Base station clustering method, device, equipment, medium and product

Abstract

The invention provides a base station clustering method and device, equipment, a medium and a product, and belongs to the technical field of wireless communication, and the method comprises the steps: obtaining base station information of a to-be-clustered region, and determining a target clustering number according to the base station information; according to the base station information and the target clustering number, performing base station clustering on the to-be-clustered region based on an improved K-means algorithm to obtain a base station clustering result; wherein the improved K-means algorithm is used for determining the shortest adjacent path distance between every two base stations according to the Wornoi diagram of the to-be-clustered region and carrying out base station clustering. According to the method, the shortest adjacent path distance is determined through the Worronoi diagram, the Euclidean distance of a traditional K-means algorithm is replaced for base station clustering, the base station clustering result is only influenced by a base station topological structure and is not influenced by the base station density, and the clustering efficiency is greatly improved regardless of the base station density. And the shortest adjacent path distance from the center base station of each cluster to the cluster edge base station can be converged to the same range, so that uniform clustering is realized.

Description

Technical Field

[0001] The present application relates to the field of wireless communication technology, and in particular to a base station clustering method, device, equipment, medium and product. Background Art

[0002] With the continuous construction and development of 4G / 5G communication networks, on the one hand, the continuous construction of existing networks has caused the network structure to change continuously, which may lead to conflicts in network resources. For example, the distance between the same frequency and the same physical cell identifier (PCI) is too close, and the tracking area code (TAC) area is too large or too small, which affects the normal communication between the terminal and the base station. It is necessary to optimize the original network resources or reallocate the entire network to optimize network performance; on the other hand, the accelerated deployment of new networks such as Narrow Band Internet of Things (NB-IoT) and 5G lightweight (RedCap) requires reasonable planning of network resources such as PCI and TAC in the early stage of construction.

[0003] Network resource optimization and planning management involve base station data analysis scenarios, which require that network resources be allocated to base stations by clusters and that the base stations be clustered relatively evenly. Relatively even means that the number of base stations included in each cluster is exactly the same or the difference is within a small range.

[0004] Existing base station clustering methods either directly use the traditional K-Means algorithm or improve the initial parameters of the K-Means algorithm. Affected by the base station density factor, the number of base stations in different clusters varies greatly, and the clustering is not uniform. Summary of the invention

[0005] The present application provides a base station clustering method, device, equipment, medium and product, which solves the defects of the prior art that the number of base stations divided into different clusters varies greatly and the clustering is not uniform due to the influence of base station density factors.

[0006] The present application provides a base station clustering method, comprising: Acquire base station information of the area to be clustered, and determine the target number of clusters according to the base station information; According to the base station information and the target number of clusters, base station clustering is performed on the to-be-clustered area based on an improved K-means algorithm to obtain a base station clustering result; The improved K-means algorithm is used to determine the shortest adjacent path distance between any two base stations in the area to be clustered according to the Voronoi diagram of the area to be clustered, and perform base station clustering according to the shortest adjacent path distance.

[0007] As an embodiment, performing base station clustering on the to-be-clustered area based on the improved K-means algorithm according to the base station information and the target number of clusters to obtain a base station clustering result includes: Determine, according to the base station information, the initialization cluster center position corresponding to the target clustering quantity and the Voronoi diagram; Determine, according to the Voronoi graph, an adjacency table of each base station in the area to be clustered; Performing a breadth-first search on each base station in the to-be-clustered area according to the adjacency relationship table to determine the shortest adjacent path distance; The base station information, the target number of clusters, the initialization cluster center point information and the shortest adjacent path distance are used as parameters of the K-means algorithm to obtain the improved K-means algorithm; or, the shortest adjacent path distance is weighted based on a predetermined adjacent path distance weight to obtain a weighted shortest adjacent path distance, and the base station information, the target number of clusters, the initialization cluster center point information and the weighted shortest adjacent path distance are used as parameters of the K-means algorithm to obtain the improved K-means algorithm; The improved K-means algorithm is executed to determine the base station clustering result of the area to be clustered.

[0008] As an embodiment, the step of determining the adjacent path distance weight includes: Determine the number of adjacent base stations in any adjacent layer of any base station based on a breadth-first search algorithm, and determine the average number of adjacent base stations in any adjacent layer according to the number of adjacent base stations; The number of adjacent layers corresponding to the average number of adjacent base stations closest to the preset target value of the number of base stations is used as the expected value of the number of adjacent layers; The adjacent path distance weight corresponding to any base station is determined according to the number of adjacent base stations corresponding to the expected value of the number of adjacent layers and the average number of adjacent base stations.

[0009] As an embodiment, executing the improved K-means algorithm to determine the base station clustering result of the area to be clustered includes: Traversing the base stations in the area to be clustered, updating the cluster to which each base station belongs according to the adjacency relationship table and the shortest adjacent path distance / the weighted shortest adjacent path distance, and performing clustering correction on isolated clustered base stations; Update the cluster center point information of each cluster, determine whether the updated cluster center information meets the preset clustering convergence condition, if yes, combine the updated clusters to obtain the base station clustering result, otherwise, return to execute the traversal of the base stations in the to-be-clustered area until the preset clustering convergence condition is met or the preset number of cycles is reached.

[0010] As an embodiment, if the parameters of the improved K-means algorithm include the weighted shortest adjacent path distance, before updating the cluster center point information of each cluster, it also includes: Determining the number of base stations in any cluster; If the number of base stations in a certain cluster is greater than a preset target number of base stations, increase the adjacent path distance weight corresponding to the cluster center base station according to the number of base stations, the target number of base stations and a preset control coefficient; If the number of base stations in a certain cluster is less than a preset base station number target value, the adjacent path distance weight corresponding to the cluster center base station is reduced according to the number of base stations, the base station number target value and a preset control coefficient.

[0011] As an embodiment, the base station information includes the latitude and longitude coordinates of the base station, and correspondingly, determining the initialization cluster center position corresponding to the target clustering quantity according to the base station information includes: Constructing a record set and a merge set, storing the longitude and latitude coordinates of all the base stations in the record set, and leaving the merge set empty; Circularly merging the latitude and longitude coordinates of the base stations in the record set according to the merged set to determine the initialization cluster center position; The number of initialized cluster centers is the same as the number of target clusters, and the merged set is used to store the coordinates generated in each merging process.

[0012] The present application also provides a base station clustering device, including: A determination module, configured to obtain base station information of a to-be-clustered area and determine a target number of clusters according to the base station information; A clustering module, configured to perform base station clustering on the to-be-clustered area based on the improved K-means algorithm according to the base station information and the target number of clusters, to obtain a base station clustering result; The improved K-means algorithm is used to determine the shortest adjacent path distance between any two base stations in the area to be clustered according to the Voronoi diagram of the area to be clustered, and perform base station clustering according to the shortest adjacent path distance.

[0013] The present invention also provides 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 computer program, any of the base station clustering methods described above is implemented.

[0014] The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the base station clustering method described above is implemented.

[0015] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements any of the above base station clustering methods.

[0016] The base station clustering method, device, equipment, medium and product provided by the present application, the Voronoi diagram belongs to the space segmentation algorithm, the Voronoi diagram is used to determine the shortest adjacent path distance between each base station in the area to be clustered, and the shortest adjacent path distance is used to replace the Euclidean distance of the traditional K-means algorithm to perform base station clustering. The clustering result is only affected by the base station topology structure and will not be affected by the base station density. Regardless of the base station density, the shortest adjacent path distance from the central base station of each cluster to the edge base station of the cluster converges to the same range, so that the number of base stations contained in each cluster is in the same range, and finally uniform clustering is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the present application or the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 a- Figure 1 f is a flowchart of the traditional K-Means algorithm.

[0019] Figure 2 It is a schematic diagram of an embodiment of using the traditional K-Means algorithm to perform base station clustering.

[0020] Figure 3 This is one of the flow charts of the base station clustering method provided in this application.

[0021] Figure 4 This is the second flowchart of the base station clustering method provided in this application.

[0022] Figure 5 It is a schematic diagram of a process for determining the number of target clusters provided in this application.

[0023] Figure 6 It is a flow chart of a preferred embodiment of initializing the cluster center point position provided by the present application.

[0024] Figure 7 It is a flowchart of a preferred embodiment of determining the shortest adjacent path distance provided by the present application.

[0025] Figure 8 It is a schematic diagram of the Voronoi diagram provided in this application.

[0026] Fig. 9 This application provides Figure 7 Schematic diagram of the process of step S303.

[0027] Fig.10 It is a schematic diagram of the shortest adjacent path distance provided by this application.

[0028] Fig.11 It is a flowchart of a preferred embodiment of determining the adjacent path distance weight provided by the present application.

[0029] Fig.12 It is a flowchart of a preferred embodiment of updating base station clustering provided by the present application.

[0030] Fig.13 It is a schematic diagram of base station clustering results of a preferred embodiment provided by the present application.

[0031] Fig.14 It is a structural schematic diagram of a base station clustering device provided in this application.

[0032] Fig.15 It is a schematic diagram of the structure of the electronic device provided by this application. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical solutions and advantages of this application clearer, the technical solutions in this application will be clearly and completely described below in conjunction with the drawings in this application. Obviously, the described embodiments are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0034] It should be noted that all actions to obtain signals, information or data in this application are carried out in compliance with the relevant data protection laws and policies of the location and with the authorization given by the owner of the corresponding device.

[0035] Existing base station clustering methods applicable to wireless network topology structures include directly using a traditional K-Means algorithm or using an improved K-Means algorithm to cluster base stations to obtain base station clustering results.

[0036] The specific steps of directly using the traditional K-Means algorithm to cluster base stations to obtain base station clustering results include: preparing base station data, the base station data includes the base station number T, the number of base stations S, the base station number, the longitude of the base station, and the latitude of the base station in each cluster. The number of clusters is calculated based on the base station data, and the number of clusters K=Int(S / T)+1, where Int is a rounding function. The number of clusters K and the longitude and latitude data of the base stations are used as parameters of the traditional K-Means algorithm, and the K-Means algorithm is executed to obtain the base station clustering results.

[0037] Figure 1 a- Figure 1 f is a flow chart of the traditional K-Means algorithm, such as Figure 1 a- Figure 1 As shown in Figure 5, the K-Means algorithm is a commonly used distance-based clustering algorithm. Its basic idea is as follows: 1. Based on the data set to be clustered, randomly select K data samples as clustering centers, such as Figure 1 As shown in b; 2. Traverse and calculate the clustering affiliation of all data samples. Specifically, for sample i, calculate its Euclidean distance with the K cluster centers respectively, and update its cluster to be consistent with the cluster center closest to it, such as Figure 1 c. Figure 1 As shown in e; 3. Update the cluster center. Specifically, traverse all k clusters, and for each cluster, take the average coordinate of all data samples as the new cluster center, such as Figure 1 As shown in d; 4. Repeat the above 2 and 3 processes until the center positions of the K clusters converge to a certain range or the maximum number of cycles is reached, thus obtaining the final clustering result, such as Figure 1 f.

[0038] The traditional K-Means algorithm is used to cluster base stations. On the one hand, this method completely randomly selects the initial cluster center point position, which easily causes the initial cluster center point position to aggregate, resulting in low algorithm efficiency. On the other hand, this method is greatly affected by the base station density factor, which easily causes large differences in the number of base stations divided into different clusters.

[0039] In actual network scenarios, the density distribution of base stations is uneven. For example, the base station density in urban areas is much higher than that in rural areas. However, when updating the clustering of base stations, the traditional K-Means algorithm uses Euclidean distance to calculate the distance from the base station to the cluster center, and updates the cluster to which the base station belongs based on the cluster number of the cluster center closest to the base station. This calculation rule determines that the common boundary of two adjacent clusters must be at the same distance from the two cluster centers. Furthermore, a cluster with a higher base station density must contain more base stations.

[0040] Figure 2 FIG. 1 is a schematic diagram of an embodiment of using the traditional K-Means algorithm to perform base station clustering. Figure 2 As shown in the figure, the distance a from the center position to the boundary of cluster 13 is the same as the distance b from the center position to the boundary of cluster 0. However, due to the different base station densities, the number of base stations in the two clusters is quite different.

[0041] Table 1: Base station clustering results based on the traditional K-Means algorithm

[0042] The above table is Figure 2 From the corresponding statistical table of the number of base stations divided into each cluster, it can be seen that the number of base stations in different clusters varies greatly.

[0043] The core idea of ​​clustering base stations using the improved K-Means algorithm is to calculate the location and number of cluster centers that meet the conditions, and then use them as the parameters of the traditional K-Means algorithm to execute the K-Means algorithm to obtain the clustering results. This method is not a thorough improvement on the traditional K-Means algorithm and still has many shortcomings. First, the base station clustering results are solidified. The location and number of the initial cluster centers are calculated using a strict formula, which completely abandons randomness and leads to the solidification of clustering results. Second, the problem of uneven clustering of base stations is not effectively solved. This solution only improves the initial parameters of the K-Means algorithm. There is no improvement in the calculation formula and process intervention during the calculation of the K-Means loop iteration. Therefore, the shortcomings of the traditional K-Means algorithm still exist. It is greatly affected by the base station density factor, which easily leads to large differences in the number of base stations divided into different clusters.

[0044] The base station clustering method, device, equipment, medium and product provided in the present application are described in detail below with reference to the accompanying drawings.

[0045] Figure 3 This is one of the flow charts of the base station clustering method provided in this application, such as Figure 3 As shown, the present application provides a base station clustering method, comprising the following steps.

[0046] Step S100, obtaining base station information of the area to be clustered, and determining the target number of clusters according to the base station information. Specifically, obtaining base station information of the area to be clustered means obtaining all base station information of the area to be clustered, that is, information of all base stations, the base station information includes base station number and base station location information, and the location information can be in the form of latitude and longitude or GPS data or base station identifier, etc.

[0047] Step S200 , performing base station clustering on the to-be-clustered area based on the improved K-means algorithm according to the base station information and the target number of clusters, to obtain a base station clustering result.

[0048] The improved K-means algorithm is used to determine the shortest adjacent path distance between any two base stations in the area to be clustered according to the Voronoi diagram of the area to be clustered, and perform base station clustering according to the shortest adjacent path distance.

[0049] It can be understood that the present application performs data analysis based on the topological structure to achieve uniform clustering, which is significantly different from the clustering method based on Euclidean distance. For wireless network scenarios, the location distribution of base stations is uneven. Using Euclidean distance to cluster base stations, in areas with high base station density, the number of base stations contained in the cluster must be large. The present application characterizes the distance between two base stations with the shortest adjacent path. This distance characteristic is only affected by the topological structure of the base station, but not by the density of the base station. For wireless network scenarios, the base station coordinates of an area are enlarged and reduced, and the base station density will change significantly, but its topological structure and base station adjacency relationship do not change, which shows that the base station adjacency relationship is related to the base station topology and is not related to the base station density. The present application uses the shortest adjacent path distance calculated based on the topological structure for the distance calculation from the base station to the cluster center in the K-Means algorithm, and then clusters. Regardless of the density of the base stations, the shortest adjacent path distance from the central base station of each cluster to the edge base station of the cluster converges to the same range, so that the number of base stations contained in each cluster is in the same range, and uniform clustering is finally achieved.

[0050] Based on the above embodiment, as an optional embodiment, the base station clustering is performed on the to-be-clustered area based on the improved K-means algorithm according to the base station information and the target number of clusters to obtain the base station clustering result, including the following steps.

[0051] According to the base station information, the initialization cluster center position corresponding to the target clustering quantity and the Voronoi diagram are determined.

[0052] According to the Voronoi diagram, an adjacency relationship table of each base station in the area to be clustered is determined.

[0053] A breadth-first search is performed on each base station in the to-be-clustered area according to the adjacency relationship table to determine the shortest adjacent path distance.

[0054] The base station information, the target number of clusters, the initialization cluster center point information and the shortest adjacent path distance are used as parameters of the K-means algorithm to obtain the improved K-means algorithm; or, the shortest adjacent path distance is weighted based on a predetermined adjacent path distance weight to obtain a weighted shortest adjacent path distance, and the base station information, the target number of clusters, the initialization cluster center point information and the weighted shortest adjacent path distance are used as parameters of the K-means algorithm to obtain the improved K-means algorithm.

[0055] The improved K-means algorithm is executed to determine the base station clustering result of the area to be clustered.

[0056] It can be understood that the present application determines the shortest adjacent path distance between each pair of base stations in the area to be clustered through the Voronoi diagram, and uses the shortest adjacent path distance to replace the Euclidean distance of the traditional K-means algorithm to cluster the base stations. The clustering result is only affected by the base station topology structure and will not be affected by the base station density. This solves the problem that the prior art is affected by the base station density factor, resulting in large differences in the number of base stations divided into different clusters and uneven clustering.

[0057] The present application provides two different improved K-means algorithms. One is to directly use the shortest adjacent path distance as the distance parameter of the traditional K-means algorithm, and to perform iterative clustering using the shortest adjacent path distance according to the clustering process of the traditional K-means algorithm. The other is to weight the shortest adjacent path distance and use the weighted shortest adjacent path distance as the distance parameter of the traditional K-means algorithm to accelerate the convergence speed. The second improved K-means algorithm is described below in conjunction with the accompanying drawings.

[0058] Figure 4 This is the second flow chart of the base station clustering method provided by this application. Figure 3 Step S100 in the embodiment includes Figure 4 Step S1 in Figure 3 Step S200 in the embodiment includes Figure 4 Steps S2 to S9 in Figure 4 As shown, the preferred embodiment provided by the present application includes: S1, calculating the target number of clusters; S2, initializing the location of the cluster center; S3, calculating the shortest adjacent path distance between two base stations based on the Voronoi diagram; S4, calculating the base station adjacent path distance weight; S5, traversing the base stations, and updating the base station clustering according to the adjacency relationship table and the weighted shortest adjacent distance; S6, correcting the base station adjacent path distance weight; S7, updating the location of the cluster center; S8, judging the clustering convergence condition; S9, outputting the base station clustering result.

[0059] Figure 5 is a flow chart of determining the number of target clusters provided by this application, such as Figure 5 As shown, obtaining base station information of a to-be-clustered area and determining a target clustering quantity according to the base station information includes the following steps.

[0060] S101: Acquire base station information in a to-be-clustered area.

[0061] The base station information data of the area to be clustered is obtained and recorded, including the base station number, base station type, base station longitude, and base station latitude information. The base station type is not limited to macro base station, indoor base station, and small base station. One of the records in the base station information record table is shown in Table 2.

[0062] Table 2: Example of base station information record table

[0063] According to the base station information record table, count the total number of base stations in the area to be clustered S site , that is, the total number of base stations in the area to be clustered.

[0064] Set the target value N for the number of base stations contained in a single cluster site , N site It is related to the specific purpose of clustering and needs to be set according to the usage scenario. For example, the purpose of clustering is used in the scenario of 5G PCI planning. There are 1008 PCIs in 5G, and 1 PCI is planned for each 5G cell. In the case of same-frequency networking, 1008 PCIs can be divided into 4 groups, and then the four-color theorem is used to allocate PCIs to specific clusters to improve the reuse distance of the same frequency and PCI. Each cluster has a maximum of 252 cells. Considering 3 cells per base station, the maximum number of base stations that can be accommodated in each cluster is 84. Considering the adjustment margin, the number of base stations in each cluster can be set to about 75. (Note: This 5G PCI planning is only used as an example to illustrate that a base station clustering is listed, and there is an actual demand scenario with requirements for the number of base stations in the cluster).

[0065] S102: Calculate the number of target clusters.

[0066] According to the number of base stations S site and the target number of base stations contained in a single cluster N site , calculate the number of target clusters N cluster , the calculation formula is as follows: ; in, It refers to the number of base stations divided by the target number of base stations and rounded up.

[0067] Based on the above embodiment, as an optional embodiment, the base station information includes the longitude and latitude coordinates of the base station, and correspondingly, the initialization cluster center position corresponding to the target clustering quantity is determined according to the base station information, including.

[0068] Construct a record set and a merge set, store the longitude and latitude coordinates of all base stations in the record set, and empty the merge set.

[0069] The latitude and longitude coordinates of the base stations in the record set are cyclically merged according to the merged set to determine the initialization cluster center position.

[0070] The number of initialized cluster centers is the same as the number of target clusters, and the merged set is used to store the coordinates generated in each merging process.

[0071] Figure 6 is a flow chart of a preferred embodiment of initializing the cluster center point position provided by the present application, such as Figure 6As shown, this application uses a cyclic merging method to obtain the initialization N based on the latitude and longitude coordinates of all base stations. cluster The location of the cluster center point includes the following steps.

[0072] S201, initializing cluster center point record set and merge set.

[0073] Use the longitude and latitude coordinates of all base stations to establish a cluster center point record set, such as {(lon1, lat1), (lon2, lat2), (lon3, lat3), ..., (lonN, latN)}. The number of coordinates contained in the initialized record set is equal to the number of base stations S. site Reuse the table structure of the record set to build a merged set and set the merged set to empty.

[0074] S202, cyclically merge the longitude and latitude coordinates of the base station, and site The longitude and latitude coordinates of the base stations are executed site -N cluster Merge times to get N cluster Longitude and latitude coordinates.

[0075] S2021. Determine the number of data items in the record set: if the number of data items is greater than 1, execute step S2022; if the number of data items is equal to 1, execute step S2023; if the number of data items is equal to 0, execute step S2024.

[0076] S2022. Randomly extract a base station longitude and latitude coordinate from the record set, select the base station longitude and latitude coordinate that is closest to it from the record set, merge the two base station longitude and latitude coordinates according to the average longitude and latitude, form a new base station longitude and latitude coordinate and store it in the merged set, delete the two base station longitude and latitude coordinates in the record set, and execute step S2025.

[0077] S2023. Take the longitude and latitude coordinates of a base station in the record set, select the longitude and latitude coordinates of the base station closest to it from the merged set, merge the longitude and latitude coordinates of the two base stations according to the average value of the longitude and latitude to form a new coordinate and store it in the merged set, delete the longitude and latitude coordinates of the two base stations in the record set and the merged set, and execute step S2025.

[0078] S2024. Transfer the longitude and latitude coordinates of the base stations in the merged set to the record set, clear the merged set, and execute step S2022.

[0079] S2025. Record the cumulative number of merges and determine whether the cumulative number of merges reaches S. site -N cluster , when the cumulative number of merges reaches S site -N cluster Exit the loop when , otherwise continue to execute step S2021.

[0080] S203: Obtain the position of the initialization cluster center point.

[0081] After exiting the loop, the longitude and latitude coordinates of the base stations in the record set and the merged set are combined to obtain N cluster The latitude and longitude coordinates of the base stations, N cluster The longitude and latitude coordinates of the base stations are used as the initial cluster center locations.

[0082] In step S2022 and step S2023, the latitude and longitude coordinates of the nearest base station can be selected by traversing and comparing distances or using a KD-Tree algorithm.

[0083] It is understood that this application is in accordance with S site -N cluster The latitude and longitude coordinates of base stations are randomly selected for a fixed number of times and merged with the latitude and longitude coordinates of the nearest base station. By setting the record set and merge set, it is ensured that the latitude and longitude coordinates of a base station are only merged once in the same round of merging, avoiding the latitude and longitude coordinates of a certain base station from being merged frequently, achieving average merging, and finally generating N cluster The traditional K-Means algorithm completely randomly extracts the initial cluster center point position. This application retains randomness and avoids the solidification of base station clustering results. The generated initial cluster center point position is relatively evenly distributed in the entire area, which is more conducive to improving the clustering effect and making the clustering more uniform.

[0084] Figure 7 is a flow chart of a preferred embodiment of determining the shortest adjacent path distance provided by the present application, such as Figure 7 As shown, based on the above embodiments, as an optional embodiment, the present application uses the shortest adjacent path distance to characterize the distance between two base stations, so that the base station topology structure, especially the base station adjacency relationship, is used as the basis for base station aggregation clustering, which is the key to achieving uniform clustering.

[0085] The steps of determining the shortest adjacent path distance include the following steps.

[0086] S301, generate a Voronoi map based on the latitude and longitude coordinates of all base stations, and perform spatial segmentation on the clustering area. The Voronoi map is as follows: Figure 8 shown.

[0087] S302: Calculate and record a base station adjacency relationship table.

[0088] Based on the Voronoi graph, the first layer adjacency relationship of each base station is traversed and calculated to form an adjacency relationship table, and an example of the adjacency relationship table is shown in Table 3. The present application does not limit the method of generating the adjacency relationship table, and those skilled in the art can make a customized selection according to the needs, such as constructing a Voronoi graph by the Delaunay triangulation method, importing the Voronoi graph into a spatial database such as PostGIS, and quickly realizing the polygon adjacency query in the Voronoi graph.

[0089] Table 3: Adjacency table

[0090] S303: Calculate the shortest adjacent path distances between any two base stations.

[0091] Calculate the shortest adjacent path distance L between base stations i and j ij When , you can arbitrarily select base station i or base station j as the root node, traverse from the root base station, visit all adjacent base stations, and for each adjacent base station, visit the adjacent base station, and so on, until the target base station is found. The number of layers traversed during the search process is the shortest adjacent path distance.

[0092] In the embodiment of the present application, a breadth-first search algorithm is used to calculate the shortest adjacent path distance, taking the calculation of the shortest adjacent path distance between base stations i and j as an example.

[0093] S3031, prioritize table lookup to obtain the shortest adjacent path distance. In the entire process of base station clustering, it is not necessary to use the shortest adjacent path distances of all base stations, so an inert calculation scheme is adopted in the embodiment of the present application. Specifically, a distance record table is set. When it is necessary to calculate the shortest adjacent path distance between two base stations, the table is prioritized for querying to obtain the result. If no result is found, it is calculated through steps S3032 to S3037.

[0094] S3032. Take base station i as the root base station and initialize a search record table. The format of the search record table is shown in Table 4.

[0095] Table 4: Search record table

[0096] S3033. Traverse all untraversed base stations in the search record table. That is, for base station m, if the adjacent base station traversal status in the search record table is no, query the adjacent relationship table generated in step S302 and traverse all adjacent base stations of base station m.

[0097] S3034. Determine whether the adjacent base station n of base station m belongs to the search record table. If base station n belongs to the search record table, it means that base station n has been analyzed, and then return to step S3033 to continue traversing the next adjacent base station n+1 of base station m; if base station n does not belong to the search record table, it means that the base station has not been analyzed and is a base station outside the area of ​​all base stations in the search record table, and then execute step S3035.

[0098] S3035. Put base station n into the search record table. The search layer number of base station n is equal to the search layer number of base station m plus 1.

[0099] S3036. Determine whether the search record table contains base station j. If the search record table does not contain base station j, it indicates that base station j is not found, and return to step S3033 to continue traversing the next adjacent base station n+1 of base station m. If the search record table contains base station j, it indicates that base station j is found, and execute step S3037.

[0100] S3037. Calculate and obtain the shortest adjacent path distance L between base stations i and j, where L is equal to the number of search layers of base station j.

[0101] S3038. The calculation result of the shortest adjacent path distance between base stations i and j is stored in a distance record table so that the table can be looked up next time the shortest adjacent path distance between base stations i and j is calculated, thereby improving calculation efficiency.

[0102] The shortest adjacent path distance is independent of the path direction. Therefore, when calculating the shortest adjacent path distance between base stations i and j, you can arbitrarily select base station i or base station j as the root node, search for base station j from base station i, and search for base station i from base station j. The two calculation results are consistent. Fig.10 As shown in the figure, base station A and base station B are in an adjacent relationship, and the shortest adjacent path distance is 1; base station A and base station C have no adjacent relationship, but have a common adjacent base station. Starting from base station A to the adjacent base station and then to base station C, after 2 layers of search, the shortest adjacent path distance is 2; by analogy, the shortest adjacent path distance between base station A and base station D is 3.

[0103] It can be understood that the Voronoi diagram belongs to a space segmentation algorithm. The present application calculates the adjacency topological relationship between base stations based on the Voronoi diagram, and then calculates the shortest adjacent path distance between each base station through a breadth-first search algorithm, and replaces the Euclidean distance algorithm in the traditional K-Means algorithm with this distance algorithm, effectively shielding the influence of base station density on clustering.

[0104] Based on the above embodiment, as an optional embodiment, the step of determining the adjacent path distance weight includes the following steps.

[0105] The number of adjacent base stations in any adjacent layer of any base station is determined based on a breadth-first search algorithm, and the average number of adjacent base stations in any adjacent layer is determined according to the number of adjacent base stations.

[0106] The number of adjacent layers corresponding to the average number of adjacent base stations that is closest to the preset target value of the number of base stations is used as the expected value of the number of adjacent layers.

[0107] The adjacent path distance weight corresponding to any base station is determined according to the number of adjacent base stations corresponding to the expected value of the number of adjacent layers and the average number of adjacent base stations.

[0108] The distribution of base stations determines the topological structure of its Voronoi diagram, which further affects the number of adjacent base stations of each base station. The number of adjacent base stations of a base station is not caused by the density of base stations, that is, the number of adjacent base stations of a base station is not related to the density of base stations. The coordinates of base stations in an area are uniformly affine transformed, that is, translated, rotated, enlarged, and reduced. The topological structure of its Voronoi diagram will not change before and after the transformation, and the number of adjacent base stations of the base station will not change, but its base station density will change with the enlargement and reduction operations. There is no obvious difference in the number of adjacent base stations of a base station in areas with high base station density and areas with low base station density.

[0109] This application proposes a method for calculating the weight of base station adjacent path distance to control the influence of network topology factors on cluster size. The adjacent path distance weight is calculated based on the number of adjacent base stations of the base station. The number of adjacent base stations of the base station is not related to the base station density. The adjacent path distance weight provided by this application is essentially different from the base station density weight.

[0110] Fig.11 : is a flowchart of a preferred embodiment of determining the distance weight of adjacent paths provided by the present application, such as Fig.11 As shown, the determination of the adjacent path distance weight includes the following steps.

[0111] S401: Calculate the average number of adjacent base stations.

[0112] Base station i is searched for adjacent base stations in layer j by breadth-first search algorithm, and the number of base stations found is recorded as , the average number of adjacent base stations for all base stations is as follows: in, The value of is the number of base stations S site .

[0113] The loop calculation starts from j=1 The value of j is incremented by 1 each time the loop is looped, and step S402 determines whether to exit the loop.

[0114] S402: Determine whether the average number of adjacent base stations meets the convergence condition, and calculate the expected value of the number of adjacent layers.

[0115] When the search reaches j=m, the average number of adjacent base stations The value of is closest to the target value N of the number of base stations contained in a single cluster site , that is, it satisfies the following formula: , If the average number of adjacent base stations meets the above convergence condition, m is recorded as the expected number of adjacent layers of the cluster center base station, the loop is exited, and step S403 is continued; if the convergence condition is not met, step S401 is returned to be executed.

[0116] S403: Calculate the adjacent path distance weight of each base station.

[0117] According to the expected value m of the number of adjacent layers of the cluster center base station, the adjacent path distance weights of all base stations are calculated, where the adjacent path distance weight calculation formula of base station i is: in, is the adjacent path distance weight of base station i.

[0118] Based on the above embodiment, as an optional embodiment, executing the improved K-means algorithm to determine the base station clustering result of the area to be clustered includes the following steps.

[0119] The base stations in the to-be-clustered area are traversed, the clusters to which each base station belongs are updated according to the adjacency relationship table and the shortest adjacent path distance / the weighted shortest adjacent path distance, and clustering correction is performed on isolated base stations in the clusters.

[0120] Update the cluster center point information of each cluster, determine whether the updated cluster center information meets the preset clustering convergence condition, if yes, combine the updated clusters to obtain the base station clustering result, otherwise, return to execute the traversal of the base stations in the to-be-clustered area until the preset clustering convergence condition is met or the preset number of cycles is reached.

[0121] Optionally, if the parameters of the improved K-means algorithm include the weighted shortest adjacent path distance, before updating the cluster center point information of each cluster, the following steps are also included.

[0122] Determine the number of base stations in any cluster.

[0123] If the number of base stations in a certain cluster is greater than a preset target number of base stations, the adjacent path distance weight corresponding to the cluster center base station is increased according to the number of base stations, the target number of base stations and a preset control coefficient.

[0124] If the number of base stations in a certain cluster is less than a preset base station number target value, the adjacent path distance weight corresponding to the cluster center base station is reduced according to the number of base stations, the base station number target value and a preset control coefficient.

[0125] Fig.12 is a flow chart of a preferred embodiment of updating base station clustering provided by the present application, such as Fig.12 As shown, the improvement of the K-Means algorithm in the present application involves the improvement of the calculation method of the cluster to which the sample points belong, and includes the following steps.

[0126] S501. Calculate a cluster center base station according to the location of the cluster center point.

[0127] For N cluster The coordinates of the cluster center points are calculated by the nearest neighbor algorithm. cluster Base stations, each cluster center point coordinate corresponds to a nearest base station, the nearest base station is recorded as the cluster center base station, a total of N cluster cluster center base station.

[0128] S502: Update the cluster ID of the base station.

[0129] Traverse all base stations and calculate the number of base stations i to N respectively. cluster The weighted adjacent path distance of the cluster center base station is taken as the minimum distance L i , and update the cluster number of base station i to the cluster number of the cluster center base station corresponding to the minimum distance.

[0130] , k∈[1,N cluster ] in, is the adjacent path distance weight of base station k, L ik is the shortest adjacent path distance from base station i to the kth cluster center base station.

[0131] S503: Modify clustering of isolated base stations.

[0132] For N cluster Starting from the cluster center base station, the breadth-first search algorithm is used to search for base stations in the cluster. If a base station in the cluster can be searched, it means that it has an adjacent relationship with the cluster center base station, otherwise the station is an isolated base station in the cluster.

[0133] For clustered isolated base stations, their Li values ​​are traced back, the next smallest value is taken, clustering is performed again, and the cluster center base station to which they belong is updated to determine whether they have an adjacent relationship with the cluster center base station. If so, the current round of clustering of the base station is completed. If not, the Li value is traced back again and the base station clustering is corrected until all base stations have an adjacent relationship with the corresponding cluster center base station.

[0134] After updating the base station belonging cluster, the number of base stations in each cluster is counted. For cluster k, the number of base stations is N. k , modify the adjacent path distance weight of the cluster center base station k according to the following rules, where β is the control coefficient, and the value range is β∈[0.01,0.1].

[0135] in, is the corrected adjacent path distance weight of base station k.

[0136] Step S7 is to update the cluster center point position, using the average of the longitude and latitude coordinates of all base stations in the cluster as the new center point coordinates.

[0137] Compare the change range of the cluster center position before and after the update in step S7 as the clustering convergence judgment condition. cluster After all cluster center positions are updated, calculate their corresponding N cluster cluster center base stations, and N before update cluster The cluster center base stations are compared. If all the corresponding cluster center base stations before and after the update are consistent, or the shortest adjacent path distances of all the corresponding cluster center base stations before and after the update are ≤1, the clustering result converges. If the clustering result does not converge, jump to step S5 to continue the K-Means main loop. If the number of K-Means main loops reaches the maximum, the loop is forced to exit.

[0138] When the clustering results converge, or the number of K-Means cycles reaches the maximum value, the process goes to step S9 to output the base station clustering results.

[0139] Fig.13 is a schematic diagram of base station clustering results in a preferred embodiment provided by the present application, such as Fig.13 As shown, the present application adopts a base station clustering method based on the Voronoi diagram and the K-means algorithm to perform base station clustering. There are 1611 base stations in the example area, 101 target base stations in each cluster, 16 clustering targets, a maximum number of clustered base stations of 106 and a minimum number of 96. Compared with the traditional K-Means algorithm, there is a significant improvement in the problem of uneven base station clustering.

[0140] Table 5 Base station clustering results

[0141] In summary, this application proposes a general uniform clustering method for location coordinate sample data in geographic analysis and spatial analysis scenarios. The typical application scenario is for 4 / 5G wireless network base station clustering to improve the efficiency of network resource planning and optimization. With the development and evolution of the network, the method of this application will also be able to be applied to new network base station clustering such as 6G in the future, and also to other scenarios with uniform clustering requirements, such as geographic analysis software, integrating the uniform clustering function designed in this application to enrich the software analysis function.

[0142] The base station clustering device provided in the present application is described below. The base station clustering device described below and the base station clustering method described above can be referred to each other.

[0143] Fig.14 is a schematic diagram of the structure of the base station clustering device provided by the present application, such as Fig.14 As shown, the present application also provides a base station clustering device, including the following modules.

[0144] A determination module 1410 is configured to obtain base station information of a to-be-clustered area and determine a target number of clusters according to the base station information; A clustering module 1420 is used to perform base station clustering on the to-be-clustered area based on the improved K-means algorithm according to the base station information and the target number of clusters, to obtain a base station clustering result; The improved K-means algorithm is used to determine the shortest adjacent path distance between any two base stations in the area to be clustered according to the Voronoi diagram of the area to be clustered, and perform base station clustering according to the shortest adjacent path distance.

[0145] The base station clustering device provided in the present application has the technical effect corresponding to the base station clustering method, which will not be described in detail.

[0146] Fig.15 An example of a physical structure diagram of an electronic device is shown in FIG. Fig.15 As shown, the electronic device may include: a processor 1510, a communication interface 1520, a memory 1530 and a communication bus 1540, wherein the processor 1510, the communication interface 1520 and the memory 1530 communicate with each other through the communication bus 1540. The processor 1510 may call the logic instructions in the memory 1530 to execute the base station clustering method.

[0147] In addition, the logic instructions in the above-mentioned memory 1530 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0148] On the other hand, the present application also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the base station clustering method provided by the above methods.

[0149] On the other hand, the present application further provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the base station clustering method provided by the above methods is implemented.

[0150] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0151] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for 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 some parts of the embodiments.

[0152] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application 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 embodiments of the present application.

Claims

1. A base station clustering method, characterized in that: include: Acquire base station information of the area to be clustered, and determine the target number of clusters according to the base station information; According to the base station information and the target number of clusters, base station clustering is performed on the to-be-clustered area based on an improved K-means algorithm to obtain a base station clustering result; The improved K-means algorithm is used to determine the shortest adjacent path distance between any two base stations in the area to be clustered according to the Voronoi diagram of the area to be clustered, and perform base station clustering according to the shortest adjacent path distance.

2. The base station clustering method according to claim 1, characterized in that: The step of performing base station clustering on the to-be-clustered area based on the improved K-means algorithm according to the base station information and the target number of clusters to obtain a base station clustering result includes: Determine, according to the base station information, the initialization cluster center position corresponding to the target clustering quantity and the Voronoi diagram; Determine, according to the Voronoi graph, an adjacency table of each base station in the area to be clustered; Performing a breadth-first search on each base station in the to-be-clustered area according to the adjacency relationship table to determine the shortest adjacent path distance; The base station information, the target number of clusters, the initialization cluster center point information and the shortest adjacent path distance are used as parameters of the K-means algorithm to obtain the improved K-means algorithm; or, the shortest adjacent path distance is weighted based on a predetermined adjacent path distance weight to obtain a weighted shortest adjacent path distance, and the base station information, the target number of clusters, the initialization cluster center point information and the weighted shortest adjacent path distance are used as parameters of the K-means algorithm to obtain the improved K-means algorithm; The improved K-means algorithm is executed to determine the base station clustering result of the area to be clustered.

3. The base station clustering method according to claim 2, characterized in that: The step of determining the adjacent path distance weight comprises: Determine the number of adjacent base stations in any adjacent layer of any base station based on a breadth-first search algorithm, and determine the average number of adjacent base stations in any adjacent layer according to the number of adjacent base stations; The number of adjacent layers corresponding to the average number of adjacent base stations closest to the preset target value of the number of base stations is used as the expected value of the number of adjacent layers; The adjacent path distance weight corresponding to any base station is determined according to the number of adjacent base stations corresponding to the expected value of the number of adjacent layers and the average number of adjacent base stations.

4. The base station clustering method according to claim 2, characterized in that: The executing the improved K-means algorithm to determine the base station clustering result of the to-be-clustered area includes: Traversing the base stations in the area to be clustered, updating the cluster to which each base station belongs according to the adjacency relationship table and the shortest adjacent path distance / the weighted shortest adjacent path distance, and performing clustering correction on isolated clustered base stations; Update the cluster center point information of each cluster, determine whether the updated cluster center information meets the preset clustering convergence condition, if yes, combine the updated clusters to obtain the base station clustering result, otherwise, return to execute the traversal of the base stations in the to-be-clustered area until the preset clustering convergence condition is met or the preset number of cycles is reached.

5. The base station clustering method according to claim 4, characterized in that: If the parameters of the improved K-means algorithm include the weighted shortest adjacent path distance, before updating the cluster center point information of each cluster, the method further includes: Determining the number of base stations in any cluster; If the number of base stations in a certain cluster is greater than a preset target number of base stations, increase the adjacent path distance weight corresponding to the cluster center base station according to the number of base stations, the target number of base stations and a preset control coefficient; If the number of base stations in a certain cluster is less than a preset base station number target value, the adjacent path distance weight corresponding to the cluster center base station is reduced according to the number of base stations, the base station number target value and a preset control coefficient.

6. The base station clustering method according to any one of claims 2 to 5, characterized in that: The base station information includes the latitude and longitude coordinates of the base station, and correspondingly, determining the initialization cluster center position corresponding to the target clustering quantity according to the base station information includes: Constructing a record set and a merge set, storing the longitude and latitude coordinates of all the base stations in the record set, and leaving the merge set empty; Circularly merging the latitude and longitude coordinates of the base stations in the record set according to the merged set to determine the initialization cluster center position; The number of initialized cluster centers is the same as the number of target clusters, and the merged set is used to store the coordinates generated in each merging process.

7. A base station clustering device, characterized in that: include: A determination module, configured to obtain base station information of a to-be-clustered area and determine a target number of clusters according to the base station information; A clustering module, configured to perform base station clustering on the to-be-clustered area based on the improved K-means algorithm according to the base station information and the target number of clusters, to obtain a base station clustering result; The improved K-means algorithm is used to determine the shortest adjacent path distance between any two base stations in the area to be clustered according to the Voronoi diagram of the area to be clustered, and perform base station clustering according to the shortest adjacent path distance.

8. 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 computer program, the base station clustering method according to any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the base station clustering method according to any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the base station clustering method according to any one of claims 1 to 6 is implemented.

Images