The time series database is used to store and record all time series data; |D| represents the number of time periods in the time series data T whose time span is the preset time span E.

[0082] The confidence level is calculated as:

[0083]

[0084] In formula (2), |A| represents the number of time periods containing item set A in the time series data T.

[0085] S105. Determine the time periods in which each frequent item in each frequent pattern tree appears, associate each frequent item with the corresponding time period, combine all associated frequent item data in each frequent pattern tree with the tree root, and obtain a frequent temporal pattern based on the confidence level. It is understood that a frequent temporal pattern refers to a set of items that frequently appear within a specific time period.

[0086] A time series database may include multiple sequences, namely time series data. A sequence may also be called a transaction, wherein each time series data may include at least one item set, each item set includes at least one item, and there is a sorting order between the item sets. Multiple item sets arranged in chronological order constitute time series data. In this embodiment, the frequency time series pattern is mined, that is, the item set that appears more than a preset threshold (frequently appears) within a specific time period is mined, which is also a time series pattern. The frequency time series pattern is an ordered arrangement of these different item sets. Therefore, the obtained frequent time series pattern contains both the temporal logic between network data and the topological relationship between network devices. At the same time, each frequency time series pattern also corresponds to a quality difference association rule. According to the quality difference association rule, the root cause of the user quality difference problem can be obtained.

[0087] The quality difference association rule means that frequent item data A can be used to infer frequent item data B under a preset confidence level. That is, the quality difference association rule "A is associated with B" can be understood as "when A occurs, B is also likely to occur." For example, when quality difference B is identified in a certain network topology layer, if situation A in the frequency time series pattern appears at this time, then B can be the root cause.

[0088] Specifically, the frequent item data belonging to the non-empty true subset of the tree root is obtained, and these frequent items are used as initial association rules. The confidence of the initial association rules is then determined, and the initial association rules whose confidence exceeds the preset confidence are used as frequent time series patterns, i.e., quality-poor association rules.

[0089] According to the above logic, in this embodiment, frequent time series patterns are converted into expert library rules and recorded and stored. The accuracy of the expert library rules can be verified by industry experts or professional operation and maintenance personnel, thereby obtaining a positioning rule table. The positioning rule table contains each frequent time series pattern, that is, each quality difference positioning rule. The positioning rule table that includes some of the quality difference positioning rules is shown in Table 2:

[0090] Table 2 Positioning rules table

[0091]

[0092]

[0093] S106. Determine the performance data and / or home network data that need to be found based on the quality difference data, and locate the quality difference problem based on the performance data and / or home network data and the frequent timing pattern to obtain the root cause of the quality difference problem, where the root cause represents the location result of the quality difference problem.

[0094] In this embodiment, after the quality difference delimitation result is obtained through the quality difference data, it will be determined whether the performance data corresponding to these network layers indicated by the quality difference delimitation result and the home network data are consistent with the quality difference association rules corresponding to the frequent timing pattern, and then the quality difference problem is located. Among them, the OLT side quality difference and line quality difference use performance data as the calculation basis, and the home side quality difference uses home side data as the calculation basis. For example, when the quality difference delimitation result is an abnormality on the OLT side, it can be checked from the performance data according to the above-mentioned positioning rule table whether the PON downstream bandwidth utilization or the PON port optical module performance index exceeds the threshold. At this time, if the PON downstream bandwidth utilization exceeds, the quality difference problem is located to PON port congestion. If the PON port optical module performance index exceeds the threshold, the quality difference problem is located to PON port optical module abnormality.

[0095] S107: Identify poor-quality users based on the root cause.

[0096] See also Figure 3According to the quality difference data, it is possible to determine which ONUs, i.e., users, in the home broadband network have quality problems. At the same time, it is possible to determine which network layers in the user's home broadband network have quality problems, i.e., which network layers are abnormal layers. Then, by locating the specific problems, i.e., the root causes, of these abnormal layers, it is possible to identify the poor quality users, the abnormal layers that cause these poor quality users to have quality problems, and the specific root causes of the poor quality problems in the abnormal layers. It can be seen that in this embodiment, quality difference detection can be performed on all terminal users in the home broadband network at the same time, which is suitable for quality difference detection of large-scale terminal users, rather than being limited to certain terminal users.

[0097] The method for demarcating and locating poor quality of home broadband services of the present invention aggregates application network key performance indicator data according to the network topology relationship of the home broadband service and extracts poor quality data from the aggregated data. Since network faults in the home broadband service propagate from the upper layer to the lower layer of the network topology, the network layer where the poor quality problem occurs can be quickly and accurately demarcated. Subsequently, the poor quality data, home network data and performance data are combined into time series data in chronological order, and a frequent pattern tree is constructed based on the time series data, and frequent time series patterns are extracted from the frequent pattern tree. The frequent time series patterns obtained in this way include both the temporal logic between network data and the topological relationship between network devices. Such an arrangement facilitates fast, efficient and low-cost geological poor location based on the quality difference demarcation results, thereby improving the accuracy and efficiency of demarcation and positioning and reducing the cost of demarcation and positioning. The method is suitable for large-scale terminal users and can be used for operator network quality detection and rapid fault location, thereby helping operator broadband operation and maintenance personnel to perceive poor quality users in advance and carry out targeted problem processing and repair, reducing user feedback complaint rate, and providing users with a better home broadband service experience.

[0098] The method may further comprise the following steps:

[0099] S201: Acquire network link data for home broadband services. For details, refer to step S101.

[0100] S2021. Based on the network topology diagram of the home broadband service, and according to the preset window time period, the application KPI data of each network layer is aggregated to obtain aggregated data within several preset window time periods corresponding to each network layer.

[0101] In this embodiment, the aggregated data includes eight data indicators: average round-trip time on the network side, average round-trip time on the user side, average jitter of the round-trip time on the network side, average jitter of the round-trip time on the user side, downlink network side packet loss rate, uplink network side packet loss rate, downlink user side packet loss rate and uplink user side packet loss rate.

[0102] More specifically, the average round-trip time on the network side is calculated as follows:

[0103]

[0104] In formula (3), networkSideAverageRtt ONU Indicates the average round-trip time of the network side of the ONU layer; networkSideAverageRtt i Indicates the average round-trip time on the network side of the i-th node in the ONU layer; count i Indicates the number of KPIs applied by the user KPI of the i-th node in the ONU layer; k represents the total number of nodes in the ONU layer.

[0105] The average round-trip time on the user side is calculated as follows:

[0106]

[0107] In formula (4), userSideAverageRtt ONU Indicates the average round-trip time of the user side of the ONU layer; userSideAverageRtt i Indicates the average round-trip time on the user side of the i-th node in the ONU layer.

[0108] The average round-trip time jitter on the network side is calculated as follows:

[0109]

[0110] In formula (5), networkSideAverageJitter ONU Indicates the average jitter of the network side round trip time of the ONU layer; networkSideAverageJitter i Indicates the average jitter of the network-side round-trip time of the i-th node at the ONU layer.

[0111] The average jitter of the user-side round-trip time is calculated as follows:

[0112]

[0113] In formula (6), userSideAverageJitter ONU Indicates the average jitter of the user-side round-trip time at the ONU layer; userSideAverageJitter i Indicates the average jitter of the user-side round-trip time of the i-th node at the ONU layer.

[0114] The calculation method for average round-trip time and average round-trip time jitter at other network layers of the home broadband service, such as the PON layer and PLT layer, is similar to that at the ONU layer. The average round-trip time or average round-trip time jitter of all nodes in a certain network layer is first aggregated and then divided by the total number of KPI applications for all nodes.

[0115] Since the packet loss rate is relatively sensitive to the value 0, in this embodiment, the packet loss rate data is not directly aggregated by taking the average.

[0116] More specifically, the packet loss rate on the downstream network side is calculated as follows:

[0117]

[0118] In formula (7), flowDownLinkNetworkSideDiscardRatio represents the downlink network side packet loss rate; flowDownLinkNetworkSideDiscardPackets represents the number of downlink network side packet loss; flowDownLinkTotalPackets represents the total number of downlink packet loss.

[0119] The packet loss rate on the upstream network side is calculated as follows:

[0120]

[0121] In formula (8), flowUplinkNetworkSideDiscardRatio represents the uplink network side packet loss ratio; flowUpLinkNetworkSideDiscardPackets represents the number of uplink network side packet loss; and flowUpLinkTotalPackets represents the total number of uplink packet loss.

[0122] The downlink user-side packet loss rate is calculated as follows:

[0123]

[0124] In formula (9), flowDownLinkUserSideDiscardRatio represents the downlink user side packet loss ratio, and flowDownLinkUsersideDiscardPackets represents the number of downlink user side packet loss.

[0125] The uplink user-side packet loss rate is calculated as follows:

[0126]

[0127] In formula (10), flowUpLinkUserSideDiscardRatio represents the uplink user side packet loss rate; flowUpLinkUsersideDiscardPackets represents the number of uplink user side packet loss.

[0128] During the data aggregation process, the eight data indicators mentioned above are aggregated to obtain a feature vector, i.e., aggregated data. The feature vector is an eight-dimensional vector, and the feature vector is:

[0129] ['networkSideAverageRtt','networkSideRttAverageJitter','userSideAverageRtt','userSideRttAverageJitter','flowDownLinkNetworkSideDiscardRatio','flowUplinkNetworkSideDiscardRatio','flowDownLinkUserSideDiscardRatio','flowUplinkUserSideDiscardRatio']

[0130] S2022. Perform traffic anomaly detection on the aggregated data of each network layer from the lower layer to the upper layer of the network topology diagram, and extract poor quality data from the aggregated data of the network layer with poor quality problems.

[0131] Specifically, process anomaly detection is performed starting from the lowest ONU layer. When it is determined that there is poor quality data in the aggregated data of the ONU layer, the poor quality data in the aggregated data of the network layer is extracted, and it is judged whether there is a quality problem in the aggregated data of the upper network layer adjacent to the network layer, namely the PON layer. Similarly, if it is not detected that there is no quality problem in the network layer, the detection can be stopped. If it is not detected that there is a quality problem in the network layer, process anomaly detection is performed upward layer by layer (ONU layer → PON layer → OLT layer → bearer network layer) until the bearer network side is detected.

[0132] Preferably, the aggregated data of each network layer is subjected to flow anomaly detection through a trained isolation forest model to extract poor quality data. Since the characteristics of each network layer of the home broadband service are different, each network layer can correspond to an isolation forest model, which is used to determine whether the aggregated data of the corresponding network layer has flow anomalies, i.e., poor quality problems, and to extract poor quality data. When used specifically, the aggregated data collected and aggregated within a preset window time period is input into the isolation forest model of the corresponding network layer. The corresponding isolation forest model will output 0 or 1, where 0 represents that the aggregated data of the corresponding network layer does not have poor quality problems within the preset window time period of collection, and 1 represents that the aggregated data of the corresponding network layer has poor quality problems within the preset window time period of collection. Subsequently, poor quality data needs to be extracted from the aggregated data.

[0133] The isolation forest model can be trained using historical aggregated data obtained by aggregating historical network link data. By fully collecting offline historical aggregated data as training data, it provides massive data support for subsequent training.

[0134] S203: The quality difference data, home network data, and performance data are combined into time series data in chronological order. The time series data includes at least one item of data, and each item of data has a corresponding occurrence time. For details, refer to step S103.

[0135] S204: Determine the support and confidence of each item in the time series data according to the preset time span, filter out frequent item data from the time series data according to the support, and establish a frequent pattern tree with the frequent item data as the root. For details, refer to step S104.

[0136] S205: Determine the time period in which each frequent item in each frequent pattern tree appears, associate each frequent item with the corresponding time period, and combine all associated frequent item data in each frequent pattern tree with the tree root to obtain a frequent temporal pattern. For details, refer to step S105.

[0137] S206: Determine the performance data and / or home network data to be searched based on the poor quality data, and locate the poor quality problem based on the performance data and / or home network data and the frequent time series pattern to obtain the root cause of the poor quality problem. For details, refer to step S106.

[0138] S207: Identify low-quality users based on the root cause. For details, refer to step S107.

[0139] The method may further comprise the following steps:

[0140] S301: Acquire network link data for home broadband services. For details, refer to step S101.

[0141] S302: Aggregate the key performance indicator data of the application network according to the network topology diagram of the home broadband service to obtain aggregated data, and extract the quality difference data from the aggregated data. For details, refer to step S102.

[0142] S303: The quality difference data, home network data, and performance data are combined into time series data in chronological order. The time series data includes at least one item of data, and each item of data has a corresponding occurrence time. For details, refer to step S103.

[0143] S3041. Divide the time series data into a number of preset time windows of equal length, and treat each event consisting of the time series data in each preset time window as a transaction.

[0144] S3042: traverse the time series data, determine the occurrence probability (number of times) of each data item in the time series data among all things according to the preset time span E, and obtain the support degree of each data item.

[0145] S3043: Item data with support not less than the preset support is regarded as frequent item data, thereby forming a frequent 1-item set F. That is, by setting a minimum support threshold, item data with support not less than the minimum support threshold is regarded as frequent item data.

[0146] S3044. Take each frequent item data as a tree root.

[0147] S3045: Acquire frequent item data that precedes the frequent item data serving as the current tree root in time sequence.

[0148] S3046. Place the frequent item data into the branches of the tree in the order of their occurrence time to generate a frequent pattern tree of the current tree root. In this way, the same number of frequent pattern trees corresponding to the frequent item data can be obtained.

[0149] Assume that s k To build the tree for the root, first scan the item s k All prefixes of the corresponding time in the time series data, that is, the time of occurrence, are located in item s kThe previous items, and during the scanning process, the root path of the frequent item data as the current tree root is scanned, that is, the root node, and the root path that has been scanned is marked to ensure that the root paths of the frequent item data of the current tree root are scanned until all root paths are marked, and the item data that is located before the frequent item data as the current tree root in the time sequence is obtained, and it is determined whether these items belong to the frequent 1-item set F, and the frequent item data belonging to the frequent 1-item set F are extracted, and the frequent item data corresponding to the frequent item data are placed in the branch of the tree according to the order of the time corresponding to the frequent item data, that is, the time of occurrence. The count of the first node of the branch is increased by 1. If there is a node of the corresponding time period in the branch, the count of the node of the corresponding time period is also increased by 1, otherwise a new node is created and the count of the node is 1. If the item s k If the prefix item of is not empty, the above process is repeated recursively. k All prefix items of are scanned, then the item s k Once established, the corresponding frequent pattern tree is also built.

[0150] The calculation method of the support degree may refer to the description of step S103.

[0151] S305: Determine the time periods in which each frequent item in each frequent pattern tree appears. The time periods can be determined by the time when the frequent item occurs. Associate each frequent item with the corresponding time period, and combine all associated frequent item data in each frequent pattern tree with the tree root to obtain a frequent temporal pattern. For details, refer to step S105.

[0152] S306: Determine the performance data and / or home network data to be searched based on the poor quality data, and locate the poor quality problem based on the performance data and / or home network data and the frequent time series pattern to obtain the root cause of the poor quality problem. For details, refer to step S106.

[0153] S307: Identify low-quality users based on the root cause. For details, refer to step S107.

[0154] The method may further comprise the following steps:

[0155] S401: Acquire network link data for home broadband services. For details, refer to step S101.

[0156] S402: Aggregate the key performance indicator data of the application network according to the network topology diagram of the home broadband service to obtain aggregated data, and extract the quality difference data from the aggregated data. For details, refer to step S102.

[0157] S403: The quality difference data, home network data, and performance data are combined into time series data in chronological order. The time series data includes at least one item of data, and each item of data has a corresponding occurrence time. For details, refer to step S103.

[0158] S404: Determine the support and confidence of each item in the time series data according to the preset time span, filter out frequent item data from the time series data according to the support, and establish a frequent pattern tree with the frequent item data as the root. For details, refer to step S104.

[0159] S4051: Determine the time period in which each frequent item data appears in each frequent pattern tree, and associate each frequent item data with the corresponding time period.

[0160] S4052, obtain the root item s in each frequent pattern tree. k The frequent item data of the non-empty true subset is collected, and the frequent items are used as initial association rules, the confidence of the initial association rules is determined, and the initial association rules whose confidence exceeds the preset confidence are used as frequent time series patterns.

[0161] S4053. Verify whether each data item in the frequent time series pattern satisfies the network topology relationship according to the network topology relationship diagram of the home broadband service, and delete the data item that does not satisfy the network topology relationship from the frequent time series pattern.

[0162] S406: Determine the performance data and / or home network data to be searched based on the poor quality data, and locate the poor quality problem based on the performance data and / or home network data and the frequent time series pattern to obtain the root cause of the poor quality problem. For details, refer to step S106.

[0163] S407: Identify low-quality users based on the root cause. For details, refer to step S107.

[0164] The following describes an apparatus provided by an embodiment of the present invention. The apparatus described below and the method described above can refer to each other.

[0165] See also Figure 4 , Figure 4 The following is a schematic diagram showing the structure of a device for locating and locating poor quality of home broadband services according to an embodiment of the present invention. The device may include:

[0166] The data acquisition module 10 is used to obtain real-time network link data of the home broadband service. The network link data includes application network KPI data, home network data and performance data.

[0167] In this embodiment, real-time data collection can be performed through various data collection devices in the network link to obtain the above-mentioned application KPI data, home network data and performance data.

[0168] The quality difference delimitation module 20 is used to aggregate the application KPI data in the network link data according to the network topology diagram of the home broadband service to obtain aggregated data, and extract quality difference data from the aggregated data, wherein the quality difference data is used to reflect network failures and to delimit the network layer where the network failure occurs in the network topology diagram, that is, the quality difference delimitation result can be obtained based on the quality difference data.

[0169] The propagation path of network faults in home broadband services is from the upper layer (layer) to the lower layer (layer) of the network topology diagram, i.e., upper → middle → lower. This means that a fault in a device in the upper layer of the network topology diagram can cause abnormal data flow in devices in the lower layer, ultimately affecting the quality of the home broadband service. The reverse is not true. Therefore, based on this network fault propagation path, once poor quality data is obtained, quality demarcation can be further performed based on whether poor quality data is detected at each network layer.

[0170] By aggregating application KPI data according to the network topology diagram, taking the data mean, and then extracting traffic anomaly data (poor quality data) from the aggregated data, quality delimitation can be performed based on this poor quality data. Specifically, the boundary for poor quality delimitation is the last network layer with an abnormal result. That is, if the data on the OLT side is normal, but the data on the PON side is abnormal, the poor quality delimitation result is: OLT side; if the poor quality delimitation result is the bearer network side, it is necessary to count the number of applications with abnormal aggregated data. Other delimitation reasons do not need to be addressed. Finally, the poor quality delimitation result corresponding to each application poor quality record is output.

[0171] The data matching module 30 is used to match the quality difference data, home network data, and performance data into time series data in chronological order. This process matches the quality difference data, performance data, and home network data in chronological order using a preset time window, and uses the ONU name as the unique identifier of the matched data.

[0172] Considering that the acquisition cycles corresponding to the multiple data acquisition devices are different, in this embodiment, the data acquisition device with the largest acquisition cycle is used as the corresponding preset time window (time period).

[0173] The tree building module 40 is used to determine the support and confidence of each data in the time series data according to a preset time span, and to filter out frequent item data from the time series data according to the support, and to build a frequent pattern tree with the frequent item data as the tree root, wherein the frequent pattern tree built based on the frequent item data will contain at least one frequent item data.

[0174] Since the time constraint is added to the time series data, in this embodiment, the corresponding support and confidence are re-determined to determine the association rules within the time span not exceeding the preset time span E.

[0175] The rule association module 50 is configured to determine the time periods in which each frequent item in each frequent pattern tree appears, associate each frequent item with the corresponding time period, combine all associated frequent item data in each frequent pattern tree with the tree root, and obtain a frequent temporal pattern based on the confidence level. It is understood that a frequent temporal pattern refers to a set of items that frequently appear within a specific time period.

[0176] The quality difference positioning module 60 is used to determine the performance data and / or home network data that needs to be found based on the quality difference data, and locate the quality difference problem based on the performance data and / or home network data and the frequent timing pattern to obtain the root cause of the quality difference problem, wherein the root cause represents the positioning result of the quality difference problem.

[0177] In this embodiment, after the quality difference delimitation result is obtained through the quality difference data, it will be determined whether the performance data corresponding to these network layers indicated by the quality difference delimitation result and the home network data are consistent with the quality difference association rules corresponding to the frequent timing pattern, and then the quality difference problem is located. Among them, the OLT side quality difference and line quality difference use performance data as the calculation basis, and the home side quality difference uses home side data as the calculation basis. For example, when the quality difference delimitation result is an abnormality on the OLT side, it can be checked from the performance data according to the above-mentioned positioning rule table whether the PON downstream bandwidth utilization or the PON port optical module performance index exceeds the threshold. At this time, if the PON downstream bandwidth utilization exceeds, the quality difference problem is located to PON port congestion. If the PON port optical module performance index exceeds the threshold, the quality difference problem is located to PON port optical module abnormality.

[0178] The user identification module 70 is used to identify poor quality users based on the root cause.

[0179] Based on the quality difference data, it is possible to determine which ONUs, i.e., users, in the home broadband network have quality problems. At the same time, it is possible to determine which network layers in the user's home broadband network have quality problems, i.e., which network layers are abnormal layers. Then, by locating the specific problems, i.e., the root causes, of these abnormal layers, it is possible to identify the poor quality users, the abnormal layers that cause these poor quality users to have quality problems, and the specific root causes of the quality problems in the abnormal layers. It can be seen that in this embodiment, quality difference detection can be performed on all terminal users in the home broadband network at the same time, which is suitable for quality difference detection of large-scale terminal users, rather than being limited to certain terminal users.

[0180] The device for demarcating and locating poor quality of home broadband services of the present invention aggregates application network key performance indicator data according to the network topology relationship of the home broadband service and extracts poor quality data from the aggregated data. Since network faults in the home broadband service propagate from the upper layer to the lower layer of the network topology, the network layer where the poor quality problem occurs can be quickly and accurately demarcated. Then, the poor quality data, home network data and performance data are matched into time series data in chronological order, and a frequent pattern tree is constructed based on the time series data, and frequent time series patterns are extracted from the frequent pattern tree. The frequent time series patterns obtained in this way include both the temporal logic between network data and the topological relationship between network devices. Such a setting facilitates fast, efficient and low-cost geological difference positioning based on the quality difference demarcation results, thereby improving the accuracy and efficiency of demarcation positioning and reducing the cost of demarcation positioning. The device is suitable for large-scale terminal users and can be used for operator network quality detection and rapid fault positioning, thereby helping operator broadband operation and maintenance personnel to perceive poor quality users in advance and carry out targeted problem processing and repair, reducing user feedback complaint rate, and giving users a better home broadband service experience.

[0181] Figure 5 An example of a physical structure diagram of an electronic device is shown below. Figure 5 As shown, the electronic device may include: a processor 510 (processor), a communication interface 520 (Communications Interface), a memory 530 (memory) and a communication bus 540, wherein the processor 510, the communication interface 520, and the memory 530 communicate with each other via the communication bus 540. The processor 510 may call the logic commands in the memory 530 to execute the method for locating and delimiting poor quality of home broadband services, which includes:

[0182] Acquire network link data for home broadband services; the network link data includes application network key performance indicator data, home network data, and performance data;

[0183] aggregating the application network key performance indicator data according to a network topology diagram of the home broadband service to obtain aggregated data, and extracting quality difference data from the aggregated data; the quality difference data is used to reflect network failures and to delimit the network layer where the network failure occurs in the network topology diagram;

[0184] Matching the quality difference data, the home network data and the performance data into time series data in chronological order; the time series data includes at least one item of data, and each item of data has a corresponding occurrence time;

[0185] Determining the support and confidence of each item in the time series data according to a preset time span, filtering out frequent item data from the time series data according to the support, and establishing a frequent pattern tree with the frequent item data as a tree root; the frequent pattern tree includes at least one of the frequent item data;

[0186] Determining a time period in which each of the frequent item data in each of the frequent pattern trees appears, associating each of the frequent item data with the corresponding time period, combining all associated frequent item data in each of the frequent pattern trees with the tree root, and obtaining a frequent time series pattern based on the confidence level; each of the frequent time series patterns corresponds to a quality difference association rule;

[0187] Determining the performance data and / or the home network data to be searched based on the poor quality data, and locating the poor quality problem based on the performance data and / or the home network data and the frequent timing pattern to obtain a root cause of the poor quality problem;

[0188] Based on the root cause, poor quality users are identified.

[0189] In addition, the logic instructions in the above-mentioned memory 530 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the 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 enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memor), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0190] On the other hand, the present invention further provides a computer program product, comprising a computer program stored on a non-transitory computer-readable storage medium, wherein the computer program comprises program instructions. When the program instructions are executed by a computer, the computer is capable of performing the home broadband service quality demarcation and positioning method provided by the above methods, the method comprising:

[0191] Acquire network link data for home broadband services; the network link data includes application network key performance indicator data, home network data, and performance data;

[0192] aggregating the application network key performance indicator data according to a network topology diagram of the home broadband service to obtain aggregated data, and extracting quality difference data from the aggregated data; the quality difference data is used to reflect network failures and to delimit the network layer where the network failure occurs in the network topology diagram;

[0193] Matching the quality difference data, the home network data and the performance data into time series data in chronological order; the time series data includes at least one item of data, and each item of data has a corresponding occurrence time;

[0194] Determining the support and confidence of each item in the time series data according to a preset time span, filtering out frequent item data from the time series data according to the support, and establishing a frequent pattern tree with the frequent item data as a tree root; the frequent pattern tree includes at least one of the frequent item data;

[0195] Determining a time period in which each of the frequent item data in each of the frequent pattern trees appears, associating each of the frequent item data with the corresponding time period, combining all associated frequent item data in each of the frequent pattern trees with the tree root, and obtaining a frequent time series pattern based on the confidence level; each of the frequent time series patterns corresponds to a quality difference association rule;

[0196] Determining the performance data and / or the home network data to be searched based on the poor quality data, and locating the poor quality problem based on the performance data and / or the home network data and the frequent timing pattern to obtain a root cause of the poor quality problem;

[0197] Based on the root cause, poor quality users are identified.

[0198] In another aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for determining and locating poor quality of home broadband services provided above is implemented. The method comprises:

[0199] Acquire network link data for home broadband services; the network link data includes application network key performance indicator data, home network data, and performance data;

[0200] aggregating the application network key performance indicator data according to a network topology diagram of the home broadband service to obtain aggregated data, and extracting quality difference data from the aggregated data; the quality difference data is used to reflect network failures and to delimit the network layer where the network failure occurs in the network topology diagram;

[0201] Matching the quality difference data, the home network data and the performance data into time series data in chronological order; the time series data includes at least one item of data, and each item of data has a corresponding occurrence time;

[0202] Determining the support and confidence of each item in the time series data according to a preset time span, filtering out frequent item data from the time series data according to the support, and establishing a frequent pattern tree with the frequent item data as a tree root; the frequent pattern tree includes at least one of the frequent item data;

[0203] Determining a time period in which each of the frequent item data in each of the frequent pattern trees appears, associating each of the frequent item data with the corresponding time period, combining all associated frequent item data in each of the frequent pattern trees with the tree root, and obtaining a frequent time series pattern based on the confidence level; each of the frequent time series patterns corresponds to a quality difference association rule;

[0204] Determining the performance data and / or the home network data to be searched based on the poor quality data, and locating the poor quality problem based on the performance data and / or the home network data and the frequent timing pattern to obtain a root cause of the poor quality problem;

[0205] Based on the root cause, poor quality users are identified.

[0206] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, or of course, by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the existing technology can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or certain parts of the embodiments.

[0207] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for locating poor quality home broadband services, characterized by: The method comprises: Acquire network link data for home broadband services; the network link data includes application network key performance indicator data, home network data, and performance data; aggregating the application network key performance indicator data according to a network topology diagram of the home broadband service to obtain aggregated data, and extracting quality difference data from the aggregated data; the quality difference data is used to reflect network failures and to delimit the network layer where the network failure occurs in the network topology diagram; Matching the quality difference data, the home network data and the performance data into time series data in chronological order; the time series data includes at least one item of data, and each item of data has a corresponding occurrence time; Determining the support and confidence of each item in the time series data according to a preset time span, filtering out frequent item data from the time series data according to the support, and establishing a frequent pattern tree with the frequent item data as a tree root; the frequent pattern tree includes at least one of the frequent item data; Determining a time period in which each of the frequent item data in each of the frequent pattern trees appears, associating each of the frequent item data with the corresponding time period, combining all associated frequent item data in each of the frequent pattern trees with the tree root, and obtaining a frequent time series pattern based on the confidence level; each of the frequent time series patterns corresponds to a quality difference association rule; Determining the performance data and / or the home network data to be searched based on the poor quality data, and locating the poor quality problem based on the performance data and / or the home network data and the frequent timing pattern to obtain a root cause of the poor quality problem; Based on the root cause, poor quality users are identified.

2. The method for locating poor quality of home broadband services according to claim 1, wherein: The process of determining the support and confidence of each item in the time series data according to a preset time span, filtering out frequent item data from the time series data according to the support, and establishing a frequent pattern tree with the frequent item data as the root, specifically includes: Divide the time series data into a number of preset time windows of equal length, and treat each event consisting of the time series data in the preset time window as a transaction; Traversing the time series data, determining the occurrence probability of each data item in the time series data among all things according to the preset time span, and obtaining the support degree of each data item; The item data whose support is not less than the preset support is regarded as frequent item data; A frequent pattern tree is established with the frequent item data as a tree root, to obtain frequent pattern trees having the same number as and corresponding to the frequent item data.

3. The method for locating poor quality of home broadband services according to claim 2, wherein: The step of establishing a frequent pattern tree with the frequent item data as the root to obtain the same number of frequent pattern trees as the frequent item data and corresponding to the frequent item data specifically includes: Taking each of the frequent item data as a tree root; Obtain the frequent item data that is before the frequent item data that is the current tree root in time order; The frequent item data are placed into the branches of the tree in the order of their occurrence time to generate a frequent pattern tree of the current root.

4. The method for locating poor quality home broadband services according to claim 3, wherein: The step of acquiring the frequent item data that precedes the frequent item data serving as the current tree root in time sequence specifically includes: Scan the root path of the frequent item data as the current tree root and mark the scanned root path; Scan the root paths of the frequent item data of the current tree root until all root paths are marked, and obtain the item data that is located before the frequent item data of the current tree root in time sequence; Obtain frequent item data whose support is not less than a preset support in the item data.

5. The method for locating poor quality of home broadband services according to claim 1, wherein: Determining the time period in which each of the frequent item data in each of the frequent pattern trees appears, associating each of the frequent item data with the corresponding time period, combining all associated frequent item data in each of the frequent pattern trees with the tree root, and obtaining a frequent temporal pattern based on the confidence level, specifically includes: Determine the time period in which each frequent item data in each frequent pattern tree appears, and associate each frequent item data with the corresponding time period; Obtain the frequent item data of the non-empty true subset of the tree root in each frequent pattern tree, and use the frequent items as the initial association rules, determine the confidence of the initial association rules, and use the initial association rules whose confidence exceeds the preset confidence as frequent time series patterns; According to the network topology diagram of the home broadband service, it is verified whether each data item in the frequent time series pattern satisfies the network topology relationship, and the data items that do not satisfy the network topology relationship are deleted from the frequent time series pattern.

6. The method for locating poor quality home broadband services according to claim 1, wherein: The aggregating the application network key performance indicator data according to the network topology diagram of the home broadband service to obtain aggregated data, and extracting the quality difference data from the aggregated data specifically includes: According to the network topology diagram, and according to the preset window time period, the application network key performance indicator data of each network layer is aggregated to obtain the aggregated data within several preset window time periods corresponding to each network layer; The aggregated data of each network layer is detected for traffic anomalies from the lower layer to the upper layer of the network topology diagram, and the poor quality data in the aggregated data of the network layer with poor quality problems is extracted.

7. The method for locating poor quality home broadband services according to claim 1, wherein: The calculation formula of the support is: Among them, |A∪B| represents the number of time periods in the time series data that contain item sets A and B and maintain the order of item sets A and B in the time series database, and the time series database is used to store and record all time series data; |D| represents the number of time periods in the time series data with a time span of the preset time span E.

8. The method for locating poor quality home broadband services according to claim 1, wherein: The confidence calculation formula is: Among them, |A∪B| represents the number of time periods in the time series data that contain item sets A and B and maintain the order of item sets A and B in the time series database. The time series database is used to store and record all time series data; |A| represents the number of time periods in the time series data that contain item set A.

9. A device for locating and determining poor quality of home broadband services, characterized by: The device comprises: A data acquisition module is used to acquire network link data of the home broadband service; the network link data includes application network key performance indicator data, home network data and performance data; a quality difference delimiting module for aggregating the application network key performance indicator data based on the network topology diagram of the home broadband service to obtain aggregated data, and extracting quality difference data from the aggregated data; the quality difference data is used to reflect network failures and delimit the network layer where the network failure occurs in the network topology diagram; a data matching module, configured to match the quality difference data, the home network data, and the performance data into time series data in chronological order; the time series data including at least one item of data, and each item of data having a corresponding occurrence time; a tree building module, configured to determine the support and confidence of each data item in the time series data according to a preset time span, filter out frequent item data from the time series data according to the support, and build a frequent pattern tree with the frequent item data as the tree root; the frequent pattern tree includes at least one frequent item data; a rule association module, configured to determine the time period in which each of the frequent item data in each of the frequent pattern trees appears, associate each of the frequent item data with the corresponding time period, combine all associated frequent item data in each of the frequent pattern trees with the tree root, and obtain a frequent time series pattern based on the confidence level; each of the frequent time series patterns corresponds to a quality difference association rule; a poor quality location module, configured to determine the performance data and / or home network data to be searched based on the poor quality data, and locate the poor quality problem based on the performance data and / or home network data and the frequent timing pattern to determine the root cause of the poor quality problem; The user identification module is used to identify poor quality users based on the root cause.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the method for locating and demarcating poor quality of home broadband services as described in any one of claims 1 to 8 are implemented.

Images