CN117376973A

CN117376973A - Regional 5G base station capacity information analysis processing method and system

Info

Publication number: CN117376973A
Application number: CN202311375208.7A
Authority: CN
Inventors: 李亦舒; 杨旭; 杨兆亿; 王铁峰; 刘佳武
Original assignee: Heilongjiang Zhiwang Technology Co ltd
Current assignee: Heilongjiang Zhiwang Technology Co ltd
Priority date: 2023-10-23
Filing date: 2023-10-23
Publication date: 2024-01-09
Anticipated expiration: 2043-10-23
Also published as: CN117376973B

Abstract

The invention discloses a method and a system for analyzing and processing capacity information of a 5G base station in a region; the method comprises the following steps of P1, data collection: various data related to 5G network performance is collected, including base station capacity, user requirements, evolution data, and the like. P2, cluster analysis: a cluster analysis algorithm is used to divide the data set into different clusters to identify different performance patterns in the network. P3, evaluating cellular automaton: and taking the topology of the 5G base station as cells, wherein each cell has spectrum resource attribute, and calculating evolution data of the base station under different time steps through a conversion function. The techniques provided by the present invention can provide more accurate capacity predictions by integrating a variety of data sources, including base station capacity, user demand, and evolution data.

Description

Regional 5G base station capacity information analysis processing method and system

Technical Field

The invention relates to the technical field of 5G base stations, in particular to a method and a system for analyzing and processing capacity information of a regional 5G base station.

Background

The 5G base station needs to process mobile devices connected to it at the same time. These devices may be smartphones, tablet computers, internet of things devices, etc. The capacity planning of the base station takes network load into account, ensuring that enough devices can be connected and serviced simultaneously without causing performance degradation or service interruption. In brief, 5G networks support high-speed data transmission, and thus the capacity of a base station is also related to the ability to provide high-speed data services. This includes download and upload speeds, as well as fluency in the user experience.

In the prior art, a base station needs to have enough bandwidth to support simultaneous data transmission of multiple users. The higher the bandwidth, the greater the number of users and data traffic supported. Wherein the capacity of a base station is also related to its coverage area. The large coverage requires more capacity to meet the user needs of different areas. And the base station capacity planning also involves the management of data traffic to ensure fair allocation and rational utilization of resources. This includes flow control and priority management to meet the needs of different users and applications. Capacity planning is thus an important aspect of 5G network design and operation, which requires consideration of future network growth and user demands to ensure that the network is able to continue to provide high quality services without congestion or performance degradation issues.

In the prior art, there are related technologies for analyzing and strategically planning capacity information of a 5G base station in a region, for example, the technology disclosed in CN201911300742.5, a distributed service caching method in a mobile edge network, and an efficient and stable game theory mechanism is designed for the problem of cost and delay sensitive service caching. Then evaluating the performance of the algorithm through simulation; the proposed appror algorithm can consider the fractional solution due to LP as a set of candidate locations for the service sl. Then, for each sl, the smallest resultant stream from the original instance of the sl to the found candidate location can be found. However, this conventional technique tends to employ a static strategy, i.e., once the cache location is determined, it is not easy to adapt to network changes and the dynamics of the user's needs. This results in inefficient resource utilization, particularly in rapidly changing environments.

And for example, in the method, the device and the equipment for planning the adjacent cells under the networking mode based on overlapping coverage 5GSA disclosed in CN202211341637.8, compared with the 5G adjacent cell planning based on the distance, the 5G adjacent cell planning based on the overlapping coverage area can more accurately reflect the signal overlapping coverage condition among cells, and the adjacent cell is more uniform by adopting a subarea segmentation method, so that the accuracy of adjacent cell configuration is improved. However, this method relies mainly on coverage information, and cannot sufficiently consider other factors, such as user distribution, channel characteristics, and the like. This results in that in some cases the planning of its capacity is still inaccurate. While overlapping coverage area is important, it is not necessarily the only determinant. Other factors such as user requirements, network topology, etc. are also considered.

For example, CN201810311473.1 discloses a networking method for comprehensive load bearing of wireless 5G forwarding and other services, where each 5G remote radio station location determines to construct a transmission access node, an active small-sized wavelength division device is deployed at the transmission access node, an AAU/RRU of the wireless base station forwarding service can be connected to the active small-sized wavelength division device through an optical fiber, a range in which the comprehensive access service can be connected to the active small-sized wavelength division device is defined as a coverage area of the transmission access node, a convergence machine room placed in a DU/BBU set is configured as required, after the convergence wavelength division device is used as a convergence transmission node, the access point networking and convergence layer networking are then determined to form a network topology graph. However, this conventional technique employs a relatively static network topology, which results in rigidity of the network topology, which can result in wasted resources if each site builds a transmission access node, especially in certain sites where there is insufficient capacity to support a transmission access node.

Summarizing, the formal technical drawbacks or improvements of the above-described conventional techniques are set out in:

(1) Limited to a single factor: the conventional technology is mainly based on the interaction relation between coverage area and capacity, and all interactions and relations among a plurality of factors are ignored. Such as user profile, channel characteristics, network topology, etc.

(2) Multi-factor interactions cannot be predicted: conventional techniques tend to be static strategies that have difficulty predicting complex interactions between multiple factors. In 5G networks, many factors interact, but it is difficult for conventional methods to capture and predict these complex relationships. This results in an inefficient network configuration because it cannot accurately predict the impact of multi-factor interactions.

(3) Lack of flexibility: conventional approaches are generally relatively fixed and rigid and difficult to cope with variations in different scenarios and requirements. The network configuration cannot be dynamically adjusted to accommodate different environments and workload conditions. This results in inefficient resource utilization and network performance degradation.

(4) Lack of comprehensiveness: conventional techniques often break down the network planning problem into different sub-problems that are handled separately without comprehensively considering the interrelationship between the problems. This results in local optimization of the sub-problem, while the globally optimal solution is ignored. Thus, the overall consideration is neglected, resulting in an inefficient network configuration.

For this purpose, a method and a system for analyzing and processing capacity information of a 5G base station are provided.

Disclosure of Invention

In view of the foregoing, embodiments of the present invention are directed to a method and a system for analyzing and processing capacity information of a 5G base station, so as to solve or alleviate the technical problems existing in the prior art, that is, limitation to a single factor, inability to predict multi-factor interaction, lack of flexibility and lack of comprehensiveness, and at least provide a beneficial choice for the same;

the technical scheme of the invention is realized as follows:

first aspect

Regional 5G base station capacity information analysis processing method

Introduction to (1)

The method aims at predicting and optimizing the capacity of the 5G network base station by integrating various factors so as to improve the network performance and the resource utilization rate. The principles, steps and key components of the method will be described in detail in this application.

(II) overview

The method for analyzing and processing the capacity information of the 5G base station in the area is a comprehensive method, and comprises the following key steps:

p1, data collection: various data related to 5G network performance is collected, including base station capacity, user requirements, evolution data, and the like.

P2, cluster analysis: a cluster analysis algorithm is used to divide the data set into different clusters to identify different performance patterns in the network.

P3, evaluating cellular automaton: and taking the topology of the 5G base station as cells, wherein each cell has spectrum resource attribute, and calculating evolution data of the base station under different time steps through a conversion function.

And P4, D-S theory verifies: the resulting information was validated using the Dempster-Shafer (D-S) theory, combining information from different sources into a more comprehensive view.

P5, capacity prediction: based on the comprehensive information, the capacity of the 5G base station is predicted, including the capacity amplification condition of the next time step.

(III) implementation

(3.1) data collection

In this step, a plurality of data needs to be collected, the data set including the performance characteristics of the connection number C, the data traffic F and the signal strength S, and the data set is constructed, and the data set is expressed as a feature vector:

sample 1 [ C1, F1, S1]

Sample 2 [ C2, F2, S2]

...

Sample N [ CN, FN, SN ]

Wherein each sample represents a base station; after the data is collected, measurement of cosine similarity is needed to measure the similarity between different base stations, and cosine similarity cosinesimilitude between every two base stations comprises:

x and Y are eigenvectors of two base stations, representing the dot product of the vectors, X/and Y/are the norms of the vectors X and Y, respectively.

(3.2) Cluster analysis

Cluster analysis is to distinguish between different performance patterns in a network to better understand network behavior. In this step, the data set is divided into different clusters using an algorithm such as K-means clustering. The principle of K-means clustering is as follows:

initializing: k cluster centers are randomly initialized.

Cluster allocation step: each sample is assigned to its nearest cluster center.

Cluster updating: the center of each cluster is recalculated, i.e., the average of the samples within the cluster is taken.

Repeating: the cluster allocation and cluster update steps are repeated until the cluster center is no longer changed or a predetermined number of iterations is reached.

In this application, two clustering schemes are provided: l1 norm or L2 norm;

1) The L1 norm: the sum of the absolute values of the vector elements is measured by manhattan distance, comprising, for an n-dimensional vector x and y:

||X||1＝∣x ₁ ∣+∣x ₂ ∣+…+∣x _n ∣

||Y||1＝∣y ₁ ∣+∣y ₂ ∣+…+∣y _n ∣

the sum of the absolute values of the elements in the vector, X, is the sum of the absolute values of the elements in the vector, X/X/1 and Y/1 _n And y _n Is the nth element of vector X;

or alternatively, the first and second heat exchangers may be,

3) The L2 norm: by the square root of the sum of squares of the individual elements of the euclidean distance vector, for an n-dimensional vector x and y, we include:

the sum of the squares of the individual elements of said X2 and Y2 is the square root, X _n And y _n Is the nth element of vector X.

Dividing the base station into K clusters, and minimizing the sum of square errors of samples in the clusters:

j is the objective function, ni is the number of data points in the ith cluster, x _j Is the jth data point in the ith cluster; i=1 denotes an index to clusters, starting from the first cluster up to the kth cluster; j=1 denotes an index to the data points within each cluster, starting from the first data point up to the ni data point;

c _i is the cluster center, is a collection of data points; the cluster center c _i Is the average or centroid of all data points in the cluster, and the sum of the distances of all data points in the cluster to the center point is the smallest, including:

c _i ＝{X1,X2,…,Xn}

the data point x _j Assigned to cluster i, if the data point x _j To the cluster center c _i The distance of (2) is the smallest, the data point x _j Belonging to cluster i; and further solving for the Davies-Bouldin index, comprising the steps of:

1) The compactness: for each cluster i, the average similarity between each pair of data points within the cluster is calculated, expressed as the average similarity of the data points within the cluster:

sim(x _j ，c _i ) Is the data point x _j And the cluster center c _i A similarity measure between;

2) The degree of separation: calculating the distance of each pair of clusters i and j, representing the degree of separation M between different clusters _ij ：

M _ij ＝sim(x _j ，c _i )

3) Calculate the Davies-Bouldin index:

R _i a ratio representing the compactness of cluster i and the separability of other clusters; sj represents the average similarity between each pair of data points within cluster j;

max _j≠i representing selecting the maximum value in the cluster j which is the least similar to the cluster i to measure the separation degree between the cluster i and other clusters;

in the calculation of the Davies-Bouldin index, for each cluster i, in order to measure the degree of separation between cluster i and other clusters, a similarity measure with other clusters j is calculated, and then the cluster j with the smallest similarity measure with cluster i is selected.

(3.3) cellular automaton evaluation

In this step, the topology of the 5G base station is regarded as cells, each cell having spectral resource properties. By defining the transfer function, the evolution data of the base station at different time steps can be calculated. The core of this step is to define a state matrix S whose dimensions are N x (m+2), where each row represents the state of one cell, and each column includes the following information:

1) Column 1 to column M: representing the spectrum resource allocation status, each element s _ij Representing a spectrum resource status of the ith cell allocated to the jth user, the spectrum resource status including 0 (unassigned), 1 (allocated to user 1), 2 (allocated to user 2);

2) Column m+1: representing topological relation information, each element s _i，M+1 Representing binary value topology of the ith cell with other cells；

3) Column m+2: representing signal strength information for each cell, each element s _i，M+2 Representing the signal strength of the ith cell;

the state matrix S is:

the Moore neighborhood:

CC represents a central cell and NO, NE, EA, SE, SO, SW, WE, NW represents eight surrounding cells.

The rule of this state matrix S is:

if one gNB point is currently in state 0 (unassigned) and at least one of its neighbors is in state 1 (assigned to user 1), then the state of that gNB point will change to 1 (assigned to user 1) at the next time step.

If one gNB point is currently in state 0 (unassigned) and at least one of its neighbors is in state 2 (assigned to user 2), then the state of that gNB point will change to 2 (assigned to user 2) at the next time step.

If one gNB point is currently in state 1 (assigned to user 1) and most (more than half) of its neighbors are in state 1, then the state of that gNB point will remain at 1 for the next time step.

If one gNB point is currently in state 2 (allocated to user 2) and most (more than half) of its neighbors are in state 2, then the state of that gNB point will remain at 2 for the next time step.

For the rest of the cases, the state of the gNB point will become 0 (unassigned).

Further comprising one of said transfer functions f:

f＝(S _t+1 ，S _t )＝X _t+1

wherein S is _t Is the current state matrix at time step t, S _t+1 Is the state matrix at time step t+1;

the evolution step of the transfer function comprises:

4) According to the current state matrix S _t To update and calculate the state matrix S for the next time step t+1 _t+1 ；

5) According to the topological relation information, through the current state matrix S _t The information column of the topological relation in (a) determines the relation between each cell and the neighbor cell;

6) Application rules: describing interaction and evolution of the cell under different states through the array information of the state matrix S; calculating a state matrix S of the next time step t+1 according to the application rule _t+1 The new state of each cell is next and stored in the next state matrix S _t+1 In (a) and (b);

4) Obtaining the evolution data D, including state matrix in all time steps, historical state and evolution track of each cell, and recording current state matrix S _t And the state matrix S of the next time step t+1 _t+1 。

(3.4) theoretical verification of D-S

D-S theory is used to combine information from different sources into a more comprehensive view. In this step, the Davies-Bouldin index in Track-1 was used as evidence A and the evolution data of Track-2 was used as evidence B using the Dempster's combination principle. Trust allocation function m (AB) for evidence a and evidence B _ij Then, the base station capacity is mapped into a section value from 0 to 1 through a sigmoid function, and the amplification condition of the 5G base station capacity in the next time step is further obtained:

1) For each m (AB) _ij It is mapped to a probability value using a sigmoid function:

p (ij) represents the probability of the 5G base station capacity increase situation for cell i at time step j;

(ij) is the input of a sigmoid function, exp is an exponential function

The key steps of the D-S theory are as follows:

trust allocation function: a trust allocation function is built for each data source to represent its trustworthiness.

Dempster's combination principle: the information of each data source is combined into a whole view angle by using the Dempster's combination principle, so that more reliable prediction of the capacity of the 5G base station is obtained.

(3.5) Capacity prediction

And finally, predicting the capacity of the 5G base station through the comprehensive information. This includes the capacity augmentation situation for the next time step to help the network operator optimize resource allocation and network configuration. Comprising the following steps:

(1) Capacity amplification calculation: and according to the mapping result of the sigmoid function, the sigmoid function is subjected to percentization, and the capacity amplification condition of the next time step is calculated. Indicating how much capacity the next time step needs to increase or decrease relative to the current capacity.

(2) Network topology adjustment: according to the capacity amplification, the network topology is considered to be adjusted. This may include increasing or decreasing the number of base stations, changing the connection relationship between base stations, or adjusting the coverage of the base stations. For example, if the capacity increase is large, it may be necessary to add base stations to meet the user demand.

(3) Spectrum resource allocation: and reallocating the spectrum resources according to the capacity amplification condition. If capacity requirements increase, more spectrum resources may need to be allocated to a particular base station or frequency band. Conversely, if the capacity demand decreases, the spectrum resources may be reallocated to increase the resource utilization.

(4) Capacity monitoring and adjustment: after implementing the capacity-increasing plan, the network performance needs to be monitored periodically to ensure the effectiveness of capacity adjustment. Further adjustments may be made if the capacity is found to remain insufficient or excessive.

(5) User experience improvement: the final goal of optimizing resource allocation and network configuration is to improve the user experience. The network operator may be concerned with indicators of user satisfaction, data speed, connection quality, etc. to ensure that the user gets better service.

(6) Capacity planning update: the capacity plan is updated periodically to reflect changes in the network and evolution of user demand. This may be a continuous process to ensure that the network is always able to meet demand.

Second aspect

Regional 5G base station capacity information analysis processing system

The system is an intelligent system for analyzing and processing the capacity information of the regional 5G base station. It is capable of performing the information analysis processing method described previously by integrating the processor, memory and program instructions to help network operators better understand and optimize 5G network performance. The following are key components and functions of the system:

(1) A processor: the present system is provided with a high performance processor for executing program instructions stored in memory. The processor is responsible for managing the data processing flow and executing the various steps.

(2) A memory: the memory is used to store program instructions, data sets, and other necessary information. This includes a variety of data sources from 5G base stations, user requirements, evolution data, and the like.

By means of automation, the system can process data in the 5G network in a large scale, and more comprehensive capacity analysis and prediction are provided. The system is able to identify performance patterns in the network in order to better understand network behavior and optimize. By integrating multi-factor and considering multi-factor interactions, the system is able to more accurately predict future capacity needs. Based on the capacity prediction results, the system provides optimization suggestions that help network operators to better configure resources and networks.

Compared with the prior art, the invention has the beneficial effects that:

1. accurate capacity prediction: the techniques provided by the present invention can provide more accurate capacity predictions by integrating a variety of data sources, including base station capacity, user demand, and evolution data. This helps the network operator to better plan network resources, avoiding problems of excessive investment or insufficient resources. Meanwhile, the technology provided by the invention can consider the interaction of various factors including user demands, network loads, spectrum resources and the like, and can perform capacity analysis in a more comprehensive mode. This helps to better understand the complexity of network performance.

2. Network performance optimization: by providing capacity prediction and optimization suggestions, the technology provided by the invention can help network operators to optimize network configuration, improve network performance and improve user experience. This helps to improve the competitiveness and sustainability of the network. Meanwhile, the technology provided by the invention is beneficial to improving the utilization rate of network resources and avoiding the waste of the resources. By adjusting the resource allocation according to the actual demands, the cost can be reduced and the efficiency can be improved.

3. Flexibility and adaptability: because the technology provided by the invention can dynamically predict the capacity requirement, the network operators can more flexibly cope with the changing user requirement and network environment. This increases the adaptability and response capability of the network. Meanwhile, the technology provided by the invention introduces the data-driven decision into the network management, so that the decision is more basis. The operator can make more informed decisions based on the data analysis results.

4. User experience improvement: by better managing capacity, optimizing network performance, improving resource utilization, the ultimate beneficiary is a network user. They will get better quality of service and more reliable connections. Meanwhile, the technology provided by the invention is beneficial to planning future network development of network operators. Through accurate capacity prediction, guidance can be provided for future expansion and upgrading.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings that are required in the embodiments or the technical descriptions will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

FIG. 1 is a schematic flow chart of the method of the present invention;

FIG. 2 is a control program diagram of an eleventh embodiment of the present invention;

FIG. 3 is a control program diagram of an eleventh embodiment of the present invention;

FIG. 4 is a control program diagram of an eleventh embodiment of the present invention;

FIG. 5 is a control program diagram of an eleventh embodiment of the present invention;

Detailed Description

In order that the above objects, features and advantages of the invention will be readily understood, a more particular description of the invention will be rendered by reference to the appended drawings. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. This invention may be embodied in many other forms than described herein and similarly modified by those skilled in the art without departing from the spirit of the invention, whereby the invention is not limited to the specific embodiments disclosed below;

Example 1

In order that the above-recited embodiments of the invention may be understood in detail, a more particular description of the invention, briefly summarized below, may be had by way of example. The present invention may be embodied in many other forms than described herein and similarly modified by those skilled in the art without departing from the spirit of the invention, so that the invention is not limited to the embodiments disclosed below.

Referring to fig. 1, a method for analyzing and processing capacity information of a 5G base station includes the following steps of implementing Track-1 and Track-2 synchronously:

track-1, cluster analysis: establishing a data set, carrying out cosine similarity measurement, initializing an estimated value of a clustering center by using K-means clustering, iterating, and measuring the compactness in clusters and the separation among clusters by using Davies-Bouldin indexes;

track-2, cellular automaton evaluation: taking a topological gNB point of a 5G base station as a cell, wherein each cell has a spectrum resource attribute, the spectrum resource attribute is set as a state matrix S, a Moore neighborhood is adopted to determine a cell neighbor relation, and evolution of the gNB point under different time steps is measured and calculated through a conversion function to obtain evolution data D;

also included is Track-3 implemented after Track-1 and Track-2:

Track-3, D-S theory verifies: using the degree of separation in Track-1 as evidence A and the evolution data of Track-2 as evidence B; the Dempster's combination principle is used to obtain the 5G base station capacity amplification at the next time step.

In this embodiment, track-1, cluster analysis, is an unsupervised learning method, whose goal is to divide the data into different clusters, so that the similarity of data points in each cluster is higher, and the similarity of data points between different clusters is lower. In the area 5G base station capability information analysis process, this step is used to identify base station clusters for different performance modes.

Specifically, first, capacity data from 5G base stations is collected and constructed into a data set, where each data point represents a base station. Cosine similarity is used to measure similarity between base stations. The cosine similarity measure considers the angle between base stations, with closer to 1 indicating higher similarity and closer to-1 indicating lower similarity. The estimate of the cluster center is initialized, typically with randomly selected K data points. Each base station is then assigned to its nearest cluster center. This step divides the base stations into different clusters.

Further, through iteration, the estimated value of the cluster center is continuously updated until convergence. The goal of the iteration is to minimize the sum of squared errors within the cluster to ensure that the similarity of data points within the cluster is high.

Further, davies-Bouldin index: the Davies-Bouldin index was used to measure the compactness within clusters and the separation between clusters. The index takes into account the degree of similarity between different clusters and the degree of dispersion of data points within a cluster. A lower Davies-Bouldin index indicates better cluster quality.

In this embodiment, track-2: cellular automaton evaluation: is a computational model that divides space into a set of cells and updates the state of the cells in discrete time steps according to a set of rules. Here, the cells represent 5G base stations, and the states thereof represent spectrum resource attributes.

Specifically, the state matrix S: each base station is regarded as a cell, each cell having spectral resource properties, which constitute a state matrix S. The matrix describes the spectrum resource allocation situation of the base station at different time steps.

Specifically, moore neighborhood relationship: a Moore neighborhood is used to determine the neighbor relationship between cells (base stations). The Moore neighborhood includes one cell and eight cells around it to represent interactions between base stations.

Specifically, the transfer function: by applying the transfer function, the evolution data of the base station at different time steps is calculated. This function takes into account the state of the base station and its neighbors, as well as the dynamic allocation rules of the spectrum resources. The specific rules of the transfer function are typically designed according to the actual requirements and network environment.

Specifically, evolution data D: and updating the state matrix S of the base station under each time step through iterative simulation to obtain evolution data D. These data describe the evolution of the spectrum resources of the base station, which can be used for subsequent analysis.

In this example, track-3:D-S theory verifies that: D-S theory is a theory used to deal with uncertainty and combine different evidence. Here it is used to combine the results of the cluster analysis and cellular automaton evaluation to provide a more comprehensive network capacity estimation.

Specifically, evidence a: the Davies-Bouldin index obtained from Track-1 is used as evidence A to indicate the degree of separation between different clusters. A lower degree of separation means a better clustering result.

Specifically, evidence B: the evolution data obtained from Track-2 is used as evidence B to describe the evolution of the base station spectrum resources. This is based on the simulation results of cellular automata.

Further, the Dempster's combination principle: evidence a and evidence B are combined using the Dempster's combination principle to get an increase in 5G base station capacity at the next time step. This step takes into account uncertainty information from two different sources, providing a more comprehensive capacity estimate.

By integrating the three tracks (Track-1, track-2 and Track-3), the regional 5G base station capacity information analysis processing method can provide more accurate and comprehensive capacity estimation, and is beneficial to better planning and optimizing resource allocation and network configuration of a network operator. The method utilizes the principles of data analysis, clustering, simulation and uncertainty processing and provides a powerful tool for 5G network management.

Summarizing, compared with the defects in the conventional technology, the present embodiment adopts the following technical features or means to achieve the solution:

(1) And (3) integrating multiple factors: the methods discussed in this embodiment explicitly emphasize the importance of integrating a variety of factors, including base station capacity, user requirements, evolution data, etc. This is achieved by using D-S theoretical verification, where information from different sources is combined into a more comprehensive view. This integration allows for a more comprehensive consideration of the aspects of network performance rather than relying on a single factor alone, thereby solving the problem of being limited to a single factor.

(2) Predicting multi-factor interactions: the new approach better predicts the behavior of the network by taking into account interactions between different factors. The use of D-S theory validation helps to capture complex relationships between different factors and incorporate these relationships into the predictive model. This may help to solve the problem of unpredictable multi-factor interactions.

(3) Flexibility: the method discussed in this embodiment appears to be more flexible and can be adjusted and optimized for the particular situation. Since it does not depend on a fixed network topology or node configuration, but can dynamically perform capacity planning and network configuration according to different scenarios. The problem of lack of flexibility is solved.

(4) Comprehensiveness: by integrating different information sources and using D-S theory verification, the method can provide a more comprehensive network capacity estimation. This helps to solve the problem of lack of comprehensiveness because it takes into account the effects of multiple factors rather than breaking down the problem into different sub-problems.

The above examples merely illustrate embodiments of the invention that are specific and detailed for relevant practical applications and are not to be construed as limiting the scope of the invention. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the invention, which are all within the scope of the invention. Accordingly, the scope of protection of the present invention is to be determined by the appended claims.

Example two

In Track-1, the present embodiment uses cosine similarity to measure similarity between different base stations to help cluster analysis;

specifically, cosine similarity is a method for measuring the similarity between two vectors, and the cosine value of the included angle between the two vectors is calculated to determine the similarity. In the area 5G base station capability information analysis process, the present embodiment uses cosine similarity to compare the performance characteristics of different base stations to determine whether they belong to similar clusters.

In this embodiment: first, each base station is represented as a feature vector, which includes performance features such as the number of connections (C), data traffic (F), signal strength (S), and the like. Each eigenvector corresponds to a base station, and the dataset is represented by eigenvectors as:

Sample 1 [ C1, F1, S1]

Sample 2 [ C2, F2, S2]

...

Sample N [ CN, FN, SN ]

Wherein each sample represents a base station;

measurement of cosine similarity: for the feature vectors X and Y of the two base stations, the present embodiment can calculate the similarity between them using a cosine similarity formula. The formula is as follows:

Further, the result of the cosine similarity calculation is a value between-1 and 1, where 1 represents complete similarity, -1 represents complete dissimilarity, and 0 represents independence. The present embodiment can measure the degree of similarity between different base stations by calculating the cosine similarity between them.

Specifically, based on the computed cosine similarity, the base stations may be grouped into different clusters, where the similarity between base stations in the clusters is higher, and the similarity between different clusters is lower. This helps identify clusters of base stations for different performance modes. Through this step, the present embodiment can use the cosine similarity measure to construct the similarity relation between the base stations, and provides a basis for subsequent cluster analysis. This helps to better understand the relationship between the base station performance characteristics for capacity analysis and optimization.

Illustratively, when the present embodiment has a data set comprising a plurality of base stations, the characteristic feature vector samples 1: [ C1, F1, S1], sample 2: [ C2, F2, S2]; the cosine similarity between these base stations is then calculated. First, a dot product (x·y) is calculated:

X·Y＝C1*C2+F1*F2+S1*S2

next, norms (modulo lengths) of the respective feature vectors are calculated:

now, the present embodiment can substitute these values into the cosine similarity formula:

this will give a cosine similarity value between the two samples, which represents their degree of similarity in the feature space. The closer the value is to 1, the more similar they are, while the closer the value is to-1, the less similar they are. This similarity value may be used to measure the similarity of performance characteristics between base stations for cluster analysis or other related tasks. One exemplary result is as follows:

sample 1 [10,200, -30]

Sample 2 [8,180, -28]

By substituting these values into cosine similarity formulas, the present embodiment can calculate cosine similarity between them, thereby measuring their degree of similarity in the performance feature space. This similarity value may help the present embodiment understand the relationship and interaction between base stations.

Example III

In this embodiment, X and Y are L1 norms; wherein:

l1 norm: the sum of the absolute values of the vector elements is measured by manhattan distance, since it is similar to the distance traveled on grid roads in a city, for an n-dimensional vector x and y, comprising:

||X||1＝∣x ₁ ∣+∣x ₂ ∣+…+∣x _n ∣

||Y||1＝∣y ₁ ∣+∣y ₂ ∣+…+∣y _n ∣

the sum of absolute values of elements in the vector, X, is the sum of the absolute values of the elements in the vector, i.e., X/1 and Y/1 _n And y _n Is the nth element of vector X;

wherein X1, X2,..xn are the 1 st, 2 nd,..n th element of vector X, respectively.

In this embodiment, the calculation logic of the L1 norm is to add the absolute value of each element in the vector. This is similar to the distance travelled along a grid road in a city and is therefore also known as manhattan distance. It takes into account the absolute contribution of each element, not just their square.

In particular, the L1 norm is more applicable than the other norms in some cases, particularly when the present embodiment is concerned with the sum of the absolute values of the vector elements. For example, in an optimization problem, L1 regularization is used to sparsify parameters in order to obtain a model with fewer non-zero elements.

Specifically, the method is applied to the analysis processing of the capacity information of the 5G base station in the area: in this context, the L1 norm may be used for normalization of different performance characteristics to ensure that they are of equal importance. When the present embodiment uses cosine similarity in Track-1, the present embodiment uses the L1 norm to calculate the modulus length (norm) of the vector to ensure that the similarity measure is based on the sum of the absolute values of the performance characteristics.

Further, as seen from the content of the second embodiment, in the area 5G base station capacity information analysis processing, the present embodiment processes the data sets with different performance characteristics, including the number of connections (C), the data traffic (F), the signal strength (S), and the like. These performance characteristics are typically of different numerical ranges and importance. These features can be normalized using the L1 norm, ensuring that they have the same weight. This is important for cluster analysis because it can prevent certain features from dominating the similarity measure, thus yielding inaccurate cluster results.

Further, cosine similarity measures are used in Track-1 to compare similarity between different base stations. By normalizing the feature vectors using the L1 norm, the present embodiment can ensure that the sum of the absolute values of each feature has a balanced effect on the similarity calculation. This means that the size of the different features does not significantly affect the similarity measure, thereby improving the accuracy of the clustering. And the regional 5G base station capability information analysis process aims to comprehensively consider different performance characteristics and identify base station clusters with similar performance characteristics. The use of the L1 norm helps to ensure that each performance feature can be treated equally, without being dominated by the absolute values of the other features. This allows the present embodiment to better understand the relationship between different performance characteristics, providing a more comprehensive view of capacity analysis.

Illustratively, there are two vectors X and Y, which are each as follows:

X＝[3,-1,5]

Y＝[-2,4,1]

now, the present embodiment will calculate their L1 norms. First, the sum of absolute values of the respective elements of X and Y is calculated:

||X||1＝|3|+|-1|+|5|＝3+1+5＝9

||Y||1＝|-2|+|4|+|1|＝2+4+1＝7

now, the present embodiment has calculated L1 norms of X and Y, 9 and 7, respectively. This means that the L1 norm of vector X is greater than the L1 norm of vector Y because the sum of the absolute values of the elements in X is greater.

This example illustrates how the L1 norm may be used to scale the vector size, and more complex vectors may be calculated in a similar manner. In the area 5G base station capability information analysis process, the L1 norm is used to normalize the performance feature vectors, ensuring that they have equal weights in the similarity measure, thus supporting accurate cluster analysis.

Example IV

This embodiment is different from the third embodiment, the sum of the X/Y/of the signals can be L2 norm; wherein:

4) L2 norm: the square root of the sum of squares of the individual elements through the euclidean distance vector, because it resembles the straight line distance between points, comprises for an n-dimensional vector x and y:

x2 and Y2 are the square root of the sum of the squares of the individual elements of the vector, X _n And y _n Is the nth element of vector X. X1, X2, xn are the 1 st, 2 nd, nth element of vector X, respectively.

In this embodiment, the L2 norm calculation logic is related to the euclidean distance. It measures the straight line distance from the origin to the end of the vector. In the area 5G base station capability information analysis process, the L2 norm may be used to measure the size of the performance feature vector regardless of the absolute value of each feature.

Specifically, the calculation of the L2 norm involves summing the squares of the individual elements of the vector and taking the square root. This makes the contribution of large elements to the norm larger, but it does not take into account the absolute value as the L1 norm does.

Further, the focus of the present embodiment is on the application area 5G base station capacity information analysis processing: the L2 norm may be used to normalize the performance feature vector, and the L2 norm is better suited to scale the size of the vector, especially when the present embodiment is concerned with the sum of squares of the individual elements of the vector. In performance feature normalization and similarity metrics, the L2 norm provides a different metric that helps to more fully understand the relationship between performance features.

In this embodiment, in view of the content of the second embodiment, in the area 5G base station capacity information analysis processing, the L2 norm is calculated in a manner related to the euclidean distance, which means that it can be used to measure the distance between different performance feature vectors. In the area 5G base station capability information analysis process, the present embodiment may concern performance differences between different base stations, such as differences in the number of connections, data traffic, and signal strength. By calculating the L2 norms between these performance feature vectors, the present embodiment can measure the similarity or difference between them, which is very useful for cluster analysis.

Further, the calculation of the L2 norm involves summing the squares of the individual elements of the vector and taking the square root. This process helps balance the contributions of the individual features, as squaring makes the contribution of large elements to the norm more pronounced. This helps to avoid that an excessive absolute value of a certain feature leads to a deviation of the similarity measure.

Further, in the area 5G base station capacity information analysis process, the present embodiment needs to comprehensively consider a plurality of performance characteristics to identify base station clusters having similar performance characteristics. The L2 norm provides a way to integrate these features that takes into account the sum of squares of the individual features, enabling the present embodiment to more fully understand the relationship between the different features.

Further, the L2 norm may be used to normalize the performance feature vectors, ensuring that they have equal weights in the similarity measure. This is important for cluster analysis because it can avoid some feature dominant similarity calculations, thus avoiding inaccurate cluster results.

Example five

In this embodiment, K-means clustering will be used to divide the data points into clusters with similar features. In the area 5G base station capacity information analysis process, the goal of K-means clustering is to divide the base stations into K clusters so that base stations of similar performance characteristics are classified into one class. This helps operators understand the performance of different base stations in the network, providing guidance for resource optimization and network configuration.

Specifically, in Track-1, K-means clustering: dividing the base station into K clusters, and minimizing the sum of square errors of samples in the clusters:

j is the objective function, ni is the number of data points in the ith cluster, x _j Is the jth data point in the ith cluster; i=1 denotes an index to a cluster, starting from the first cluster and continuingTo the kth cluster; j=1 denotes an index to the data points within each cluster, starting from the first data point up to the ni data point;

c _i is the cluster center, is a collection of data points; cluster center c _i Is the average or centroid of all data points in the cluster, and the sum of the distances of all data points in the cluster to the center point is the smallest, including:

c _i ＝{X1,X2,…,Xn}

data point x _j Assigned to cluster i, if data point x _j To cluster center c _i Is the smallest, data point x _j Belonging to cluster i;

further, K initial cluster centers are selected, and can be randomly selected or initialized by other methods. These cluster centers represent the K initial clusters of base stations. For each base station, its distance from each cluster center is calculated (typically using Euclidean distance or other distance metric), and then the base stations are assigned to the clusters corresponding to the closest cluster center. For each cluster, the average of all base stations within the cluster is calculated to obtain a new cluster center. This new cluster center better represents the performance characteristics of the cluster. The above steps are repeated until the cluster center is no longer changed or changed little, or a predetermined number of iterations is reached. This will ensure that the cluster center converges to the optimal position. The quality of the clusters is evaluated using an objective function J. The objective function J represents the sum of squares of the distances of the data points within a cluster to the center of the cluster to which it belongs, with the objective of minimizing this value. Smaller J values indicate that the data points within the cluster are more tightly packed together.

Further, K in K-means clustering is the number of clusters specified in advance. The elbow rule and profile coefficients may be used to select the appropriate K value. This is the prior art and will not be described in detail.

In this embodiment, the performance characteristics of the base station are key information, including the number of connections, data traffic, signal strength, and the like. K-means clustering can classify base stations with similar performance characteristics into the same cluster, so that network operators are helped to better understand the performance of different base stations. Wherein K-means clustering can also be used to detect abnormal base stations. A base station may be considered an abnormal base station when its performance characteristics are significantly different from other base stations in its cluster, requiring special attention. This helps to find out the problem base station in the network in time and take corresponding measures.

Further, in the present embodiment, J is an objective function; in K-means clustering, an objective function J is expressed as the sum of distances from data points in a cluster to the center of the cluster to which the data points belong; the present embodiment thus illustratively provides another composition of the objective function J:

wherein J _i Representing the objective function value of the i-th cluster. The meaning of this new objective function is that the present embodiment wishes to minimize the sum of the distances of the data points within each cluster from the center of the cluster to which it belongs. This means that data points should be assigned to cluster centers closest to them so that the data points within each cluster are more similar, thereby maximizing compactness within the cluster. The total objective function J is thus the objective function J of all clusters _i The sum is that:

this new (total) objective function J _i The performance of the overall K-means clustering process is shown, and the goal of this example is to find a set of cluster centers, minimizing J, to get the proper cluster partitioning.

Summarizing, this new (total) objective function J _i For the core logic of this embodiment, K-means clustering provides guidance for resource optimization by dividing the base stations into different clusters. The operator can adjust the resource allocation according to the performance characteristics of the base stations in the cluster to improve the overall performance and capacity utilization of the network.

In this embodiment, the iterative process:

1) Cluster allocation: for data point x _j Calculate its center with each cluster center clusterc _i Distance of data point x _j Assigned to cluster i to which the cluster center closest thereto belongs:

i＝argmin _k ||x _j -c _i || ²

2) Cluster update: at this step, for each cluster i, the center c of that cluster is recalculated _i . Center c _i Is all data points x in the cluster _j Is calculated by the following formula:

this process updates the center of the cluster to the average of all the data points within the cluster to ensure that the cluster center can represent the data points within the cluster.

If the cluster center does not change significantly any more or reaches the maximum iteration number, stopping iteration, otherwise, continuing cluster allocation and cluster updating; after the iteration is finished, each data point is distributed to the final cluster center, and K-means clustering is finished.

Exemplary, in cluster allocation, it is also possible to x for each data point _j Calculate it and each cluster center c _i Is a distance of (3). The distance is measured here using the square of the euclidean distance, since the computation is faster and the same effect. The distance calculation formula is:

then, data point x _j Assigned to the cluster center c nearest to it _i Cluster i to which it belongs. This process is accomplished by selecting the minimum distanceI to:

This step assigns the data point to the nearest cluster center, completing a round of cluster reassignment.

In this embodiment, the goal of K-means clustering is to minimize the objective function J, i.e., the sum of the distances of the data points within a cluster to the center of the cluster to which it belongs. This is achieved by constantly performing cluster allocation and cluster update steps. Eventually, each data point will be assigned to the cluster most suitable for it, forming a clustered result. It comprises the following steps:

(1) Clustering data points: k-means clustering divides data points into K clusters, each cluster containing data points with similar characteristics. This helps to efficiently organize and categorize large amounts of base station data for better understanding of the different subgroups in the network.

(2) Similarity analysis: each cluster represents a similar set of base stations because the data points within the cluster have smaller intra-cluster distances. This enables the network operator to better analyze and understand the similarities between these base stations and their performance in the network.

(3) Performance comparison: by grouping the base stations into different clusters, the performance characteristics of the base stations within the different clusters, such as the number of connections, data traffic and signal strength, can be easily compared. This helps operators to identify performance differences in the network and take corresponding measures to optimize.

(4) Network optimization: based on the performance characteristics of the different clusters, operators may take targeted network optimization strategies. For example, in clusters of high data traffic, resource allocation may be enhanced, while in clusters of low signal strength, signal enhancement measures may be taken.

(5) And (3) resource allocation: grouping base stations into different clusters can help operators allocate resources more efficiently to meet the needs of the different clusters. This helps to improve resource utilization and reduce network operating costs.

Summarizing, in the 5G base station capability information analysis process, K-means clustering can be used to partition base stations into different clusters for further analysis and optimization of network configuration. The nature of the different clusters may help operators to better understand network performance and user requirements and to take corresponding measures to implement planning of network capacity.

Example six

In this embodiment, the Davies-Bouldin index comprises:

1) Tightness: the closeness measures the similarity or degree of aggregation between data points within a cluster. For cluster i, the present embodiment calculates the average similarity between each pair of data points within the cluster, expressed as the average similarity S of the data points within the cluster _i . The similarity measure herein typically uses cosine similarity or other suitable similarity measure. For each cluster i, an average similarity between each pair of data points within the cluster is calculated, expressed as an average similarity S of the data points within the cluster _i ：

Where ni represents the number of data points within cluster i; sim (x) _j ，c _i ) Is the data point x _j And cluster center c _i A similarity measure between;

2) Degree of separation: next, the present embodiment calculates the distance between each pair of clusters i and j to represent the degree of separation M between different clusters _ij . Here, theThe distance is typically measured by a similarity measure sim (x _j ，c _i ) I.e. the similarity measure between the center of cluster i and the center of cluster j.

Calculating the distance between each pair of clusters i and j to represent the separation M between different clusters _ij ：

M _ij ＝sim(x _j ，c _i )

3) Calculate the Davies-Bouldin index: finally, the present embodiment uses the Davies-Bouldin index to consider compactness and separability in combination to measure cluster quality. For each cluster i, the present embodiment calculates a similarity measure s with other clusters j _i +s _j Degree of separation M _ij Then selecting the maximum value in cluster j which is least similar to cluster i to obtain R _i 。

R _i A ratio indicating the compactness of cluster i and the separability of other clusters; s is S _j Representing the average similarity between each pair of data points within cluster j;

max _j≠i representing selecting the maximum value in the cluster j which is the least similar to the cluster i to measure the separation degree between the cluster i and other clusters; in the calculation of the Davies-Bouldin index, for each cluster i, in order to measure the degree of separation between cluster i and other clusters, a similarity measure with other clusters j is calculated, and then the cluster j with the smallest similarity measure with cluster i is selected.

It will be appreciated that this represents the degree of separation between cluster i and other clusters, R _i The smaller the value of (c) is, the better the separation between cluster i and other clusters is, and the higher the similarity between data points in the clusters is. The Davies-Bouldin index is the average of the maximum values of Ri for all clusters.

Illustratively, there are 3 clusters: cluster 1, cluster 2, and cluster 3, each cluster having two samples. The aim of this example is to calculate the Davies-Bouldin index while illustratively calculating the closeness, separation of one cluster.

1) Tightness: for cluster 1, let cluster center beC1＝[2，3]And the two samples are x1= [1,2, respectively]And X2= [3,4]. The present embodiment may use euclidean distance as a similarity measure. Average similarity S _i ：

Also, for cluster 2 and cluster 3, the present embodiment can calculate their compactness.

2) Calculating the separation degree:

sim(C ₁ ，C ₂ )＝0.5

sim(C ₁ ，C ₃ )＝0.3

sim(C ₂ ，C ₃ )＝0.4

calculating the separation degree of the cluster 1 from other clusters:

M ₁₂ ＝sim(C ₁ ，C ₂ )＝0.5

M ₁₃ ＝sim(C ₁ ，C ₃ )＝0.3

calculating the separation degree of the cluster 2 from other clusters:

M ₂₁ ＝sim(C ₂ ，C ₁ )＝0.5

M ₂₃ ＝sim(C ₂ ，C ₃ )＝0.4

calculating the separation degree of the cluster 3 from other clusters:

M ₃₁ ＝sim(C ₃ ，C ₁ )＝0.3

M ₃₂ ＝sim(C ₃ ，C ₂ )＝0.4

3) Calculate the Davies-Bouldin index: j is 2 or 3, taking the maximum value. For simplicity of description in this example, the present embodiment demonstrates the Davies-Bouldin index of only one cluster to demonstrate its principle:

from this demonstration, the present example can see how to calculate the compactness, the degree of separation and the final Davies-Bouldin index of a cluster. In practice, the present embodiment will calculate the Davies-Bouldin index for all clusters and select the cluster number with the smallest Davies-Bouldin index as the best cluster number. This helps to better divide the base stations in 5G networks for resource allocation and network optimization.

Summarizing, by calculating the Davies-Bouldin index, the present embodiment can evaluate the quality of each cluster in the K-means cluster to determine the optimal cluster number K. The K value with the smallest Davies-Bouldin index is generally considered the best cluster number because it represents the best balance of intra-cluster compactness and inter-cluster separation, helping to better partition the base stations or data points. This can be used in 5G networks to determine the partitioning of clusters for resource allocation and network optimization.

Example seven

In Track-2 of this embodiment, this embodiment focuses on the cellular automaton model, where each cell represents a 5G base station and has spectral resource properties. These attributes record the state information of the base station, including spectrum resource allocation state, topology relationship information, and signal strength information. This information is critical to analyzing and optimizing the capacity and performance of 5G networks.

In this embodiment, the cells have a spectrum resource attribute, which is a state matrix S; the state matrix includes: n cells i and M users, the dimension of the state matrix S is N x (m+2), where each row represents the state of one cell, and each column includes the following information:

1) Column 1 to column M: representing the spectrum resource allocation status, each element s _ij Representing the spectrum resource status of the ith cell allocated to the jth user;

specifically, these columns contain the status of the spectrum resources allocated to each user by the base station. Each element Sij represents the spectrum resource status allocated to the jth user by the ith cell. These states may be generally represented as numbers, e.g., 0 for unassigned, 1 for assigned to user 1,2 for assigned to user 2, and so on. This part of information reflects the spectrum resource allocation situation between the base station and the user, and is very important for analysis and optimization of network capacity.

2) Column m+1: representing topological relation information, each element s _i，M+1 Representing the binary value topological relation between the ith cell and other cells;

specifically, this column contains binary topology information between each cell and other cells. Each element s _i，M+1 Indicating the connection state of the i-th cell with other cells. This information can be used to determine the connection between the base stations and facilitate the understanding and analysis of the network topology by the cellular automaton model.

3) Column m+2: representing signal strength information for each cell, each element s _i，M+2 Representing the signal strength of the ith cell; may be represented by a numerical value representing the strength or quality of the signal;

specifically, this column contains the signal strength information for each cell. Each element s _i，M+2 The signal strength of the i-th cell is typically represented by a numerical value. Signal strength is a key performance indicator that can impact user experience and network capacity. By analyzing the signal strength information, signal coverage in the network can be determined, helping to identify potential signal vulnerabilities and optimize the networkAnd (5) resource allocation.

The dimension of the state matrix S is N x (m+2), where N represents the number of base stations and M represents the number of users. The content of the state matrix S records the state information of each base station, which is critical for the analysis and processing of the 5G base station capacity information. Through a cellular automaton model, the embodiment can simulate the evolution process of the base station, and further perform capacity analysis and performance optimization. This detailed status information helps the network operator to better understand and manage its 5G network. Specific:

The rules of this state matrix S are as follows:

(1) Rule 1: if one gNB point is currently in state 0 (unassigned) and at least one of its neighbors is in state 1 (assigned to user 1), then the state of that gNB point will change to 1 (assigned to user 1) at the next time step. This means that when there is a base station around an unassigned base station to which user 1 has been assigned, the base station will attempt to assign to user 1.

(2) Rule 2: if one gNB point is currently in state 0 (unassigned) and at least one of its neighbors is in state 2 (assigned to user 2), then the state of that gNB point will change to 2 (assigned to user 2) at the next time step. This means that when there is an unassigned base station surrounding that to which user 2 has been assigned, the base station will attempt to assign to user 2.

(3) Rule 3: if one gNB point is currently in state 1 (assigned to user 1) and most (more than half) of its neighbors are in state 1, then the state of that gNB point will remain at 1 for the next time step. This means that if a majority of the base stations around a base station that has been allocated to user 1 have been allocated to user 1, then that base station will continue to remain allocated to user 1.

(4) Rule 4: if one gNB point is currently in state 2 (allocated to user 2) and most (more than half) of its neighbors are in state 2, then the state of that gNB point will remain at 2 for the next time step. Again, this means that if most of the base stations around a base station that has been allocated to user 2 have been allocated to user 2, then that base station will continue to remain allocated to user 2.

(5) Rule 5: for the rest of the cases, the state of the gNB point will become 0 (unassigned). This means that if there are not enough neighbor base stations in the same state around a base station, or if no neighbor base stations are in an allocated state, then the base station will return to an unallocated state.

These rules help simulate the spectrum resource allocation and state evolution process between 5G base stations, and through the cellular automaton model, the performance and capacity of the 5G network can be better understood and optimized. This is very valuable for network operators. As they can predict and adjust the state of the base station according to these rules to meet changing network demands.

It should be noted that, in this embodiment, the rule of the state matrix S mentioned above conforms to the actual logic of capacity management and spectrum resource allocation of the 5G base station, and includes:

(1) Spectrum resource allocation logic: rule 1 and rule 2 relate to how the base station allocates spectrum resources to different users. This reflects the limited and competitive spectrum resources in 5G networks. When there is already one user 1 or 2 around one base station, the base station will give priority to assigning it to the corresponding user to meet the user demand to the greatest extent.

(2) Collaborative logic between base stations: rule 3 and rule 4 consider the cooperativity between base stations. They indicate that if most base stations around a base station have been allocated to the same user, the base station will continue to maintain the same user allocation status. Such synergistic logic helps to optimize the utilization of network resources, avoiding frequent state changes and handovers.

(3) And (3) adaptability logic: rule 5 indicates that if there are not enough neighbor base stations in the same state around a base station, or if no neighbor base stations are in an assigned state, then the base station will return to an unassigned state. Such adaptive logic helps to cope with the ever changing distribution and demand of users in the network. When there are not enough users in a certain area or spectrum resources need to be allocated, the base station will remain unallocated to wait for more demands.

(4) State evolution logic: together, these rules constitute the evolution logic of the base station state. They simulate interactions and competition between base stations and how decisions are made based on surrounding circumstances. Such state evolution logic helps network operators to better manage base station states to improve network performance and capacity.

In this embodiment, the cellular automaton selects the Moore neighborhood, which includes:

Specifically, in cellular automata, moore neighborhood is used to determine how the state of a cell at a certain moment is affected by the states of its surrounding cells. The state evolution rules for each cell are typically determined based on its own state and the states of neighboring cells within the Moore neighborhood. This neighborhood structure allows cellular automata to mimic various complex spatial and temporal patterns, as each cell interacts with its surrounding cells.

In particular, the use of Moore's neighborhood helps to simulate various natural and social phenomena, including propagation phenomena, fluid mechanics, urban traffic, etc. In the 5G base station capacity information analysis processing method, moore neighborhood is used to determine neighboring base stations around each base station in order to evaluate spectrum resource allocation and network capacity according to its status. This neighborhood structure provides an efficient way to model the interrelationship and impact between base stations.

Example eight

In this embodiment, the conversion function of the cellular automaton includes:

f＝(S _t+1 ，S _t )＝X _t+1

the evolution step of the transfer function comprises:

1) According to the current state matrix S _t To update and calculate the state matrix S for the next time step t+1 _t+1 The method comprises the steps of carrying out a first treatment on the surface of the In the 5G base station capacity information analysis process, the state matrix S represents the current state of each base station, including the spectrum resource allocation state, topology relationship information, and signal strength information. By updating the state matrix, the state evolution between base stations can be simulated, e.g. allocating spectrum resources to users or changing the topology.

2) According to the topological relation information, through the current state matrix S _t The information column (M+1 column) of the topological relation in the network, and determining the relation between each cell and the neighbor cells; the topology relationship information column (m+1 column) is used to determine the relationship of each base station to its neighbor base stations. This is critical in 5G networks because the interplay between base stations affects signal propagation and network capacity. Through the topological relation information, the connection relation between the base stations can be determined, so that the capacity analysis of the base stations is affected.

3) Application rules: by means of the array information of the state matrix S, the interaction and evolution of the cells under different states are described, namely based on the current state matrix S _t And a function of a neighbor relation; calculating a state matrix S of the next time step t+1 according to the application rule _t+1 The new state of each cell is next and stored in the next state matrix S _t+1 In (a) and (b); rules describe the interaction and evolution of the base station in different states. In 5G networks, these rules may relate to reallocation of spectrum resources, adjustment of signal propagation, etc. The calculation of the application rule is based on the current state matrix S _t And a neighbor relation to determine the new state of each base station at the next time step t+1.

4) Obtaining evolution data D comprising state matrix in all time steps, historical state and evolution track of each cell, and recording current state matrix S _t And the state matrix S of the next time step t+1 _t+1 The method comprises the steps of carrying out a first treatment on the surface of the The evolution data D includes a state matrix in all time steps, a history state of each base station, and an evolution track. Such data is useful for analyzing the changes and trends in 5G base station capacity and can help network operators optimize resource allocation and network configuration.

Exemplary: let the embodiment have a state matrix S _t It represents the current state of a group of base stations, each of which has three possible states: 0 (unassigned), 1 (assigned to user 1), 2 (assigned to user 2). The present embodiment now updates this state matrix according to some rules to obtain the state moment S of the next time step t+1 _t+1 . Example State matrix S _t (let it contain 3 base stations):

the rules are identical to those described above for the present embodiment, and for the first rule, the present embodiment checks the status of each base station in state 0 and its neighbor, and if there is a base station in state 1 in the neighbor, the status of the base station will become 1. After applying the rule, this embodiment obtains a temporary state matrix Temp1.

For the second rule, the present embodiment again checks the state of each base station in state 0 and its neighbors, and if a base station in state 2 is in the neighbor, the state of that base station will become 2. After applying the rules, this embodiment obtains a temporary state matrix Temp2.

For the third and fourth rules, the present embodiment examines each base station in states 1 and 2, and if most of their neighbors are also in the same state, their states will remain unchanged. After applying the rules, the embodiment obtains the final state matrix S _t+1 。

This is according to the rule f= (S _t+1 ，S _t )＝X _t+1 One example algorithm for updating the state matrix. In this way, the embodiment can simulate and describe the evolution of the base station state, and is helpful for 5G base station capacity information analysis and network optimization.

Example nine

In this embodiment, the Dempster's combination principle is a method for merging and reasoning about information of different evidences. In the analysis processing of the capacity information of the zone 5G base station, the damps-Bouldin index in Track-1 is used as evidence A and the evolution data in Track-2 is used as evidence B by using the Dempster's combination principle so as to comprehensively consider information from different sources.

In this example, the Davies-Bouldin index in Track-1 was used as evidence A, and the evolution data of Track-2 was used as evidence B, using the Dempster's combination principle:

evidence a: the Davies-Bouldin index was used as a result of Track-1 to measure compactness and segregation in clusters.

Evidence B: the evolution data in Track-2 includes a state matrix for each time step, a historical state for each cell, and an evolution trace.

1) Trust allocation function:

1.1 Trust allocation function for evidence a): this function is used to assign confidence levels to hypothetical cases for different capacity-amplified cases. In this scenario, it is assumed that the situation is about the possibility of different capacity increases. The confidence is inversely proportional to the Davies-Bouldin index, since the lower the Davies-Bouldin index, the better the clustering result. The trust allocation function is as follows: assigning credibility for evidence A to hypothetical cases of different capacity-increasing cases, the number of hypothetical cases being n, the trust allocation function being denoted m (A), where A represents the element m (A) of evidence A, m (A) _i Representing the confidence level of the ith hypothesis;

for evidence A, the confidence level is inversely proportional to the Davies-Bouldin index, let H be _i Indicating the ith capacity-increasing case, the confidence level m (A) _i Expressed as:

DB therein _i Represents the Davies-Bouldin index associated with the ith capacity increase event. Thus, the assumption that the Davies-Bouldin index is lower will result in higher confidence.

1.2 Trust allocation function for evidence B):

for evidence B, the present embodiment needs to build a trust allocation for each time step of each cell i. This relates to the trust function of the state matrix S of the cells (see below for a definition of trust function). The trust allocation function is expressed as:

Trust assignment m (B) at time step j for the ith cell _ij Expressed as:

m(B) _ij =trust function (S _ij )

S _ij Representing the state matrix S of cell i at time step j;

2) Dempster's combination principle:

once there is a trust allocation function, evidence A and evidence B can be combined using the Dempster's combination principle to obtain a comprehensive confidence level. The combination process is as follows: the combination for two evidence A and B is denoted m (AB), where A and B are the trust allocation functions m (AB) for evidence A and evidence B, respectively _ij ：

X represents all possible hypothesis sets;

y represents all possible evidence sets;

c is a subset of X and Y;

m(A) _C indicating the trustworthiness of evidence a on hypothesis set C.

KK is a normalization factor.

In this embodiment, the core idea of the Dempster's combination principle is to take into account the uncertainty of two different pieces of evidence, reducing the uncertainty by combining their confidence levels. In the area 5G base station capacity information analysis process, the method can be used for comprehensively considering the information of the clustering quality and the base station state so as to provide more comprehensive and reliable data and facilitate network optimization and resource allocation decision.

It should be noted that in the above procedure, the trust function is used to determine the trust level of each cell at a particular time step. The choice of the trust function depends on the specific application scenario and the problem. In the area 5G base station capability information analysis process, this trust function should reflect some attribute or feature of the cell state in order to better describe network state and performance. In this embodiment, the following six schemes of trust functions are exemplarily provided:

(1) Load case trustworth based on number of subscriber connections: the trust level function is inversely proportional to the number of user connections of the base station. When the number of users connected to one base station is small, the trust level is high, and vice versa. The formula can be expressed as follows:

this is added 1 to avoid zero denominator.

(2) Load situation trustworth based on spectrum resource utilization: the trust function is set to be inversely proportional to the spectrum resource utilization of the base station. When the spectrum resources of one base station are sufficient, the trust level is higher, and vice versa.

Here, 0.01 is added to prevent the denominator from being too small

(3) Load case trustworth based on data transfer delay: the trust function is set to be inversely proportional to the data transmission delay of the base station. When the data transmission delay of a base station is low, the trust level is high, and vice versa.

(4) Stability-based historical state trustworth: the trust level function is defined according to the connection stability of the base station in the past time step. If a base station is connected stable in the most recent time step, the confidence level is high and vice versa.

Here Stability Score is the base station connection Stability Score and Maximum Stability Score is the highest possible Score

(5) Historical state trustworth based on performance history: the trust level function is defined based on the performance history of the base station in the past time step. If a base station performs well in past time steps, the confidence level is high and vice versa.

Where Performance Score is the performance score and Maximum Performance Score is the highest possible score.

(6) Historical state trustworth based on historical data traffic: the trust level function is defined based on the history of data traffic at the base station in the past time step. If the data traffic of a base station is always kept at a high level, the confidence level is high and vice versa.

Here Historical Data Flow is the historical data traffic and Maximum Data Flow is the maximum possible data traffic.

In this embodiment, KK is a normalization factor used to ensure that the value of the trust function is within a reasonable range. Its specific value can be set according to the application scenario and form of trust function. The selection of KK requires balancing the relationship between the different trust function values to ensure that they are reasonably combined.

Exemplary:

(1)KK＝1：

application scene: if the value of the confidence function is already within the proper range, no additional normalization is required. This is typically used when the range of trust functions is already between 0, 1.

(2)KK＝0.1：

Application scene: when the value of the trust function is small and the difference between small values needs to be emphasized, a small KK value may be selected. For example, for a trust function of spectrum resource utilization, kk=0.1 may be used to emphasize different degrees in the case of resource shortage.

(3)KK＝10：

Application scene: when the value of the trust function is large and the difference between the large values needs to be emphasized, a larger KK value may be selected. For example, for a trust function of the number of user connections, kk=10 may be used to emphasize the performance of a base station of high connection number.

(4) Determining KK according to the actual data range:

application scene: the KK value may be determined according to a specific trust function and data scope in the application. For example, if the value of the confidence function ranges between [0,1000], then KK may be chosen to account for the confidence function's value being between [0,1], and kk=0.001 may be chosen.

In this embodiment, the Dempster's combination principle is of practical significance and importance in TracKK-3, as it allows for efficient combination of evidence from different sources to provide more comprehensive and trusted information, which is critical to the effectiveness of the 5G base station capability information analysis processing method. In TracKK-3, this example uses the Dempster's combination principle to combine two different types of evidence: evidence A (Davies-Bouldin index) and evidence B (evolution data in TracKK-2). The two types of evidence provide information on different aspects, the evidence A measures the compactness and the separation degree of the base station clusters, and the evidence B contains evolution data of the base station states.

Specific:

(1) Integrity and robustness: the Dempster's combination principle allows combining information from different sources together, providing more comprehensive, robust information. This helps to reduce misleading that may be caused by some type of evidence being inadequate or inaccurate.

(2) Reducing uncertainty: the Dempster's principle of combination can reduce uncertainty by combining different evidences. Evidence a and evidence B may complement each other to help determine the capacity augmentation at the next time step, thereby more accurately predicting network capacity demand.

(3) Decision support: the combined information may be used for decision making by the network operator. Based on the comprehensive information, operators can plan network resources better, optimize configuration and meet the requirements of different areas.

(4) Performance improvement: by combining the Davies-Bouldin index and evolution data together, network operators can better understand network performance, identify possible problems and bottlenecks, and take corresponding actions to improve network performance.

(5) The adaptability: the Dempster's combination principle can accommodate different network scenarios and changes as it allows for different types of evidence to be dynamically combined. This enables the method to perform well in different network deployments and environments.

Examples ten

In the present embodiment, the trust allocation function m (AB) is calculated by using a sigmoid function _ij Mapping to a probability value, then carrying out weighted average on each cell i, and calculating an estimated value of the capacity amplification condition of the cell i at the next time step j+1, thereby realizing the prediction of the capacity of the 5G base station. Thereby helping the staff to implement the regulation strategy.

It should be noted that, in this embodiment, the "cells" described below refer to corresponding base stations.

In this embodiment, the distribution function m (AB) will be trusted _ij Mapping the sigmoid function into a section value from 0 to 1, and representing the possibility of capacity amplification of a certain monoblock under the next time step j+1, thereby obtaining the amplification of the capacity of the 5G base station under the next time step:

(ij) is the input of a sigmoid function, exp is an exponential function;

2) For each cell i, P (ij) is weighted averaged using a weight function, where the weight (i) may be determined based on the importance of the cell or other factors. This weighted average yields an estimate of the capacity increase of cell i at the next time step j+1, a+ (ij+1):

A+(ij+1)＝∑ _i p (ij) weight (i)

A+ (ij+1) represents an estimate of the capacity increase of cell i (i.e., the corresponding base station) at the next time step j+1. This value reflects an estimate of the change in cell capacity based on the calculation of the trust allocation function M (AB) ij and Sigmoid function at a given time step.

Demonstrative: first, consider the form of a power function of e:

substituting the power function of e into the Sigmoid function:

3. taking the reciprocal of the score in the denominator:

it maps the input x to a probability value between 0 and 1. Next, the present embodiment considers weighted averaging of P (ij) and a+ (ij+1) = Σ _i Derivation of the weight (i):

for each cell i, the present embodiment has a probability value P (ij) indicating the probability that the cell i has capacity increased at time step j. The present embodiment introduces a weight function that represents the weight of each cell, which may be determined based on the characteristics of the cell or other factors. Let the weight of cell i be weight (i). Then, the present embodiment multiplies P (ij) by the weight (i), sums all the cells i, and obtains a+ (ij+1) = Σ _i P (ij). Weight (i).

Further illustratively, regarding a+ (ij+1) = Σ _i P (ij). Weight (i):

P(1j)＝0.8

P(2j)＝0.6

P(3j)＝0.4

meanwhile, the weight of each cell is set as follows:

weight (1) =0.3

Weight (2) =0.5

Weight (3) =0.2

The present embodiment will now use these parameters to calculate an estimate of the capacity increase at time step j+1:

a (1j+1) =p (1 j) ·weight (1) =0.8·0.3=0.24=24%

A (2j+1) =p (2 j) ·weight (2) =0.6·0.5=0.3=30%

A (3j+1) =p (3 j) ·weight (3) =0.4·0.2=0.08=8%

Thus, the present embodiment obtains an estimated value of the capacity increase condition of each cell at the time step j+1. At time step j+1, the capacity increase of cell 1 was estimated to be 24%, the capacity increase of cell 2 was estimated to be 30%, and the capacity increase of cell 3 was estimated to be 8%. These data are critical to the base station staff, as the regulation strategy for the base station capacity can be formulated in advance. For example:

(1) Resource allocation optimization: the base station staff can adjust the resource allocation strategy according to the estimated capacity amplification condition. For those cells with higher estimated capacity increases, more resources may be allocated to meet future capacity demands. In contrast, for cells with lower estimated capacity increase, resource allocation can be reduced to avoid resource waste.

(2) Capacity expansion plan: if the capacity increase estimate for some cells is very high, the base station operator may consider deploying additional base stations in these areas or performing capacity expansion to ensure that the network will meet the needs of the user in the future.

(3) Load balancing: the estimated capacity-amplified data may be used by base station operators for load balancing. If the capacity increase estimate for a cell is high, an attempt may be made to transfer some users from that cell to other cells with sufficient capacity to balance the network load.

(4) Network optimization: the estimated capacity-amplified data may be used by base station operators for network optimization. They can adjust antenna direction or signal strength to improve coverage or optimize signal modulation to increase capacity.

Further, in this embodiment, the weight (i) needs to be adjusted based on specific network performance and requirements. The weights can be changed according to different application scenarios and optimization objectives; it should follow any one of the following rules:

(1) Load balancing and optimizing: if the network wishes to achieve load balancing to ensure that the load distribution of the individual cells is uniform, the weight (i) may be set to the current load level of the cells. The higher the load of a cell, the higher its weight. This may help to transfer the user from a high load cell to a low load cell, enabling load balancing.

(2) Historical performance considerations: if the network wishes to take into account the historical performance of the cells in order to better accommodate future demand changes, the weights (i) may be set according to the historical performance data of the cells. For example, where the history of cells is better, their weights may be higher to better accommodate future capacity demands.

(3) Optimizing the resource utilization efficiency: if the optimization objective of the network is to maximize the resource utilization efficiency, the weights (i) may be set based on the resource utilization of the cells. Cells with higher resource utilization may obtain higher weights in order to more efficiently utilize network resources.

(4) And (3) guaranteeing service quality: the weights may be set according to Service Level Agreements (SLAs) or user requirements if the network needs to ensure the quality of service of certain specific areas or users. For areas or users requiring high quality of service, higher weights may be assigned to ensure that their capacity requirements are met.

(5) Regional importance: if certain regions are important to the overall performance of the network, weights may be set according to the importance of the regions. Important areas may be weighted higher to ensure that their capacity requirements are met preferentially.

Illustratively, the present embodiment provides several options for the weights (i) as follows:

(1) Load level parameters:

parameter values: the weight of the high load cell is 0.9, the weight of the medium load cell is 0.7, and the weight of the low load cell is 0.5.

Application scene: the method is used for load balancing, and ensures that more capacity is allocated in a high-load area so as to improve user experience.

(2) Historical performance parameters:

parameter values: the weight of the cells with good performance is 0.8, the weight of the cells with normal performance is 0.6, and the weight of the cells with poor performance is 0.4.

Application scene: for allocating capacity based on historical performance data to improve overall network performance.

Resource utilization efficiency parameters:

(3) Parameter values: the weight of the cells with high resource utilization efficiency is 0.9, the weight of the cells with general resource utilization efficiency is 0.7, and the weight of the cells with low resource utilization efficiency is 0.5.

Application scene: the method is used for maximizing the utilization efficiency of the resources and ensuring the effective utilization of the resources.

Service Level Agreement (SLA) parameters:

(3) Parameter values: the weight of the cells of the high SLA level area is 0.9, the weight of the cells of the medium SLA level area is 0.7, and the weight of the cells of the low SLA level area is 0.5.

Application scene: for allocating capacity to meet SLA commitments according to the requirements of different service levels.

Regional importance parameters:

(4) Parameter values: the weight of the cells in the important region is 0.9, the weight of the cells in the general important region is 0.7, and the weight of the cells in the non-important region is 0.5.

Application scene: the capacity requirements for ensuring critical areas are preferentially met.

(5) User density parameters:

parameter values: the weight of the cells in the high user density area is 0.9, the weight of the cells in the medium user density area is 0.7, and the weight of the cells in the low user density area is 0.5.

Application scene: for allocating capacity according to the user density difference to meet the demands of different areas.

Example eleven

Regional 5G base station capacity information analysis processing system

The system is an intelligent system for analyzing and processing the capacity information of the regional 5G base station. It is capable of performing the previous information analysis processing method by integrating the processor, memory and program instructions to help network operators better understand and optimize 5G network performance. The following are key components and functions of the system:

Further, regarding the program instructions:

the program instructions stored in the memory include detailed implementation steps of the regional 5G base station capability information analysis processing method. These program instructions are executed by a processor to perform the following key functions:

(1) And (3) data collection: the program instructions are responsible for collecting data from various data sources including base station capacity data, user demand data, evolution data, and the like.

(2) And (3) cluster analysis: using a cluster analysis algorithm, program instructions divide the data set into different clusters to distinguish between different performance patterns in the network.

(3) Cellular automaton evaluation: the program instructions treat the 5G base station topology as cells and apply a transfer function to calculate the evolution data of the base station at different time steps.

(4) D-S theory verifies: the trust of the individual data sources is verified using D-S theory and combined into an overall view to provide a more comprehensive estimate of network performance.

(5) Capacity prediction: based on the comprehensive information, the program instructions execute a prediction of the capacity of the 5G base station, including the capacity increase condition of the next time step.

Further, referring to fig. 2-5, there are shown the program instructions described above, which exhibit logic in the form of c++ pseudo code, the principle of which is as follows:

(1) cosinesimilityfunction: this function calculates the cosine similarity between the two vectors, which is used to measure the similarity between different base stations. Cosine similarity is a common metric used in K-means clustering to measure the distance between data points.

(2) kMeansCluster function: this function performs the main logic of the K-means clustering algorithm. It accepts as input a data set and the number of clusters k to be split into and performs the following steps:

S1, initializing a cluster center.

S2, an iterative clustering process is carried out, each base station is distributed to the nearest cluster, and the cluster center is updated until the maximum iterative times are reached.

S3, returning the distribution result of the cluster to which each base station belongs.

(3) main function: this is the entry point for the program. It defines an example base station dataset and invokes a kmeans clustering function to perform K-means clustering. Then, it outputs the cluster to which each base station belongs.

(4) the transitionFunction: the state transition function of the cellular automaton calculates a state matrix of the next time step according to the current state matrix and the neighbor relation. And performs the steps of:

s1, traversing each cell, calculating the states of surrounding neighbors, and updating the states of the cells according to the conversion rule.

S2, the conversion rule comprises the processing of the unassigned, assigned to the user 1 and assigned to the user 2 states, and the state of the cell at the next time step is determined according to the state of the neighbor.

S3, if the cell is not allocated, the cell is allocated according to whether the neighbor in the state of the user 1 or the user 2 exists in the neighbors.

S4, if the cell is allocated to the user 1 or the user 2, determining whether to keep the allocation state or change to the unallocated state according to the number of neighbors with the same state in the neighbors.

(5) sigmoid function: this is a sigmoid function that maps the input values to probability values between 0 and 1. sigmoid function is used to map the value of an element in the state matrix to a probability value between 0 and 1, which represents the estimated probability of the cell capacity increase.

(6) These key functions play an important role in the computation of cellular automata and capacity estimation. the transitionFunction is responsible for the evolution of the state, while the sigmoidFunction is responsible for mapping the state values into probabilities for computing an estimate of the capacity increase.

Summarizing, the present embodiment provides more comprehensive capacity analysis and prediction by enabling the system to process data in a 5G network on a large scale in an automated manner. The system is able to identify performance patterns in the network in order to better understand network behavior and optimize. By integrating multi-factor and considering multi-factor interactions, the system is able to more accurately predict future capacity needs. Based on the capacity prediction results, the system provides optimization suggestions that help network operators to better configure resources and networks.

In this embodiment, the regional 5G base station capacity information analysis processing system is a powerful tool that can be used for 5G network performance optimization and resource management. The method integrates a processor, a memory and program instructions, and can execute a detailed information analysis processing method, thereby helping network operators to better meet the demands of users, improving the network performance, optimizing the resource allocation and providing better services. The introduction of this system helps to drive the development and optimization of 5G networks.

The above examples merely illustrate embodiments of the invention that are specific and detailed for the relevant practical applications, but are not to be construed as limiting the scope of the invention. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the invention, which are all within the scope of the invention. Accordingly, the scope of protection of the present invention is to be determined by the appended claims.

Claims

1. The regional 5G base station capacity information analysis processing method is characterized by comprising the following Track-1 and Track-2 which are synchronously implemented:

track-2, cellular automaton evaluation: taking a topology gNB point of a 5G base station as a cell, wherein each cell has a spectrum resource attribute, the spectrum resource attribute is set as a state matrix S, a Moore neighborhood is adopted to determine a cell neighbor relation, and evolution of the gNB point under different time steps is measured and calculated through a conversion function to obtain evolution data D;

Also included is Track-3 implemented after the Track-1 and the Track-2:

2. The information analysis processing method according to claim 1, wherein: in Track-1, the data set includes the performance characteristics of the connection number C, the data traffic F, and the signal strength S, and the data set is constructed, and the data set is represented by a feature vector as:

sample 1 [ C1, F1, S1]

Sample 2 [ C2, F2, S2]

...

Sample N [ CN, FN, SN ]

Each of the samples in the dataset represents a base station;

measurement of the cosine similarity: for measuring similarity between different base stations, the cosine similarity cosinesimilitude between every two base stations includes:

3. The information analysis processing method according to claim 2, wherein: the/X/and the/Y/are L1 norm or L2 norm, wherein:

||X||1＝∣x ₁ ∣+∣x ₂ ∣+…+∣x _n ∣

||Y||1＝∣y ₁ ∣+∣y ₂ ∣+…+∣y _n ∣

the sum of the absolute values of the individual elements in the vector is the sum of the X/1 and Y/1, the X _n And said y _n Is the nth element of the vector X;

or alternatively, the first and second heat exchangers may be,

2) The L2 norm: by the square root of the sum of squares of the individual elements of the euclidean distance vector, for an n-dimensional vector x and y, we include:

the sum of the squares of the individual elements of the vectors, X2 and Y2, is the square root of the sum of the squares of the individual elements of the vectors, X _n And said y _n Is the nth element of the vector X.

4. The information analysis processing method according to claim 2, wherein: in the Track-1, the K-means clustering: dividing the base station into K clusters, and minimizing the sum of square errors of samples in the clusters:

the J is an objective function, the ni is the number of data points in the ith cluster, the x _j Is the jth data point in the ith cluster; i=1 denotes an index to a cluster; the j=1 represents an index to a data point within each cluster;

the c _i Is the cluster center, is a collection of data points; the cluster center c _i Is the average or centroid of all data points in the cluster, and the sum of the distances of all data points in the cluster to the center point is the smallest, including:

c _i ＝{X1,X2,…,Xn}

The data point x _j Assigned to cluster i, if the data point x _j To the cluster center c _i The distance of (2) is the smallest, the data point x _j Belonging to said cluster i;

the iterative process:

1) Cluster allocation: for the data point x _j Calculating the cluster center c with each cluster center _i Distance of the data point x _j Assigned to the cluster i to which the cluster center closest thereto belongs:

i＝argmin _k ||x _j -c _i || ²

2) Cluster update:

stopping iteration if the cluster center reaches the maximum iteration number, otherwise continuing cluster allocation and cluster updating; after the iteration is finished, each data point is distributed to the final cluster center, and the K-means clustering is completed.

5. The information analysis processing method according to claim 4, wherein: the Davies-Bouldin index comprises:

1) The compactness: for each cluster i, calculating the average similarity between each pair of data points in the cluster, which is expressed as the average similarity S of the data points in the cluster _i ：

M _ij ＝sim(x _j ，c _i )

3) Calculating the Davies-Bouldin index:

the R is _i A ratio representing the compactness of the cluster i and the separability of other clusters; the S is _j Representing the average similarity between each pair of data points within cluster j;

the max is _j≠i Representing the selection least similar to cluster iTo measure the degree of separation between the cluster i and other clusters;

in the calculation of the Davies-Bouldin index, for each cluster i, in order to measure the degree of separation between the cluster i and other clusters, a similarity measure with other clusters j is calculated, and then the cluster j with the smallest similarity measure with the cluster i is selected.

6. The information analysis processing method according to claim 1, wherein: in the Track-2:

the cells have spectrum resource attributes, and the spectrum resource attributes are one state matrix S; the state matrix includes: n cells i and M users, the dimension of the state matrix S is N x (m+2), where each row represents the state of one cell, and each column includes the following information:

the state matrix S is:

the Moore neighborhood:

the CC represents a central cell, and the NO, the NE, the EA, the SE, the SO, the SW, the WE, and the NW represent surrounding eight cells.

7. The information analysis processing method according to claim 6, wherein: the transfer function includes:

f＝(S _t+1 ，S _t )＝X _t+1

wherein the S is _t Is the current state matrix at time step t, S _t+1 Is the state matrix at time step t+1; the evolution step of the transfer function comprises:

1) According to the current state matrix S _t To update and calculate said state matrix S for the next said time step t+1 _t+1 ；

2) According to the topological relation information, through the current state matrix S _t The array information of the topological relation in (a) determines the relation between each cell and the neighbor cells;

3) Application rules: setting interaction and evolution of the cells under different states through the array information of the state matrix S; calculating the state matrix S of the next time step t+1 according to the application rule _t+1 The new state of each cell is next stored in the next state matrix S _t+1 In (a) and (b);

4) Obtaining the evolution data D, including state matrix, history state and evolution track of each cell in all time steps, and recording the current state matrix S _t And the state matrix S of the next time step t+1 _t+1 。

8. The information analysis processing method according to claim 1, wherein: in the Track-3: the Dempster's combination principle uses the Davies-Bouldin index in Track-1 as evidence A and uses the evolution data of Track-2 as evidence B:

1) Trust allocation function:

1.1 The trust allocation function for the evidence a): assigning trustworthiness for said evidence a to hypothetical cases of different capacity amplification cases, said falseLet the number of cases be n, the trust allocation function be denoted m (A), where A represents the element m (A) of evidence A, m (A) _i Representing the confidence level of the ith hypothesis;

for evidence A, the confidence level is inversely proportional to the Davies-Bouldin index, let H be _i Indicating the i-th capacity-increasing condition, DB _i Representing the Davies-Bouldin index associated therewith, the confidence level m (A) _i Expressed as:

1.2 For evidence B) the trust allocation function:

i said cells and a number of time steps, said trust allocation function for said evidence B will construct a trust allocation for each time step of each said cell:

trust assignment m (B) at time step j for the ith cell _ij Expressed as:

m(B) _ij =trust function (S _ij )

S _ij Representing the state matrix S of the cell i at time step j;

2) Dempster's combination principle:

the combination for two evidence A and B is denoted m (AB), where A and B are the trust allocation functions m (AB) for evidence A and evidence B, respectively _ij ：

The X represents all hypothesis sets;

the Y represents all evidence sets;

said C is a subset of X n Y;

the m (A) _C Representing the trustworthiness of evidence a on hypothesis set C;

and KK is a normalization factor.

9. The information analysis processing method according to claim 8, wherein: assigning the trust allocation function m (AB) _ij Mapping the sigmoid function into a section value from 0 to 1, and further obtaining the amplification condition of the capacity of the 5G base station in the next time step:

1) For each said m (AB) _ij Mapping it to a probability value using the sigmoid function:

said P (ij) represents the probability of a 5G base station capacity increase situation for said cell i at time step j; the (ij) is the input of a sigmoid function, and the exp is an exponential function;

2) For each of said cells i, calculating an estimate of its capacity increase at the next time step j+1 by weighted averaging of said P (ij):

A+(ij+1)＝∑ _i p (ij) weight (i)

The a+ (ij+1) represents an estimated value of the capacity increase condition of the cell i at the next time step j+1.

10. A regional 5G base station capacity information analysis processing system is characterized in that: the system includes a processor, a memory coupled to the processor, the memory having stored therein program instructions that, when executed by the processor, cause the processor to perform the information analysis processing method of any of claims 1-9.