WO2024067093A1 - Communication load forecasting method and apparatus, device, and storage medium - Google Patents

Abstract

The present disclosure provides a communication load forecasting method and apparatus, a device, and a storage medium. The method comprises: acquiring real-time communication load data of a first cell; and obtaining a communication load forecasting result on the basis of the real-time communication load data of the first cell and a communication load forecasting model corresponding to a target cell cluster to which the first cell belongs, the communication load forecasting model corresponding to the target cell cluster being constructed on the basis of historical communication load data of at least one cell in the target cell cluster.

Description

Communication load prediction method, device, equipment and storage medium

This disclosure claims priority to Chinese patent application No. 202211200891.6, filed on September 29, 2022, the entire contents of which are incorporated by reference into this application.

Technical Field

The present disclosure relates to the field of communication technology, and in particular to a communication load prediction method, device, equipment and storage medium.

Background technique

Wireless communication technology has developed rapidly in recent years and has become an important part of human life. In order to ensure the normal operation of the wireless communication system, it is necessary to formulate a load control strategy for the wireless communication system. Load control is to monitor the load of the wireless communication system in real time during the operation of the wireless communication system (that is, to continuously monitor the users (or services) that have been connected to the wireless communication system in real time). When it is detected that the communication load of the wireless communication system is too high and affects the stable operation of the wireless communication system, the load control strategy is implemented to reasonably allocate communication resources to ensure the stable operation of the wireless communication system.

Summary of the invention

In a first aspect, an embodiment of the present disclosure provides a communication load prediction method. The method includes:

Acquire real-time communication load data of the first cell;

A communication load prediction result is obtained based on the real-time communication load data of the first cell and the communication load prediction model corresponding to the target cell cluster to which the first cell belongs. The communication load prediction model corresponding to the target cell cluster is constructed based on the historical communication load data of at least one cell in the target cell cluster.

In a second aspect, an embodiment of the present disclosure provides a communication load prediction device. The device includes: a communication unit, configured to obtain real-time communication load data of a first cell;

A processing unit is used to obtain a communication load prediction result based on real-time communication load data of the first cell and a communication load prediction model corresponding to a target cell cluster to which the first cell belongs. The communication load prediction model corresponding to the target cell cluster is constructed based on historical communication load data of at least one cell in the target cell cluster.

In a third aspect, an embodiment of the present disclosure provides an electronic device, which includes: a processor and a memory, wherein the memory stores instructions executable by the processor, and when the processor is configured to execute the instructions, the electronic device implements the method provided in the first aspect above.

In a fourth aspect, an embodiment of the present disclosure provides a computer-readable storage medium. The computer-readable storage medium stores computer instructions, and when the computer instructions are executed on a computer, the computer executes the method provided in the first aspect.

In a fifth aspect, an embodiment of the present disclosure provides a computer program product comprising computer instructions, which, when executed on a computer, enables the computer to execute the method provided in the first aspect.

BRIEF DESCRIPTION OF THE DRAWINGS

The accompanying drawings are used to provide a further understanding of the technical solution of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the technical solution of the present invention and do not constitute a limitation on the technical solution of the present invention.

FIG1 is a schematic diagram of the structure of a communication system according to some embodiments;

FIG2 is a flow chart of a communication load prediction method according to some embodiments;

FIG3 is a flow chart of another communication load prediction method according to some embodiments;

FIG4 is a flow chart of another communication load prediction method according to some embodiments;

FIG5 is a schematic diagram of an elbow method positioning according to some embodiments;

FIG6 is a flow chart of another communication load prediction method according to some embodiments;

FIG7 is a flow chart of another communication load prediction method according to some embodiments;

FIG8 is a flow chart of another communication load prediction method according to some embodiments;

FIG9 is a flow chart of another communication load prediction method according to some embodiments;

FIG10 is a flow chart of another communication load prediction method according to some embodiments;

FIG11 is a schematic diagram showing the composition of a communication load prediction device according to some embodiments;

FIG. 12 is a schematic structural diagram of an electronic device according to some embodiments.

Detailed ways

In order to enable those skilled in the art to better understand the technical solutions of the embodiments of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present disclosure.

In the description of the present disclosure, unless otherwise specified, "/" means "or", for example, A/B can mean A or B. "And/or" in this article is merely a description of the association relationship of associated objects, indicating that there may be three relationships, for example, A and/or B can mean: only A, only B, and A and B. The terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present disclosure, unless otherwise specified, "at least one" means one or more, and "multiple" means two or more.

In the description of the present disclosure, it should be noted that, unless otherwise clearly specified and limited, the terms "connected" and "connection" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection. For ordinary technicians in this field, the meaning of the above terms in the present disclosure can be understood according to the situation. In addition, when describing the pipeline, the "connected" and "connection" used in the present disclosure have the meaning of conduction. The meaning needs to be understood in conjunction with the context.

In the embodiments of the present disclosure, words such as "exemplarily" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "exemplarily" or "for example" in the embodiments of the present disclosure should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplarily" or "for example" is intended to present related concepts in a detailed manner.

With the development of wireless communication technology, especially with the continuous evolution of the fourth generation of mobile communication technology (4G) and the fifth generation of mobile communication technology (5G), wireless communication systems are facing more and more serious load problems. The load of a wireless communication system is an important indicator for measuring the network operation capability. When the load of a wireless communication system reaches a certain threshold, there will be a capacity bottleneck, resulting in network delay or packet loss. In order to ensure the user's perception of network usage, it is necessary to implement a load control strategy for the wireless communication system, so that when the wireless communication system is about to be overloaded, an alarm prompt can be issued in time and the network can be optimized, thereby alleviating the problem of high load of the wireless communication system.

When formulating a load control strategy, it is necessary to predict the communication load of the wireless communication system, and then formulate a load control strategy based on the predicted results of the communication load. Currently, there are two ways to predict the communication load of the wireless communication system: one is for network managers to predict based on manual experience. The accuracy of the communication load prediction in this solution is not high. Another solution is to establish a communication load prediction model for each cell of the wireless communication system. In this solution, the decision is too detailed, resulting in excessive consumption of computing resources and increased management difficulty and cost. Therefore, how to increase the communication load Improving prediction accuracy and reducing the consumption of computing resources are issues that need to be addressed urgently.

Based on this, the embodiment of the present disclosure provides a communication load prediction method, which obtains the real-time communication load data of a cell, and then obtains the communication load prediction result of the cell based on the real-time communication load data of the cell and the communication load prediction model corresponding to the target cell cluster to which the cell belongs. Compared with the communication load prediction result obtained by the management personnel based on manual experience, the accuracy of the communication load prediction is improved. At the same time, since the above-mentioned communication load model is constructed based on the historical communication load data of at least one cell in the target cell cluster, that is, each cell in the target cell cluster can share the communication load prediction model corresponding to the cell cluster to predict the communication load, there is no need to establish a communication load prediction model corresponding to each cell separately, which reduces the consumption of computing resources.

In some embodiments, the technical solution of the present disclosure can be applied to various communication systems. For example: global system for mobile communications (GSM), code division multiple access (CDMA) system, wideband code division multiple access (WCDMA) system, long term evolution (LTE) system, new radio (NR) system, frequency division duplexing (FDD) system, time division duplexing (TDD) system, etc. For the convenience of description, various communication systems to which the technical solution of the present disclosure can be applied are illustrated by taking the communication system shown in FIG. 1 as an example.

FIG1 is a schematic diagram of the structure of a communication system according to an exemplary embodiment. The communication system includes a communication load prediction device 10, multiple base stations (such as base station 21 and base station 22) and multiple terminal devices (such as terminal device 31, terminal device 32, terminal device 33 and terminal device 34). The communication load prediction device 10, multiple base stations and multiple terminal devices can be connected through a wired network or a wireless network. The wired network or wireless network may include a router, a switch, a base station, or other devices that facilitate communication between the communication load prediction device 10, multiple base stations and multiple terminal devices, which is not limited by the embodiments of the present disclosure.

In some embodiments, the communication load prediction device 10 may obtain information of multiple base stations. For example, the type of each base station, the number of multiple base stations, the number of cells served by a base station, the time series data generated by each cell in the process of the base station providing network services to the cells served by the base station, etc. After obtaining the time series data of each cell, the communication load prediction device 10 may cluster each cell based on the time series data of each cell to obtain multiple cell clusters, and then establish a communication load prediction model corresponding to each cell cluster.

In some embodiments, the communication load prediction device 10 may be a computer device or a server. The server may be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms. In some embodiments, the communication load prediction device 10 is a device including a display screen, which can display the clustering results of the cell clusters and the communication load prediction results of the cell when predicting the communication load of the cell.

In some embodiments, the communication load prediction device 10 carries a network management system, which is used to manage multiple base stations connected to the communication load prediction device 10. For example, the network management system is used to collect time series data generated by each cell in the process of each base station providing network services to the cell served by the base station.

In some embodiments, the base station is used to provide wireless access services for terminal devices. Specifically, each base station provides a service coverage area (also called a cell). Terminal devices entering the area can communicate with the base station via wireless signals. Accept the wireless access service provided by the base station. The service coverage areas of the base stations may overlap, and the terminal devices in the overlapping areas can receive wireless signals from multiple base stations.

In some embodiments, each of the multiple base stations can be connected to multiple terminal devices. For example, base station 21 is connected to terminal device 31 and terminal device 32. Terminal device 31 and terminal device 32 can be located in the same cell, or terminal device 31 and terminal device 32 can be located in different cells. That is, a base station can provide network services to a terminal device in one cell, or can provide network services to terminal devices in multiple cells at the same time.

In some embodiments, each of the multiple base stations (e.g., base station 21) may be an evolution node B (eNB), a next generation node B (gNB), a transmission receive point (TRP), a transmission point (TP), and any other access node. According to the size of the service coverage area provided, the base station can be divided into a macro base station for providing macro cells (Macro cells), a micro base station for providing micro cells (Pico cells), and a femto base station for providing femto cells (Femto cells). With the continuous evolution of wireless communication technology, future base stations may also adopt other names.

In some embodiments, the base station may be referred to as a network element. The above communication system may also include other network elements, such as a mobility management entity (MME) network element, a serving gateway (SGW) network element, etc.

In some embodiments, each of the multiple terminal devices (e.g., terminal device 31) may be a device with wireless transceiver function, such as a mobile phone, a tablet computer, a wearable device, a vehicle-mounted device, an augmented reality (AR)/virtual reality (VR) device, a notebook computer, an ultra-mobile personal computer (UMPC), a netbook, a personal digital assistant (PDA), etc. The disclosed embodiments do not limit the type of terminal device.

It is understood that FIG1 is an exemplary structural diagram, and the number of devices included in the communication system shown in FIG1 is not limited. For example, the number of base stations is not limited, and the number of terminal devices is not limited. In addition, in addition to the devices shown in FIG1, the communication system shown in FIG1 may also include other devices, which is not limited in the embodiments of the present disclosure.

Next, Fig. 2 is a flow chart of a communication load prediction method according to an exemplary embodiment, the method is performed by a communication load prediction device. The communication load prediction device may be the communication load prediction device 10 in the communication system shown in Fig. 1, and the method includes the following steps.

S101. Obtain real-time communication load data of a first cell.

In some embodiments, when formulating a load control strategy for a cell (eg, a first cell), the communication load prediction device may obtain real-time communication load data of the first cell from a base station to which the first cell belongs through a network management system.

In some embodiments, each base station or network element in the communication system shown in Figure 1 above periodically reports real-time communication load data of at least one first cell it serves to the communication load prediction device, so that when formulating a load control strategy for a first cell, the communication load prediction device can quickly obtain the real-time communication load data of the first cell.

Real-time communication load data includes one or more of the following data types: number of new radio (NR) carrier radio resource control (RRC) connections, NR carrier uplink physical resource block (PRB) utilization rate, NR carrier downlink PRB utilization rate, cell group uplink PRB utilization rate, cell group downlink PRB utilization rate, long term evolution (LTE) dynamic spectrum sharing (DSS) cell group uplink PRB utilization rate, LTE DSS cell group downlink PRB utilization number, LTE DSS cell group RRC connection number, LTE cell uplink PRB utilization number, LTE cell downlink PRB utilization number and LTE cell RRC connection number, etc.

S102: Obtain a communication load prediction result based on the real-time communication load data of the first cell and a communication load prediction model corresponding to the target cell cluster to which the first cell belongs.

In some embodiments, the communication load prediction device pre-stores a communication load prediction model corresponding to the cell cluster to which each first cell belongs and a first correspondence. The first correspondence includes a correspondence between identifiers of multiple first cells and identifiers of multiple cell clusters. An identifier of a first cell is used to uniquely indicate a first cell, for example, it can be the name of the first cell, and an identifier of a cell cluster is used to uniquely indicate a cell cluster, for example, it can be the name of the cell cluster.

In some embodiments, a cell cluster includes at least one cell.

After obtaining real-time communication load data of a first cell, the communication load prediction model can determine the identifier of the cell cluster corresponding to the identifier of the first cell based on the identifier of the first cell and the first corresponding relationship, and then use the cell cluster corresponding to the identifier of the cell cluster as the target cell cluster.

After determining the target cell cluster, the real-time communication load data of the first cell can be input into the communication load prediction model corresponding to the target cell cluster to which the first cell belongs to obtain a communication load prediction result. The communication load prediction result is used to characterize the communication load change trend of the first cell at a future time, or the communication load prediction result is used to characterize the communication load data of the first cell at a future time.

Exemplarily, if the real-time communication load data of a first cell is the cell group uplink PRB utilization, the real-time cell group uplink PRB utilization of the first cell is output to the communication load prediction model, and the changing trend of the cell group uplink PRB utilization of the first cell at future times or the cell group uplink PRB utilization of the first cell at future times can be obtained.

In some embodiments, the communication load prediction model corresponding to the target cell cluster is constructed based on historical communication load data of at least one cell in the target cell cluster. Regarding how to construct the communication load prediction model corresponding to the target cell cluster based on historical communication load data of at least one cell in the target cell cluster, reference can be made to the description of S401 to S402 below, which will not be repeated here.

Based on the embodiment shown in FIG2 , at least the following beneficial effects are brought about: a communication load prediction method provided by the embodiment of the present disclosure obtains the real-time communication load data of a first cell, and then obtains the communication load prediction result based on the real-time communication load data of the first cell and the communication load prediction model corresponding to the target cell cluster to which the cell belongs. Compared with the communication load prediction result obtained by the network management personnel based on manual experience, the accuracy of the communication load prediction is improved. So that when formulating a load control strategy for the first cell, the load control strategy of the first cell can be formulated according to the communication load prediction result with higher accuracy of the first cell, which improves the rationality of the formulation of the load control strategy. And the above-mentioned communication load prediction model is constructed based on the historical communication load data of at least one cell in the target cell cluster, that is, at least one cell in the target cell cluster can share the communication load prediction model for communication load prediction, and there is no need to establish a communication load prediction model corresponding to each cell separately, which improves the accuracy of the communication load prediction while reducing the consumption of computing resources.

The above embodiments focus on the use process of the communication load prediction model corresponding to the target cell cluster. In some embodiments, a communication load prediction method provided by the embodiments of the present disclosure also includes a process for determining the cell cluster. As shown in FIG3 , the method also includes the following steps.

S201. Obtain historical communication load data of multiple second cells.

In one embodiment, before establishing a communication load prediction model corresponding to a cell cluster, the communication load prediction device needs to obtain historical communication load data of multiple second cells, so as to predict the communication load of multiple second cells according to the historical communication load data of the multiple second cells. The second cells are clustered to determine the cell cluster to which each second cell belongs, and then a communication load prediction model corresponding to the cell cluster is established for each cell cluster. The first cell is one of the plurality of second cells.

Exemplarily, the communication load prediction device may collect historical communication load data of at least one cell served by each base station or network element from each base station side, base station portrait or network element in the communication system shown in FIG1 at a granularity of CollectDataStep (unit/minute) through the network management system. The collection duration may be CollectDataTime (unit/day).

In some embodiments, the plurality of second cells may be cells located in the same area or in different areas. The area may be a street, a township, a municipal district or a city, and the embodiments of the present disclosure do not limit the scope of the division of the area. The cells may be cells of different types. For example, the types of cells may be colleges and universities, subway platforms, subway lines, ground road transportation hubs, residential areas, commercial areas, urban villages, parks, office buildings and large supermarkets, and the embodiments of the present disclosure do not limit this.

In some embodiments, for each second cell among multiple second cells, the historical communication load data of a second cell may be time series data after a communication load prediction device collects the communication load data of the second cell at multiple historical moments and pre-processes the communication load data of the second cell at multiple historical moments.

Exemplarily, the communication load device pre-processes the communication load data of a second cell at multiple historical moments, including one or more of the following.

1) Data timeline processing.

It is understandable that during the data collection process, some data may be missed at certain granularities. If some data on the time axis is missing, the data time axis is filled, and the corresponding filled data is empty.

In some embodiments, after filling the data time axis, the communication load prediction device can also perform data time axis deduplication. It is understandable that during the data collection process, repeated collection may occur within certain granularities, and the repeatedly collected data needs to be deduplicated.

In some embodiments, the deduplication rule may be to retain the data that appears for the first time in the data set and delete the duplicate data after the first appearance. The deduplication rule may also include the data that appears for the last time in the data set and delete the duplicate data before the last appearance. The embodiments of the present disclosure do not limit the setting of the deduplication rule.

2) Data completion processing.

In some embodiments, if the missing data is at the head of the data set, the empty data at the head will be uniformly filled with the first non-empty data from the head; if the missing data is at the tail of the data set, the empty data at the tail will be uniformly filled with the first non-empty data from the tail; if the missing data is in the middle of the data set, the first non-empty data can be searched forward and backward respectively, and linear interpolation filling can be performed (mean filling and other schemes can also be used).

3) Sequence alignment and truncation processing.

In some embodiments, since the communication load of the cell is repeated in cycles based on weeks, the communication load data needs to be aligned according to the same relative time starting point between weeks (for example, all collection starts from Monday), and the load data CollectDataTime_aligned of a preset period (for example, two weeks) is retained as a training data set.

4) Data vectorization processing.

In some embodiments, the communication load prediction device removes the communication load data with the time axis mark from the time axis and expands it into a length of The one-dimensional data samples are vectorized and the time series data of the cell is obtained.

The communication load prediction device predicts the communication load data of each second cell in the plurality of second cells at a plurality of historical moments. By performing the above preprocessing, the time series data of each second cell can be obtained, that is, the historical communication load data of each second cell can be obtained.

S202: Cluster the multiple second cells based on historical communication load data of the multiple second cells to obtain at least one cell cluster.

In some embodiments, as shown in FIG. 4 , S202 may be implemented as the following steps, for example.

S2021. Determine the number of cell clusters based on historical communication load data of multiple second cells.

In some embodiments, after acquiring the historical communication load data of the plurality of second cells, the communication load prediction device may determine the number of cell clusters based on the historical communication load data of the plurality of second cells and a preset cluster number determination algorithm. The preset cluster number determination algorithm includes an elbow method, a silhouette coefficient method, a cluster evaluation index (Calinski Harabasz) method, and other cluster number determination algorithms.

Exemplarily, the embodiment of the present disclosure takes the preset cluster number determination method as the elbow method as an example for illustration.

As shown in Figure 5, the embodiment of the present disclosure adopts the elbow method to locate the critical point of the curve of the sum of squares error (SSE) curve by increasing the number of clusters to reduce the intra-cluster variance, that is, the critical point where increasing or decreasing the number of clusters can no longer obtain a significant partitioning benefit, thereby determining the number of cell clusters to facilitate subsequent clustering of multiple second cells.

In some embodiments, determining the number of cell clusters according to the elbow method and historical communication load data of a plurality of second cells may include the following steps.

A1. Calculate the intra-cluster variance sum.

It can be understood that the purpose of clustering is to cover as many samples as possible under the premise of having the strongest generalization ability, so the embodiment of the present disclosure uses SSE for measurement, and SSE can be shown as the following formula.

In the above formula, k is the number of clusters, C is the number of clusters, _Ci is the i-th cluster among C clusters, _mi is the cluster center of the i-th class, and j is the data point of the i-th class.

As the number of classification clusters increases, the intra-cluster variance and the sum will gradually decrease, and the rate of decrease will change. The disclosed embodiment uses the number of clusters corresponding to the peak point of the slowing trend of the intra-cluster variance and the increase of the number of clusters as the number of clusters of the clustering algorithm (ie, the number of cell clusters).

In some embodiments, the intra-cluster variance and intra-cluster density may be replaced by other measurement methods including but not limited to intra-cluster average Euclidean distance, intra-cluster density, and the like.

A2. Determine the number of cell clusters based on the elbow method diagram.

The elbow method is a general method for determining the optimal number of clusters. The main idea of the elbow method is to increase the number of clusters until the moment of diminishing returns stops, which is reflected in the graph as an elbow with a sudden increase in curvature.

The calculation method of the elbow method includes: first calculate the number from 1 to ( In order to obtain the upper limit of the range of cluster numbers, n can be the number of at least one first cell, or it can be pre-set by the management personnel) and the corresponding intra-cluster variances, and then the number of clusters corresponding to the point where the curve descends most slowly, that is, the point where the curvature of the elbow method image is the largest, is taken as the number of cell clusters.

Therefore, the number of cell clusters can be obtained through the above A1 and A2.

S2022. For any two cells among the multiple second cells, determine the distance between the two second cells based on historical communication load data of the two cells.

In some embodiments, after determining the number of cell clusters, it is also necessary to determine the distance between any two cells in the plurality of second cells, and define a distance metric to measure the intra-class similarity difference and the inter-class difference, so as to obtain the maximum intra-class similarity and the minimum inter-class difference. The clustering result with the smallest similarity between classes is the one with the largest difference.

The distance between the two cells is used to characterize the similarity between the communication load variation trends of the two second cells.

In some embodiments, the communication load prediction device can determine the distance between any two second cells based on historical communication load data of multiple second cells and a preset distance algorithm. The preset distance algorithm includes distance algorithms such as the Euclidean distance (ED) algorithm, the dynamic time warping distance (DTW) algorithm, the Hausdorff distance algorithm, the hidden Markov model distance (HMM Distance) algorithm, and the longest common subsequence distance (LCS Distance) algorithm.

Exemplarily, the embodiments of the present disclosure are described by taking the preset distance algorithms as the Euclidean distance algorithm and the DTW distance algorithm as examples.

In some embodiments, when the preset distance algorithm is a Euclidean distance algorithm, determining the distance between any two cells in the plurality of second cells based on historical communication load data of the plurality of second cells and the preset distance algorithm may include the following steps.

B1. Time series intercepted by equal length.

It can be understood that the calculation of the Euclidean distance between sequences in the field of time series is a distance measurement between corresponding point pairs at the same timestamp, so the sequences must be of equal length to calculate the Euclidean distance between any two cells.

Exemplarily, for two time series of unequal lengths Seq ₁ =(x ₁ , x ₂ ,…, x _n ) and Seq ₂ =(y ₁ , y ₂ ,…, y _m ), when m and n are both integers greater than 1 and m is greater than n, in order to ensure that the two sequences are equal in length, the first n sample points of the long sequence Seq ₂ are retained so that Seq ₂ =(y ₁ , y ₂ ,…, _yn ), making Seq ₂ the same length as Seq ₁ .

B2. Calculate the Euclidean distance.

Exemplarily, taking Seq ₁ as cell 1 and Seq ₂ as cell 2 as an example, the Euclidean distance between cell 1 and cell 2 can be expressed by the following formula.

In the above formula, Dist_euclidean is the Euclidean distance between cell 1 and cell 2, that is, the Euclidean distance between any cells.

In some embodiments, when the preset distance algorithm is a DTW distance algorithm, determining the distance between any two cells in the plurality of second cells based on historical communication load data of the plurality of second cells and the preset distance algorithm may include the following steps.

C1. Calculate the distance matrix.

For two time series, the DTW distance sets aside the limitation of the Euclidean distance. The idea is to find a continuous, mutually corresponding matching relationship that includes all the points in the two time series (this matching can be the i-th point corresponding to the j-th point). The DTW distance is more accurate for waveform fitting. The disclosed embodiment still uses the distance between each pair of "points" in the two sequences to calculate the similarity, even if the number of points in the two sequences may be different. However, because the time axis can be warped, we do not take a pair of points in the two sequences in turn to calculate the distance, but each point may correspond to multiple points in the other sequence. Each point must be used and cannot be skipped. It must be in the original order, and the point pairs cannot cross. Dynamic programming is usually used to complete the calculation.

Exemplarily, for two time series of unequal lengths Seq ₃ =(x ₁ ,x ₂ ,..., _xi ) and Seq ₄ =(y ₁ ,y ₂ ,...,y _j ), when i and j are both integers greater than 1 and i is greater than j, the distance matrix between each point in the two time series is first calculated.

C2. Dynamic programming to find the optimal solution.

Find a path from the upper left corner to the lower right corner of the matrix so that the sum of the elements on the path is minimized. This is a dynamic programming algorithm with the starting condition being In the above formula, is the distance between two time series, M is the distance matrix. The recursive rule is expressed by the following formula.

Then the DTW distance between the two time series is obtained. The above two time series are understood as two cells, that is, the DTW distance between the two cells is obtained.

By performing the above-mentioned processing of B1 and B2, or the processing of C1 and C2 on any two cells among the multiple cells, the distance between any two cells among the multiple second cells can be obtained.

It should be noted that, for the above S2021 and S2022, S2021 may be executed first and then S2022; S2022 may be executed first and then S2021; S2021 and S2022 may also be executed simultaneously. The embodiment of the present disclosure does not limit the execution order of S2021 and S2022.

S2023: Based on the number of cell clusters and the distance between any two cells in the plurality of second cells, cluster the plurality of second cells to obtain at least one cell cluster.

In some embodiments, after determining the number of cell clusters and the distance between any two cells among a plurality of second cells, the plurality of second cells can be clustered based on the number of cell clusters, the distance between any two cells among a plurality of second cells, and the historical communication load data of the plurality of second cells, in combination with a preset clustering algorithm, to obtain at least one cell cluster.

In some embodiments, the preset clustering algorithm includes a k-means clustering algorithm and a k-shape clustering algorithm, and at least one cell cluster may be at least one cell cluster with the largest similarity and the smallest difference within the cluster and the smallest similarity and the largest difference between clusters. A cell cluster includes at least one second cell.

In some embodiments, the k-shape clustering algorithm is an efficient and accurate time series clustering method. The k-shape clustering algorithm optimizes the distance calculation method, the centroid calculation method, and introduces a method for extracting frequency domain features. It can improve the computational efficiency of time series while supporting amplitude scaling and translation invariance. The k-shape clustering algorithm uses a shape-based distance (SBD) based on cross-correlation measurement to identify the same pattern while ignoring the differences in amplitude and phase.

In some embodiments, when the preset clustering algorithm is the k-means clustering algorithm, the clustering process includes the following steps.

D1. Randomly initialize and generate clusters.

Make sure to use the same random seed to generate random numbers to ensure the stability of the initial clustering results each time, that is, each experiment has the same initial cluster selection and distribution.

D2. Data normalization.

The TimeSeriesScalerMeanVariance method is used to normalize the data into a standard distribution with a mean of 0 and a standard deviation of 1, eliminating the influence of the dimension and facilitating the calculation of cross-correlation.

D3. k-means clustering.

K cluster centers are initialized according to the number of cell clusters k, and the distance between any two cells in the multiple second cells (such as Euclidean distance) is used as a metric to perform k-means clustering on the historical communication load data of all second cells in the training set, and the training is performed until convergence.

In some embodiments, the clustering process may further include: D4, cluster visualization, and calculation of mutual correlations.

It is understandable that the clustering process is visualized to facilitate staff to observe the clustering situation. The number and type of cells in each cluster (such as the type of cells described in S201) are counted during the clustering process, and the correlation between the clustering results is calculated to measure the Measures the accuracy of the k-means clustering algorithm on the data set.

In some embodiments, when the preset clustering algorithm is a k-shape clustering algorithm, the clustering process includes the following steps.

E1. Randomly initialize and generate clusters.

Make sure to use the same random seed to generate random numbers to ensure the stability of the initial clustering results each time, that is, each experiment has the same initial cluster selection and distribution.

E2. Use the TimeSeriesScalerMeanVariance method to normalize the data into a standard distribution with a mean of 0 and a standard deviation of 1 to eliminate the impact of the dimension and facilitate the calculation of cross-correlation.

E3. k-shape clustering.

According to the number of clusters k, k cluster centers are initialized, and the shape-based distance (SBD) distance based on cross-correlation measurement is used as the metric to perform k-shape clustering on the time series data of multiple first cells in the training set, and the training is carried out until convergence.

For two time series Seq ₁ = (x ₁ , x ₂ , …, x _n ) and Seq ₂ = (y ₁ , y ₂ , …, y _m ), the cross-correlation measure is a statistical metric. Even if the two time series Seq ₁ and Seq ₂ are not aligned, the cross-correlation measure can be used to determine the similarity between the two time series Seq ₁ and Seq ₂ .

In order to achieve translation invariance, Seq ₂ is kept unchanged when calculating the cross-correlation, and Seq ₁ is slid on Seq ₂ to calculate the inner product of each displacement of Seq ₁ , which is recorded as CCw(Seq ₁ ,Seq ₂ )=R _wm (Seq ₁ ,Seq ₂ ),w∈1,2,……,2m-1. The calculation method of R _wm (Seq ₁ ,Seq ₂ ) can be expressed by the following formula.

The goal is to calculate w that maximizes CCw(Seq ₁ , Seq ₂ ), that is, to obtain the best move of Seq ₁ relative to Seq ₂ .

In some embodiments, the calculation method of the SBD distance is represented by the following formula.

In some embodiments, the clustering process may further include: E4, cluster visualization and calculation of mutual correlation.

It can be understood that the clustering process is visualized, which makes it easier for network managers to observe the clustering situation, count the number and type of cells in each cluster during the clustering process, and calculate the cross-correlation of the clustering results to measure the accuracy of the k-shape clustering algorithm on the data set.

It should be noted that, compared with the k-means clustering algorithm, the k-shape clustering algorithm is suitable for load scenarios where the load change trend is learned and the influence of the absolute value of the load can be ignored.

In some embodiments, if the preset distance algorithm used to determine the distance between any two cells in the plurality of second cells in S2022 is the Euclidean distance algorithm, then the k-means clustering algorithm may be used for clustering the plurality of second cells. If the preset distance algorithm used to determine the distance between any two cells in S2022 is the DTW distance algorithm, then the k-shape clustering algorithm may be used for clustering the plurality of second cells.

It is understandable that the DTW distance algorithm has higher requirements for computing resources than the Euclidean distance algorithm. In scenarios where there are high-precision requirements for clustering results and computing devices with high computing power, the DTW distance algorithm can be used. If there are no high-precision requirements for clustering results and/or no computing devices with high computing power, the Euclidean distance algorithm can be used.

In this way, clustering of the plurality of second cells based on the number of cell clusters and the distance between any two cells in the plurality of second cells is completed to obtain at least one cell cluster. The target cell cluster is one of the at least one cell cluster.

In some embodiments, if the number of cells in a cell cluster is too small, for example, the cell cluster includes only two cells, it is not meaningful to establish a unified communication load prediction model for the cell cluster. The communication load prediction device can regard at least one cell included in the cell cluster as an outlier and mark it, so that the network management personnel can adjust the load control strategy of at least one cell included in the cell cluster in a targeted manner. At the same time, the communication load prediction device can establish a communication load prediction model corresponding to each cell based on the historical communication load data of each cell in the cell cluster, so as to formulate a load control strategy corresponding to each cell in the cell cluster.

In some embodiments, if the communication load prediction device identifies that a cell has too few connected users, the communication load prediction device may issue a prompt message to modify the networking or antenna coverage of the cell, so as to prompt the network administrator to take corresponding measures for the cell.

The above embodiment introduces the process of how to cluster multiple second cells to obtain at least one cell cluster. In some embodiments, a communication load prediction method provided by the embodiment of the present disclosure, after determining at least one cell cluster, if the first cell is a cell newly added to the communication system shown in Figure 1 above, it is necessary to determine the cell cluster to which the first cell belongs. As shown in Figure 6, the method also includes the following steps.

S301. Obtain historical communication load data of the first cell.

In some embodiments, when the first cell joins the communication system shown in Figure 1, the communication load prediction device can obtain the communication load data of the first cell at multiple historical moments through the base station to which the first cell belongs, and then perform the preprocessing described in S201 on the communication load data of the first cell at multiple historical moments to obtain the historical communication load data of the first cell.

In some embodiments, when the first cell joins the communication system shown in Figure 1, the base station to which the first cell belongs actively reports the communication load data of the first cell at multiple historical moments to the communication load prediction device, and the communication load prediction model performs the preprocessing described in S201 on the communication load data of the first cell at multiple historical moments to obtain the historical communication load data of the first cell.

Exemplarily, the plurality of historical moments is 28 days.

S302: Determine the distance between the first cell and the cluster center of each cell cluster based on the historical communication load data of the first cell.

In some embodiments, after acquiring the historical communication load data of the first cell, the communication load prediction device can determine the distance between the first cell and the cluster center of each cell cluster based on the historical communication load data of the first cell and a preset distance algorithm, that is, determine the similarity between the first cell and the cluster center of each cell.

For the description of how the communication load prediction device determines the distance between the first cell and the cluster center of each cell cluster based on the historical communication load data of the first cell and a preset distance algorithm, please refer to the description of how to determine the distance between two second cells in S2022 above, which will not be repeated here.

S303: Take the cell cluster whose cluster center is closest to the first cell as the cell cluster to which the first cell belongs.

It can be understood that if the first cell is closest to the cluster center of a cell cluster, it means that the first cell has the highest similarity with the cluster center of the cell cluster. Therefore, the cell cluster whose cluster center is closest to the first cell can be used as the cell cluster to which the first cell belongs.

The above embodiments introduce the use process of the communication load prediction model corresponding to the target cell cluster, the clustering process of multiple second cells, and the determination process of the cell cluster to which a cell belongs. In some embodiments, a communication load prediction method provided by an embodiment of the present disclosure also involves the establishment process of the communication load prediction model corresponding to the target cell cluster, as shown in FIG7 , the method also includes the following steps.

S401. Determine a cell that meets a preset condition from the target cell cluster based on the cluster center of the target cell cluster.

In some embodiments, in order to improve the rate at which the communication load prediction model corresponding to the target cell cluster is established, the communication load prediction device can determine the cells that meet the preset conditions from the target cell cluster based on the cluster center of the target cell cluster. Then, based on the historical communication load data of the cells that meet the preset conditions in the target cell cluster, a communication load prediction model corresponding to the target cell cluster is established, so that the amount of data in the data set of the communication load prediction model can be reduced, and the rate at which the communication load prediction model is established can be improved.

In some embodiments, as shown in FIG. 8 , S401 may be implemented as the following steps, for example.

S4011. Determine the distance between each cell in the target cell cluster and the cluster center.

For the description of how to determine the distance between each cell in the target cell cluster and the cluster center in S4011, reference may be made to the description of how to determine the distance between two cells in S2022, which will not be described in detail here.

S4012: All cells in the target cell cluster whose distances to the cluster center are less than or equal to a preset threshold are regarded as cells meeting a preset condition.

The preset threshold may be preset by a network administrator.

It can be understood that the closer a cell is to the cluster center, that is, the higher the similarity between the cell and the cluster center, the more representative the change trend of the cell's communication load is for at least one cell in the target cell cluster corresponding to the cluster center. In other words, the change trend of the cell's communication load can represent the change trend of the communication load of other cells in the target cell cluster. Therefore, all cells in the target cell cluster whose distance to the cluster center is less than or equal to the preset threshold can be regarded as cells that meet the preset conditions.

In some embodiments, as shown in FIG. 9 , S401 may also be implemented as the following steps, for example.

S4013: Determine the distance between each cell in the target cell cluster and the cluster center.

Similarly, for the description of how to determine the distance between each cell in the target cell cluster and the cluster center in S4013, reference may be made to the description of how to determine the distance between two cells in S2022, which will not be repeated here.

S4014: Take the first N cells in the target cell cluster that are closest to the cluster center as cells that meet the preset conditions.

N is a positive integer, for example, N is 2.

It can be understood that the closer a cell is to the cluster center, that is, the higher the similarity between the cell and the cluster center, the more representative the change trend of the communication load of the cell is for at least one cell in the target cell cluster corresponding to the cluster center, that is, the change trend of the communication load of the cell can reflect the change trend of the communication load of other cells in the target cell cluster to a certain extent. The change trend of the communication load of the first N cells closest to the cluster center in the target cell cluster can represent the change trend of the communication load of at least one cell in the target cell cluster. Therefore, the first N cells closest to the cluster center in the target cell cluster can be regarded as cells that meet the preset conditions.

S402: Establish a communication load prediction model corresponding to the target cell cluster based on historical communication load data of cells that meet preset conditions.

In some embodiments, as shown in FIG. 10 , S402 may be implemented as the following steps, for example.

S4021. Generate multiple training samples based on historical communication load data of cells that meet preset conditions.

In some embodiments, after determining the cells that meet the preset conditions, the historical communication load data of the cells that meet the preset conditions can be modeled and processed, including but not limited to manual adjustment strategies, load prediction methods based on window filtering, and load prediction methods based on machine learning, to generate multiple training samples, so as to optimize the historical communication load data of the cells that meet the preset conditions, strengthen the characteristics of the historical communication load data of the cells that meet the preset conditions, and improve the generalization performance of the communication load prediction model completed by subsequent training.

S4022: Based on multiple training samples, train the initial model to obtain a communication load prediction model corresponding to the trained target cell cluster.

Based on multiple training samples, the initial model is trained until the error of the initial model converges, and a communication load prediction model corresponding to the trained target cell cluster is obtained.

In some embodiments, the communication load prediction model is built based on long short-term memory networks (LSTM).

In some embodiments, after the communication load prediction model corresponding to the target cell cluster is trained, the communication load prediction device may send the communication load prediction model to each cell of at least one cell included in the target cell cluster.

After receiving the communication load prediction model sent by the communication load prediction device, a cell can combine its own real-time communication load data to numerically restore the parameters of the communication load prediction model, and then use the numerically restored communication load prediction model as the communication load prediction model of the cell and store it for use in subsequent formulation of load control strategies for the cell.

Based on the embodiment shown in FIG8 , at least the following beneficial effects are brought about: It can be understood that each cell in a cell cluster is obtained through clustering, which means that each cell in a cell cluster has a high similarity in the communication load change trend. The communication load prediction model corresponding to a cell cluster is constructed based on the historical communication load data of the cells in the cell cluster that meet the preset conditions, and the change trend of the communication load of the cells that meet the preset conditions is representative for each cell in the cell cluster, so the historical communication load data of the cells that meet the preset conditions can be used as a representative of the historical communication load data of each cell in the cell cluster to construct the communication load prediction model corresponding to the cell cluster. In addition, each cell in the cell cluster can share the communication load prediction model corresponding to the cell cluster to predict the communication load, without having to establish a communication load prediction model corresponding to each cell for each cell, thereby reducing the consumption of computing resources.

The above mainly introduces the solution provided by the embodiment of the present disclosure from the perspective of the method. In order to achieve the above functions, it includes hardware structures and/or software modules corresponding to the execution of each function. Those skilled in the art should easily realize that, in combination with the units and algorithm steps of each example described in the embodiment disclosed in this article, the present disclosure can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present disclosure.

FIG11 is a schematic diagram of the composition of a communication load prediction device according to some embodiments. The communication load prediction device is used to execute the communication load prediction method described above. The communication load prediction device 2000 includes a communication unit 2001 and a processing unit 2002. In some embodiments, the communication load prediction device 2000 may also include a storage unit 2003.

In some embodiments, the communication unit 2001 is used to obtain real-time communication load data of the first cell.

Processing unit 2002 is used to obtain a communication load prediction result based on the real-time communication load data of the first cell and the communication load prediction model corresponding to the target cell cluster to which the first cell belongs. The communication load prediction model corresponding to the target cell cluster is constructed based on the historical communication load data of at least one cell in the target cell cluster.

In some embodiments, the processing unit 2002 is further used to determine cells meeting preset conditions from the target cell cluster based on the cluster center of the target cell cluster; and to establish a communication load prediction model corresponding to the target cell cluster based on historical communication load data of the cells meeting the preset conditions.

In some embodiments, the processing unit 2002 is used to: determine the distance between each cell in the target cell cluster and the cluster center; and take all cells in the target cell cluster whose distance to the cluster center is less than or equal to a preset threshold as cells meeting preset conditions.

In some embodiments, the processing unit 2002 is used to: determine the distance between each cell in the target cell cluster and the cluster center; and take the first N cells in the target cell cluster that are closest to the cluster center as cells that meet preset conditions, where N is a positive integer.

In some embodiments, the processing unit 2002 is used to: generate multiple training samples based on historical communication load data of cells that meet preset conditions; train the initial model based on the multiple training samples to obtain a communication load prediction model corresponding to the trained target cell cluster.

In some embodiments, the communication load prediction model is constructed based on a long-short time neural network.

In some embodiments, the communication unit 2001 is further used to obtain historical communication load data of the first cell.

The processing unit 2002 is further configured to: determine the distance between the first cell and the cluster center of each cell cluster based on the historical communication load data of the first cell; and take the cell cluster whose cluster center is closest to the first cell as the cell cluster to which the first cell belongs.

In some embodiments, the communication unit 2001 is further used to obtain historical communication load data of multiple second cells, and the first cell is one of the multiple second cells.

The processing unit 2002 is further configured to: perform clustering processing on the multiple second cells based on historical communication load data of the multiple second cells to obtain at least one cell cluster, wherein the target cell cluster is one of the at least one cell cluster.

In some embodiments, the processing unit 2002 is used to: determine the number of cell clusters based on historical communication load data of multiple second cells; for any two cells among the multiple second cells, determine the distance between the two second cells based on the historical communication load data of the two cells, and the distance between the two second cells is used to characterize the similarity between the two second cells in the communication load change trend; based on the number of cell clusters and the distance between any two second cells among the multiple second cells, cluster the multiple second cells to obtain at least one cell cluster.

In some embodiments, the above-mentioned real-time communication load data includes one or more of the following data types: the number of new radio interface NR carrier radio resource control RRC connections, the NR carrier uplink physical resource block PRB utilization rate, the NR carrier downlink PRB utilization rate, the cell group uplink PRB utilization rate, the cell group downlink PRB utilization rate, the Long Term Evolution LTE dynamic spectrum sharing DSS cell group uplink PRB usage number, the LTE DSS cell group downlink PRB usage number, the LTE DSS cell group RRC connection number, the LTE cell uplink PRB usage number, the LTE cell downlink PRB usage number and the LTE cell RRC connection number.

In some embodiments, the storage unit 2003 is used to store real-time communication load data of the first cell.

In some embodiments, the storage unit 2003 is also used to store the communication load prediction model corresponding to each cell cluster.

In some embodiments, the storage unit 2003 is also used to store historical communication load data of each cell.

The units in Fig. 11 may also be referred to as modules. For example, a processing unit may be referred to as a processing module.

If the various units in FIG. 11 are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present disclosure is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor (processor) to perform all or part of the steps of the method described in each embodiment of the present disclosure. The storage medium for storing computer software products includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc. Various media that can store program codes.

In the case of implementing the functions of the above-mentioned integrated modules in the form of hardware, the embodiment of the present disclosure provides a structural schematic diagram of an electronic device, which may be the above-mentioned communication load prediction device. As shown in FIG. 12 , the electronic device 3000 The electronic device includes: a processor 3002 , a communication interface 3003 and a bus 3004 . In some embodiments, the electronic device may further include a memory 3001 .

The processor 3002 may be a device that implements or executes various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of the present disclosure. The processor 3002 may be a central processing unit, a general-purpose processor, a digital signal processor, an application-specific integrated circuit, a field programmable gate array, or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. The processor 3002 may be a device that implements or executes various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of the present disclosure. The processor 3002 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, and the like.

The communication interface 3003 is used to connect with other devices through a communication network. The communication network can be Ethernet, wireless access network, wireless local area network (WLAN), etc.

The memory 3001 may be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store desired program codes in the form of instructions or data structures and can be accessed by a computer, but is not limited to these.

As an example, the memory 3001 may exist independently of the processor 3002, and the memory 3001 may also be connected to the processor 3002 via the bus 3004 to store instructions or program codes. When the processor 3002 calls and executes the instructions or program codes stored in the memory 3001, the communication load prediction method provided in the embodiment of the present disclosure can be implemented.

As another example, the memory 3001 may also be integrated with the processor 3002 .

The bus 3004 may be an extended industry standard architecture (EISA) bus, etc. The bus 3004 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, FIG12 only uses one thick line, but does not mean that there is only one bus or one type of bus.

Through the description of the above implementation mode, the technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the communication load prediction device can be divided into different functional modules to complete all or part of the functions described above.

The embodiment of the present disclosure also provides a computer-readable storage medium. All or part of the processes in the above method embodiments can be completed by computer instructions to instruct the relevant hardware, and the program can be stored in the above computer-readable storage medium. When the program is executed, it can include the processes of the above method embodiments. The computer-readable storage medium can be the memory or memory of any of the above embodiments. The above computer-readable storage medium can also be an external storage device of the above communication load prediction device, such as a plug-in hard disk, a smart memory card (smart media card, SMC), a secure digital (secure digital, SD) card, a flash card (flash card), etc. equipped on the above communication load prediction device. Further, the above computer-readable storage medium can also include both the internal storage unit of the above communication load prediction device and an external storage device. The above computer-readable storage medium is used to store the above computer program and other programs and data required by the above communication load prediction device. The above computer-readable storage medium can also be used to temporarily store data that has been output or is to be output. The readable storage medium includes a non-transient computer-readable storage medium.

The present disclosure also provides a computer program product. The computer product includes a computer program. When the program product runs on a computer, the computer executes any one of the communication load prediction methods provided in the above embodiments.

Although the present disclosure is described herein in conjunction with various embodiments, in the process of implementing the claimed disclosure, those skilled in the art may understand and implement other variations of the disclosed embodiments by viewing the drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other components or steps, and "one" or "an" does not exclude multiple situations. A single processor or other unit may implement several functions listed in a claim. Certain measures are recorded in different dependent claims, but this does not mean that these measures cannot be combined to produce good results.

Although the present disclosure has been described in conjunction with features and embodiments thereof, it is apparent that various modifications and combinations may be made thereto without departing from the spirit and scope of the present disclosure. Accordingly, this specification and the drawings are merely exemplary illustrations of the present disclosure as defined by the appended claims, and are deemed to have covered any and all modifications, variations, combinations or equivalents within the scope of the present disclosure. Obviously, those skilled in the art may make various modifications and variations to the present disclosure without departing from the spirit and scope of the present disclosure. Thus, if these modifications and variations of the present disclosure fall within the scope of the claims of the present disclosure and their equivalents, the present disclosure is also intended to include these modifications and variations.

The above are only specific embodiments of the present disclosure, but the protection scope of the present disclosure is not limited thereto. Any changes or substitutions within the technical scope disclosed in the present disclosure should be included in the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be based on the protection scope of the claims.

Claims (13)

1. A communication load prediction method, comprising:

   Acquire real-time communication load data of the first cell;

   A communication load prediction result is obtained based on the real-time communication load data of the first cell and the communication load prediction model corresponding to the target cell cluster to which the first cell belongs. The communication load prediction model corresponding to the target cell cluster is constructed based on the historical communication load data of at least one cell in the target cell cluster.
2. The method according to claim 1, wherein the communication load prediction model corresponding to the target cell cluster is constructed by the following steps:

   Based on the cluster center of the target cell cluster, determining a cell that meets a preset condition from the target cell cluster;

   Based on the historical communication load data of the cells meeting the preset conditions, a communication load prediction model corresponding to the target cell cluster is established.
3. The method according to claim 2, wherein the determining, based on the cluster center of the target cell cluster, a cell that meets a preset condition from the target cell cluster comprises:

   Determining the distance between each cell in the target cell cluster and the cluster center;

   All cells in the target cell cluster whose distances from the cluster center are less than or equal to a preset threshold are taken as cells meeting the preset condition.
4. The method according to claim 2, wherein the determining, based on the cluster center of the target cell cluster, a cell that meets a preset condition from the target cell cluster comprises:

   Determining the distance between each cell in the target cell cluster and the cluster center;

   The first N cells in the target cell cluster that are closest to the cluster center are used as the cells that meet the preset conditions, where N is a positive integer.
5. The method according to any one of claims 2 to 4, wherein the establishing the communication load prediction model corresponding to the target cell cluster based on the historical communication load data of the cell that meets the preset conditions comprises:

   Generate multiple training samples using the historical communication load data of the cells meeting the preset conditions;

   Based on the multiple training samples, the initial model is trained to obtain a communication load prediction model corresponding to the target cell cluster that has been trained.
6. The method according to claim 5, wherein the communication load prediction model is constructed based on a long-short time neural network.
7. The method according to any one of claims 1 to 4, further comprising:

   Acquire historical communication load data of the first cell;

   Determining, based on historical communication load data of the first cell, distances between the first cell and cluster centers of each cell cluster;

   The cell cluster whose cluster center is closest to the first cell is used as the cell cluster to which the first cell belongs.
8. The method according to any one of claims 1 to 4, further comprising:

   Acquire historical communication load data of a plurality of second cells, the first cell being one of the plurality of second cells;

   Based on the historical communication load data of the plurality of second cells, clustering processing is performed on the plurality of second cells to obtain at least one cell cluster, and the target cell cluster is one of the at least one cell cluster.
9. The method according to claim 8, wherein the clustering of the plurality of second cells based on the historical communication load data of the plurality of second cells to obtain at least one cell cluster comprises:

   determining the number of cell clusters based on the historical communication load data of the plurality of second cells;

   For any two cells among the plurality of second cells, based on the historical communication load data of the two cells, determine determining a distance between the two second cells, where the distance between the two second cells is used to characterize a similarity between the two second cells in a communication load change trend;

   Based on the number of cell clusters and the distance between any two second cells in the multiple second cells, clustering processing is performed on the multiple second cells to obtain the at least one cell cluster.
10. The method according to claim 1, wherein the real-time communication load data includes one or more of the following data types:

    Number of new air interface NR carrier radio resource control RRC connections, NR carrier uplink physical resource block PRB utilization rate, NR carrier downlink PRB utilization rate, cell group uplink PRB utilization rate, cell group downlink PRB utilization rate, Long Term Evolution LTE dynamic spectrum sharing DSS cell group uplink PRB usage number, LTE DSS cell group downlink PRB usage number, LTE DSS cell group RRC connections number, LTE cell uplink PRB usage number, LTE cell downlink PRB usage number and LTE cell RRC connections number.
11. A communication load prediction device, comprising:

    A communication unit, configured to obtain real-time communication load data of the first cell;

    A processing unit is used to obtain a communication load prediction result based on the real-time communication load data of the first cell and a communication load prediction model corresponding to a target cell cluster to which the first cell belongs, wherein the communication load prediction model corresponding to the target cell cluster is constructed based on historical communication load data of at least one cell in the target cell cluster.
12. An electronic device comprises: a processor and a memory for storing instructions executable by the processor;

    The processor is configured to execute the instructions so that the electronic device performs the communication load prediction method as described in any one of claims 1-10.
13. A computer-readable storage medium, wherein computer instructions are stored on the computer-readable storage medium, and when the computer instructions are executed on an electronic device, the electronic device executes the communication load prediction method as described in any one of claims 1-10.