Dynamic load forecasting method, device, base station and storage medium

Abstract

The present disclosure relates to a dynamic load forecasting method, device, base station, and storage medium, which includes the steps of: determining a cross-layer unidirectional group of a target base station cell in a plurality of base station cells located on a predetermined load transfer route; obtaining a real-time load of the target base station cell at a current time and a historical predicted load corresponding to a previous time; receiving cross-layer neighboring cell load data transmitted by each of the left adjacent cell unidirectional group and the right adjacent cell unidirectional group of the target base station cell cross-layer unidirectional group, wherein the corresponding cross-layer neighboring cell load data both include historical load data and real-time load data; and performing load forecasting based on the real-time load, historical predicted load, and cross-layer neighboring cell load data using a preset load forecasting model, obtaining a forward predicted load and a backward predicted load corresponding to the target base station cell, and determining a corresponding load forecast result based on the forward predicted load and the backward predicted load.

Description

[Technical Field]

[0001] This disclosure claims priority to Chinese patent application CN202211028169.9, entitled "Dynamic Load Forecasting Method, Apparatus, Base Station, and Storage Medium," filed on August 25, 2022, the entire contents of which are incorporated by reference.

[0002] The present disclosure relates to the field of base station systems, and in particular to a dynamic load prediction method, apparatus, base station, and storage medium. [Background technology]

[0003] In related technologies, the term "mobile network load" typically refers to the number of users (terminals) accessing a network base station cell or the occupancy rate of various resources within the network base station cell. Network load is a key data indicator in the operation and maintenance of mobile communication networks, and is a key indicator for real-time network monitoring. It is directly related to stable network operation, user perception, and energy conservation, making it a research hotspot and application focus for mobile communication networks. In related technologies, load forecasting can support and lead network base stations to implement various energy-saving measures, timely wake-up, proactive load balancing, and high-traffic admission control, thereby ensuring stable base station operation and user perception of services.

[0004] In related technologies, the Auto Regressive Integrated Moving Average (ARIMA) model is commonly used for load forecasting. It uses historical load data from network base station cells to train and regress the model, predicting future loads for each base station cell. Other load forecasting methods employ deep neural network models, such as recurrent neural network (RNN) models and improved long short-term memory (LSTM) models. However, conventional forecasting methods generally require sufficient time and historical data for effective training and accurate predictions. Furthermore, load forecasting requires that base station cell load changes exhibit periodic regularity. In mobile communication scenarios where there is no clear time regularity, the network topology and user movement trajectories are complex, or the network topology structure is relatively simple but users move quickly, and the load is sudden and unpredictable, related load forecasting methods cannot achieve accurate and effective load forecasts.

[0005] In the related art, there is currently no effective solution to the problem that load prediction methods cannot be applied to high-speed mobile communication scenarios, such as users moving at high speed, loads being sudden, and lacking precise time period regularity. Summary of the Invention [Problem to be solved by the invention]

[0006] The present disclosure provides a dynamic load forecasting method, apparatus, base station, and storage medium. [Means for solving the problem]

[0007] In a first aspect, the present disclosure provides a dynamic load prediction method applied to a target base station cell under a high-speed private network, comprising the steps of: determining a cross-layer unidirectional group in a plurality of base station cells located on a predetermined load migration route, wherein base station cells located at both ends of the cross-layer unidirectional group are both K-th layer primary neighbor cells corresponding to the target base station cell, K is a predetermined value; obtaining a real-time load at a current time and a historical predicted load corresponding to a previous time, wherein the historical predicted load is used to characterize a load migration change that occurs at the current time predicted at the previous time; and receiving cross-layer neighbor cell load data transmitted by each of a left neighbor cell unidirectional group and a right neighbor cell unidirectional group of the cross-layer unidirectional group, wherein the corresponding The corresponding cross-layer neighboring cell load data all include historical load data and real-time load data, the historical load data being used to characterize the corresponding predicted load of the Kth layer primary neighboring cell of the corresponding neighboring cell unidirectional group at a previous time, and the real-time load data being used to characterize the actual load of the Kth layer primary neighboring cell of the corresponding neighboring cell unidirectional group at a current time; and a dynamic load forecasting method comprising the steps of: using a preset load forecasting model to perform load forecasting based on the real-time load, the historical predicted load, and the cross-layer neighboring cell load data, obtaining a forward predicted load and a backward predicted load corresponding to the target base station cell, and determining a corresponding load forecast result based on the forward predicted load and the backward predicted load.

[0008] In a second aspect, the present disclosure provides a dynamic load prediction device applied to a target base station cell under a high-speed private network, comprising: a determination module, an acquisition module, a reception module, and a prediction module, wherein the determination module is configured to determine a cross-layer unidirectional group among a plurality of base station cells located on a predetermined load migration route, and the base station cells located at both ends of the cross-layer unidirectional group are both K-th layer primary neighbor cells corresponding to the target base station cell, where K is a default value, the acquisition module is configured to acquire a real-time load at a current time and a historical predicted load corresponding to a previous time, and the historical predicted load is used to characterize a load migration change that occurs at the current time predicted at the previous time, and the reception module is configured to receive a load signal transmitted by each of a left adjacent cell unidirectional group and a right adjacent cell unidirectional group of the cross-layer unidirectional group. The present invention provides a dynamic load prediction device, the dynamic load prediction device being configured to receive cross-layer neighboring cell load data, wherein the corresponding cross-layer neighboring cell load data both include historical load data and real-time load data, the historical load data being used to characterize the corresponding predicted load of a primary neighboring cell of the Kth layer in the corresponding neighboring cell unidirectional group at a previous time, and the real-time load data being used to characterize the actual load of the primary neighboring cell of the Kth layer in the corresponding neighboring cell unidirectional group at a current time, the prediction module being configured to use a preset load prediction model to perform load prediction based on the real-time load, the historical predicted load, and the cross-layer neighboring cell load data, obtain a forward predicted load and a backward predicted load corresponding to the target base station cell, and determine a corresponding load prediction result based on the forward predicted load and the backward predicted load.

[0009] In a third aspect, there is provided a base station comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus, the memory stores a computer program, and the processor performs the dynamic load prediction method according to the first aspect when the program stored in the memory is executed.

[0010] In a fourth aspect, there is provided a computer readable medium having stored thereon a computer program that, when executed by a processor, performs the dynamic load prediction method according to the first aspect.

[0011] To more clearly illustrate and describe other features, objects, and advantages of the present disclosure, the details of one or more embodiments of the disclosure are set forth in the drawings and description that follow. [Brief explanation of the drawings]

[0012] The drawings herein are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure, and together with the specification, serve to explain the principles of the disclosure. In order to more clearly describe the technical solutions in the embodiments of the present disclosure or related art, the drawings that need to be used in the description of the embodiments or related art are briefly described below. It is obvious to those skilled in the art that other drawings can be obtained based on these drawings without exerting any creative effort.

[0013] [Figure 1] FIG. 1 is a conceptual flow diagram of a dynamic load forecasting method provided by an embodiment of the present disclosure.

[0014] [Figure 2] FIG. 2 is a conceptual diagram illustrating the configuration of a cross-layer unidirectional group according to an embodiment of the present disclosure.

[0015] [Figure 3]FIG. 3 is a conceptual diagram of a dynamic load prediction device provided by an embodiment of the present disclosure.

[0016] [Figure 4] FIG. 4 is a conceptual diagram of a base station according to an embodiment of the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0017] In order to make the objectives, technical solutions and advantages of the embodiments of the present disclosure clearer, the technical solutions in the embodiments of the present disclosure will be clearly and thoroughly described below in combination with the drawings of the embodiments of the present disclosure. Obviously, the described embodiments are only some of the embodiments of the present disclosure, but not all of the embodiments. Based on the embodiments of the present disclosure, other embodiments obtained by those skilled in the art without exerting their creative powers are all within the scope of protection of the present disclosure.

[0018] Before describing the embodiments of the present disclosure, the names of related technologies related to the present disclosure will be described below.

[0019] A base station (BS) belongs to the basic unit of a mobile network.

[0020] A cell is the basic unit of a base station. In the cloverleaf structure of a conventional cellular base station, one base station has three cells covering a 120° sector, while in a high-speed rail private network, one base station has two or more cells.

[0021] User equipment (UE).

[0022] 5G: the Fifth-Generation mobile communication.

[0023] 4G: the Fourth-Generation mobile communication.

[0024] New Radio (NR:NEW Ratio).

[0025] Long Term Evolution (LTE).

[0026] Autoregressive Integrated Moving Average (ARIMA) is an improved time series model based on ARMA.

[0027] Recurrent Neural Network (RNN).

[0028] Long Short-term Memory (LSTM) is an improved neural network model based on RNN.

[0029] Hereinafter, the technical solutions in the embodiments of the present disclosure will be described in combination with the drawings in the embodiments of the present disclosure.

[0030] 1 is a conceptual flow diagram of a dynamic load forecasting method provided by an embodiment of the present disclosure. As shown in FIG. 1, the embodiment of the present disclosure provides a dynamic load forecasting method applied to a target base station cell under a high-speed private network, and the method includes the following steps S101 to S104.

[0031] Step S101: For a plurality of base station cells located on a predetermined load transfer route, a cross-layer unidirectional group is determined, and the base station cells located at both ends of the cross-layer unidirectional group are all K-th layer primary neighbor cells corresponding to the target base station cell, where K is a default value.

[0032] In an embodiment of the present disclosure, the dynamic load forecasting method forecasts the load to be generated at the next time based on the associated load at the current time and the associated load at the previous time of the corresponding base station cell, and obtains the changes, load change trend, and load change amount that will occur in the corresponding load of the base station cell in the future through the load forecast. In this embodiment, the load forecasting does not primarily aim to make the predicted load closer to the actual load, i.e., does not aim at the accuracy of the forecast, but rather takes into consideration the predicted load and the actual load comprehensively to obtain the load change trend and change amount in the corresponding load of the base station cell in the future, thereby providing effective reference and guidance for operations such as energy saving, load control, and call guarantee of the base station cell. In this embodiment, the load forecast is performed on the target base station cell, that is, one of multiple base station cells located in the load transition direction.

[0033] In this embodiment, the devised high-speed private network is a high-speed private network corresponding to a linear high-speed railroad line, so that each base station cell of a plurality of base station cells under the high-speed private network only needs to have one corresponding cross-layer co-directional group. For a base station cell located at the central intersection of a T-shaped or cross-shaped high-speed railroad line, there should be two corresponding cross-layer co-directional groups; for a base station cell located at the central intersection of a X-shaped high-speed railroad line, there should be three cross-layer co-directional groups; and for a base station cell located at the central intersection of a U-shaped high-speed railroad line, there should be four cross-layer co-directional groups.

[0034] In this embodiment, before determining the cross-layer co-directional group corresponding to the target base station cell, a plurality of base station cells C installed on a section of the high-speed private network corresponding to the high-speed railway line are iFor each base station cell in (i=1, 2, ..., N), determine the primary neighboring cell, that is, for the left and right neighboring cells corresponding to the base station cell, screen and determine the primary neighboring cell of each base station cell based on the "neighbor cell pair" switching count statistical data, for example, base station cell C i is the time-granularity statistical data of the number of "neighbor cell pair" switchings for a certain period of time (for example, 168 hours corresponding to one cycle) and each of its neighboring cells (the number of times switching to a neighboring cell and the number of times the neighboring cell is a base station cell C) i Based on the number of times the neighboring cells have switched to the base station cell C, the neighboring cells are divided into two types according to a clustering algorithm. The neighboring cells with a higher number of switching times are designated as the base station cell C. i Of course, in some alternative embodiments, the network planner or optimizer may decide to use the base station cell C i and the base station cell C based on the network engineering parameter information of the neighboring cells or by further referring to the switching statistical data. i Manually screen the nearest neighboring cells on both the left and right sides along the direction of the high-speed railway track, and select base station cell C i The base station cell C can be the primary neighboring cell of i The nearest neighboring cells on both the left and right sides are base station cell C. i Then, a cross-layer co-directional group corresponding to each base station cell is determined based on the relationship between the co-directional groups corresponding to multiple base station cells. For example, if C2 is the right adjacent cell of the C1 co-directional group and C3 is the right adjacent cell of the C2 co-directional group, C3 can be related to C1 as the right adjacent cell crossing the 1st layer, and similarly, C1 is the right adjacent cell crossing the Kth layer. k {C1,C2,...C K} constitutes a cross-layer same direction group in which C1 crosses the K layer.

[0035] In the embodiments of the present disclosure, the cross layer co-directional group corresponding to the determined target base station cell includes two cross layer co-directional groups, for example, a left adjacent cell co-directional group and a right adjacent cell co-directional group, where the left adjacent cell co-directional group and the right adjacent cell co-directional group each include a K-th layer primary neighbor cell, where K is a default value, for example, the target base station cell C i If it is desired to predict future loads earlier and detect loads over longer distances, K can be set to a larger value. However, this increases the overhead and difficulty of cross-layer transmission of load information, and also increases random uncertainty. Therefore, the K value must be set taking into account the balance between the target requirements for load forecasting and the overhead cost, for example, K = 9.

[0036] Step S102: Obtain a real-time load at the current time and a historical predicted load corresponding to the previous time, and the historical predicted load is used to characterize the load transition change that occurs at the current time and that was predicted at the previous time.

[0037] In an embodiment of the present disclosure, when predicting the load for the next time, two dimensions, the current time and the previous time, must be considered from the time dimension. Here, there is a set time period T between the previous time, the current time, and the next time. The time period T is determined by the typical average base station cell coverage distance in a high-speed rail private network and the average speed of high-speed rail trains. Its possible value range may be 10 to 60 seconds, with T=30 seconds being an available option. From the load information dimension, two dimensions, the real-time load and the predicted load, must be considered. The predicted load refers to the load that will occur at the current time that has already been predicted at the previous time. From the base station cell target dimension, the base station cells participating in the load prediction include two dimensions: a target base station cell and one cross-layer unidirectional group.

[0038] In this embodiment, the load corresponding to the target base station cell includes the real-time load at the current time and the predicted load predicted corresponding to the previous time. Since the predicted load is a prediction of the load change that may occur in the base station cell in the future, the difference between the real-time load at the current time and the predicted load is determined based on the corresponding predicted load at the previous time, thereby characterizing whether the load change trend of the target base station cell itself matches the predicted change trend.

[0039] In addition, when considering the time dimension, the time period T needs to be set taking into account the time situation, for example, if T is too small, the overhead of load prediction and load information transmission will be large, and if T is too large, there is a high possibility that a high-speed train will pass through the target base station cell and some of the adjacent cells related to the target base station cell, so that the corresponding base station cell cannot sample the true actual load, and the target base station cell will not be able to achieve effective load prediction. When considering the real-time load, the real-time load corresponding to the current time of each base station cell needs to consider the real-time load that is located in the base station cell and corresponds to the base station cell in the forward load transition direction and is transitioning in the reverse direction to the base station, so the real-time load of the base station cell is the load that was transitioned from the corresponding base station cell at the previous time minus the detected real-time load, and only in this way can the real-time load at the current time be matched to the load transition situation.

[0040] Step S103: Receive cross-layer neighboring cell load data transmitted by each of the left neighboring cell neighboring group and the right neighboring cell neighboring group of the cross-layer unidirectional group, and the corresponding cross-layer neighboring cell load data both include historical load data and real-time load data, where the historical load data is used to characterize the corresponding predicted load of the Kth layer primary neighboring cell of the corresponding neighboring cell unidirectional group at the previous time, and the real-time load data is used to characterize the actual load of the Kth layer primary neighboring cell of the corresponding neighboring cell unidirectional group at the current time.

[0041] In this embodiment, the target base station cell C i If the target base station cell C is located at the beginning or end of the high-speed private network, i There is only one cross-layer co-directional group corresponding to the left adjacent cell co-directional group and the right adjacent cell co-directional group. At this time, the target base station cell C i performs only forward load prediction or reverse load prediction.

[0042] Step S104: Use a preset load prediction model to perform load prediction based on the real-time load, historical prediction load, and cross-layer adjacent cell load data, obtain the forward direction prediction load and reverse direction prediction load corresponding to the target base station cell, and determine the corresponding load prediction result based on the forward direction prediction load and reverse direction prediction load.

[0043] In the embodiment of the present disclosure, the preset load forecasting model sets a corresponding calculation function, and then performs forward load forecasting and reverse load forecasting on the real-time load, historical predicted load, and corresponding cross-layer adjacent cell load data corresponding to the target base station cell based on the calculation function, respectively, to obtain the corresponding forward predicted load and reverse predicted load, calculate the total predicted load, and then determine the corresponding forecast result based on the total predicted load.

[0044] The above steps S101 to S104 result in "a step of determining a cross layer unidirectional group in a plurality of base station cells located on a predetermined load migration route, wherein the base station cells located at both ends of the cross layer unidirectional group are both K-th layer primary neighbor cells corresponding to the target base station cell, K being a predetermined value; a step of acquiring a real-time load at a current time and a historical predicted load corresponding to a previous time, wherein the historical predicted load is used to characterize a load migration change occurring at the current time predicted at the previous time; and a step of receiving cross layer neighbor cell load data transmitted by each of a left neighbor cell unidirectional group and a right neighbor cell unidirectional group of the cross layer unidirectional group, wherein the corresponding cross layer neighbor cell load data both include historical load data and real-time load data, and the historical load data is used to characterize the corresponding predicted load at the previous time of the K-th layer primary neighbor cell included in the corresponding neighbor cell unidirectional group." The present invention employs the steps of: using real-time load data to characterize the actual load at the current time of the primary neighbor cell of the Kth layer in the corresponding neighbor cell unidirectional group; and using a preset load prediction model to perform load prediction based on the real-time load, historical predicted load, and cross-layer neighbor cell load data, to obtain the forward direction predicted load and backward direction predicted load corresponding to the target base station cell, and determining the corresponding load prediction result based on the forward direction predicted load and backward direction predicted load. This solves the problem in the related art that load prediction methods cannot be applied to high-speed mobile communication scenarios in which users move at high speed, the load is sudden, and there is no precise time period regularity. The present invention enables the base station cell to accurately detect and sense the user load and change trend of users arriving at neighbor cells that are farther away in advance. The combination of the predicted load and actual load of the base station cell provides a beneficial effect of providing effective reference and guidance for operations such as energy saving, load control, and call guarantee of the base station cell.

[0045] In some embodiments, the steps of using a preset load prediction model to perform load prediction based on real-time load, historical predicted load, and cross-layer adjacent cell load data, and obtaining forward direction predicted load and backward direction predicted load corresponding to the target base station cell are realized by the following steps 21 to 22.

[0046] Step 21: Using the load forecasting model, perform forward load calculation on the real-time load, the historical forecast load, and the historical load data and real-time load data corresponding to the left adjacent cell same-direction group to obtain the forward forecast load.

[0047] In this embodiment, on the corresponding load transfer route, "from left to right" is defined as the forward direction, and "from right to left" is defined as the reverse direction. In addition, when calculating the forward load, in order to determine the accurate real-time load of the target base station cell, it is also necessary to consider the load transmitted to the target base station cell at the previous time by the first-layer primary neighbor cell in the right neighbor cell same-direction group. This makes the predicted load closer to the load change trend, that is, it can better reflect the change trend of the target base station cell.

[0048] Step 22: Using the load forecasting model, perform reverse load calculation on the real-time load, the historical forecast load, and the historical load data and real-time load data corresponding to the right adjacent cell same-direction group to obtain the reverse forecast load.

[0049] In this embodiment, when calculating the reverse load, in order to determine the accurate real-time load of the target base station cell, it is also necessary to consider the load transmitted to the target base station cell at the previous time by the first-layer primary neighbor cell in the left neighbor cell same-direction group, thereby making the predicted load closer to the load change trend, that is, better reflecting the change trend of the target base station cell.

[0050] In some embodiments, the following steps 31 to 34 are further performed to obtain an accurate real-time load of each base station cell.

[0051] Step 31: Obtain a first actual load at the current time of each base station cell located on a predetermined load transfer route.

[0052] In this embodiment, the first actual load includes the partial load transferred to the target base station cell at a previous time and the corresponding partial load for load prediction. In order to make the predicted load more consistent with the foreseeable load prediction and make the predicted load change trend more accurate, in this embodiment, the partial load transferred to the target base station cell at a previous time needs to be excluded before making the corresponding prediction.

[0053] Step 32: Determine the nearest left neighboring cell and nearest right neighboring cell corresponding to each base station cell, and respectively obtain the nearest right neighboring cell corresponding to the corresponding base station cell and the actual load corresponding to the previous time of the nearest left neighboring cell, and the positive direction is from the left end to the right end of the predetermined load transfer route, and the nearest left neighboring cell is used to characterize the first base station cell located on the left side of the corresponding base station cell in the positive direction, and the nearest right neighboring cell is used to characterize the first base station cell located on the right side of the corresponding base station cell in the positive direction.

[0054] In this embodiment, the nearest left neighboring cell is the first base station cell located to the left (rear) of the corresponding base station cell in the correspondingly set forward direction, and the nearest right neighboring cell is the first base station cell located to the right (i.e., front) of the base station cell in the correspondingly set forward direction, and the nearest left neighboring cell receives a load that moves in the reverse direction from the base station cell, and the nearest right neighboring cell receives a load that moves in the forward direction from the base station cell.

[0055] Step 33: Based on the actual load corresponding to the nearest right neighboring cell at the previous time, determine a first transition load of the nearest right neighboring cell to be transitioned to the base station cell at a first predetermined rate at the previous time, and based on the actual load corresponding to the nearest left neighboring cell at the previous time, determine a second transition load of the nearest left neighboring cell to be transitioned to the base station cell at a second predetermined rate at the previous time.

[0056] Step 34: Based on the difference between the first actual load and the first transitional load, determine the real-time load corresponding to when the base station cell performs forward prediction at the current time, and based on the difference between the first actual load and the second transitional load, determine the real-time load corresponding to when the base station cell performs backward prediction at the current time.

[0057] In this embodiment, the true load corresponding to the forward prediction or backward prediction of the base station cell at the current time is determined based on the difference value between the first actual load and the first transitional load and the difference value between the first actual load and the second transitional load.Therefore, the load at the current time determined in this manner can better reflect the changing trend of the load of the base station cell in the predicted load obtained by the load prediction calculation, and for example, the time of arrival of the load corresponding to the corresponding base station cell can be more accurately predicted.

[0058] Among the steps, the steps include a step of acquiring a first actual load at the current time of each base station cell located on a predetermined load transfer route, a step of determining a nearest left adjacent cell and a nearest right adjacent cell corresponding to each base station cell, and acquiring actual loads corresponding to a previous time of the nearest right adjacent cell and the nearest left adjacent cell corresponding to the corresponding base station cell, and a step of determining a first transfer load to transfer to the base station cell at a first predetermined rate at the previous time of the nearest right adjacent cell based on the actual load corresponding to the nearest right adjacent cell at the previous time, and By the steps of determining a second transition load to be transitioned to the base station cell at a second predetermined rate at the previous time of the adjacent cell, and determining a real-time load corresponding to when the base station cell performs forward prediction at the current time based on the difference between the first actual load and the first transition load, and determining a real-time load corresponding to when the base station cell performs backward prediction at the current time based on the difference between the first actual load and the second transition load, accurate real-time load of each base station cell can be obtained, and further providing effective reference and guidance for operations such as energy saving, load control, and call guarantee of the base station cell.

[0059] TIFF2025527012000002.tif32170

[0060] TIFF2025527012000003.tif216170

[0061] In this embodiment, the base station cell is a base station cell corresponding to a linear high-speed rail network, and the number of corresponding cross-layer unidirectional groups is P=1. In this embodiment, the default values of α, β, and γ are all 0.5.

[0062] TIFF2025527012000004.tif69170

[0063] The time interval between the current time and the previous time is the cycle T for transmitting load forecast and load information, which is determined by the average coverage distance of a typical base station cell in a high-speed rail private network and the average speed of a high-speed rail train. The possible value range of T is approximately 10 to 60 seconds, with T = 30 seconds being a selectable value. If T is too small, the overhead of load forecast and load information transmission will be large. If T is too large, the high-speed train will likely pass through the target base station cell and some of the primary neighbor cells related to the target base station cell. As a result, the corresponding base station cell will not be able to sample the true real-time load, and the target base station cell will not be able to achieve effective load forecasting.

[0064] Furthermore, when T is set to match the base station cell coverage distance and the speed of high-speed rail trains, the larger α and β are, i.e., the larger the load weight and actual load weight of the target base station cell itself, the higher the load prediction accuracy, but the shorter the load detection distance, which is unfavorable for sensing loads over longer distances and is unfavorable for predicting the arrival of loads earlier. The smaller α and β are, i.e., the smaller the load weight and actual load weight of the target base station cell itself, the longer the load detection distance, but the lower the load prediction accuracy. In the embodiments of the present disclosure, the objective of load prediction is not to make the predicted load and the actual load as close as possible, i.e., not to focus on improving prediction accuracy as the core objective, but to combine and consider the predicted load and the actual load comprehensively to provide effective reference and guidance for operations such as energy saving, load control, and call guarantee of base station cells.

[0065] In some embodiments, the step of determining the corresponding load prediction result based on the forward predicted load and the backward predicted load is realized by the following steps 41 to 42.

[0066] Step 41: The forward predicted load and the backward predicted load are summed to obtain a first predicted total load that the target base station corresponds to along a predetermined load transfer route.

[0067] In this embodiment, the first predicted total load is calculated using the following formula.

[0068] TIFF2025527012000005.tif26170

[0069] Step 42: Accumulate the first forecasted total loads corresponding to the plurality of predetermined load transfer routes to generate a total forecasted load, where the load forecast result includes the total forecasted load.

[0070] TIFF2025527012000006.tif13170

[0071] TIFF2025527012000007.tif18170

[0072] In this embodiment, in the case of a linear high-speed rail network, the total predicted load is equal to the first predicted total load, and in the case of a cross-shaped or U-shaped high-speed rail network, the total predicted load corresponding to the target base station cell located at the intersection of multiple load transfer routes is the sum of multiple first predicted total loads.

[0073] Among the above steps, the step of calculating the sum of the forward direction predicted load and the backward direction predicted load to obtain the first predicted total load corresponding to the target base station on a predetermined load transfer route, and the step of accumulating the first predicted total loads corresponding to multiple predetermined load transfer routes to generate a total predicted load, with the load prediction result including the total predicted load, realizes a summary calculation of the total load predicted to occur in the target base station cell at the next time at the current time.

[0074] In some embodiments, after determining the corresponding load prediction result, the following steps 51 to 53 are further performed.

[0075] Step 51: Obtain the total predicted load corresponding to the current time and the real-time load corresponding to the current time, and obtain the historical total predicted load corresponding to the previous time and the historical actual load corresponding to the previous time.

[0076] In this embodiment, the predicted load corresponding to the target base station cell at the current time is obtained by acquiring and summing up the forward direction predicted load and the reverse direction predicted load, and calculating the sum of the predicted total loads corresponding to multiple load transfer routes.

[0077] Step 52: Generate a variation function of the load forecast based on at least two of the actual load corresponding to the current time, the total forecasted load corresponding to the current time, the historical total forecasted load corresponding to the previous time, and the historical actual load corresponding to the previous time, and determine a calculated value corresponding to the corresponding variation function, where the variation function is used to characterize the predictive performance of the load forecast.

[0078] In this embodiment, the expression method of different load forecasting sensitivity parameters is determined by a preset calculation function that optimizes the parameters of the load forecasting model. For example, the difference value between the real-time load corresponding to the current time and the total forecasted load corresponding to the current time may be adopted as the variation function.

[0079] In this embodiment, "generating a variation function of load prediction" refers to defining a preset variation function using at least two of the actual load corresponding to the current time, the total predicted load corresponding to the current time, the historical total predicted load corresponding to the previous time, and the historical actual load corresponding to the previous time as corresponding parameters. "Determining a calculated value corresponding to the variation function" refers to using the function value obtained by applying the correspondingly defined variation function to one of the four parameters.

[0080] Step 53: Determine whether the calculated value corresponding to the variation function is greater than a predetermined threshold, and adjust the weighting coefficient corresponding to the load prediction model according to the determination result, and the weighting coefficient corresponding to the load prediction model is adjusted to the target base station cell C i The weights include the weight α of the load of the base station itself, the weight β of the actual load of each base station cell, and the first predetermined ratio γ.

[0081] In this embodiment, the load prediction model determines whether it is sensitive to load prediction by determining whether the calculated value corresponding to the variation function is greater than a predetermined threshold (i.e., a corresponding threshold). If the variation function value is positive and exceeds a predetermined threshold (a set upper threshold), it indicates that the prediction is fast and sensitive, and α and β are increased by a predetermined step size to reduce the weights of the neighboring cell load and the predicted load. Conversely, if the variation function value is negative and lower than a predetermined threshold (a set lower threshold), it indicates that the prediction is slow and insensitive, and α and β are decreased by a predetermined step size to increase the weights of the neighboring cell load and the predicted load. In this embodiment, the default setting value of the predetermined step size is 0.05.

[0082] Among the above steps, the steps of respectively acquiring the total predicted load corresponding to the current time and the real-time load corresponding to the current time, and respectively acquiring the historical total predicted load corresponding to the previous time and the historical actual load corresponding to the previous time, generating a variation function of the load prediction based on at least two of the actual load corresponding to the current time, the total predicted load corresponding to the current time, the historical total predicted load corresponding to the previous time, and the historical actual load corresponding to the previous time, and determining a calculated value corresponding to the corresponding variation function, wherein the variation function is used to characterize the prediction performance of the load prediction, and determining whether the calculated value corresponding to the variation function is greater than a predetermined threshold and adjusting a weighting coefficient corresponding to the load prediction model based on the determination result, realize real-time adjustment of the model parameters of the load prediction model based on the prediction result, thereby achieving both accuracy of the load prediction and an extension of the detection distance corresponding to the load prediction.

[0083] In some alternative embodiments, the variation function includes one of a difference value between the actual load corresponding to the current time and the historical total predicted load corresponding to the previous time, a difference value between the actual load corresponding to the current time and the historical total predicted load corresponding to the previous time and a ratio of the actual load corresponding to the current time, a difference value between the historical total predicted load corresponding to the previous time and the total predicted load corresponding to the current time, a difference value between the actual load corresponding to the current time and the historical actual load corresponding to the previous time and a difference value between the total predicted load corresponding to the current time and the historical total predicted load corresponding to the previous time, and a load difference value generated by weighting the difference value between the actual load corresponding to the current time and the historical actual load corresponding to the previous time.

[0084] In this embodiment, each base station cell is supported to optimize the corresponding weight coefficient after determining the corresponding load forecasting result, that is, to maintain and optimize the model parameters of the load forecasting model, and form a personalized dynamic load forecasting model that suits itself. In some alternative embodiments, the following formula is adopted to optimize the corresponding model parameters, that is, to define and generate a variation function according to the following formula:

[0085] TIFF2025527012000008.tif28170

[0086] TIFF2025527012000009.tif12170

[0087] TIFF2025527012000010.tif9170

[0088] TIFF2025527012000011.tif25170

[0089] TIFF2025527012000012.tif17170

[0090] TIFF2025527012000013.tif12170

[0091] TIFF2025527012000014.tif7170

[0092] TIFF2025527012000015.tif31170

[0093] In the second method, the model parameters are adjusted in real time based on changes in the actual and predicted loads to achieve both improved prediction accuracy and a longer detection and sensing distance. When the predicted load is increasing, i.e., when the train load is approaching, α and β are increased by a predetermined step size to better focus on the load of the target base station cell itself and the actual load, thereby improving prediction accuracy. When the predicted load is decreasing, i.e., when the train load is moving away, α and β are decreased by a predetermined step size to better focus on the load of adjacent cells and the predicted load, thereby extending the distance for detecting and sensing the load of adjacent cells of the base station.

[0094] TIFF2025527012000016.tif14170

[0095] TIFF2025527012000017.tif56170

[0096] In this embodiment, each base station cell determines the center of gravity of the overall load of the high-speed private network segment based on its own actual load and the actual load of each base station neighbor cell in the cross-layer unidirectional group that crosses the corresponding K layer. The center of gravity of the overall load is then determined based on the change in the center of gravity of the overall load, along with the direction of train movement on the high-speed rail track corresponding to the high-speed private network, and the resulting direction of load movement. If it is determined that the center of gravity of the overall load of the high-speed private network segment is moving in a positive direction from left to right, γ is decreased by a predetermined step size; otherwise, it is increased. In this embodiment, the predetermined step size can be 0.05, and the adjustment of γ must not exceed a predetermined value.

[0097] In some embodiments, the step of receiving the cross-layer neighboring cell load data transmitted by each of the left neighboring cell neighboring group and the right neighboring cell neighboring group of the cross-layer neighboring group is realized by the following steps 61 to 62.

[0098] Step 61: After the nearest left neighboring cell detects the first target load data from the first neighboring cell group load data it receives, it combines the first target load data, its own corresponding actual load at the current time and its own historical predicted load corresponding to the previous time into first cross-layer neighboring cell load data in a predetermined format and transmits it to the corresponding target base station cell, where the first neighboring cell group load data includes the cross-layer neighboring cell load data corresponding to the left neighboring cell co-directional group of the nearest left neighboring cell itself, and the first target load data includes the cross-layer neighboring cell load data corresponding to all primary neighboring cells after excluding its Kth layer primary neighboring cell from the left neighboring cell co-directional group of the nearest left neighboring cell itself.

[0099] In some alternative embodiments, the target base station cell is C8, the left neighbor cell co-directional group corresponding to the target base station cell C8 is {C2, C3, C4, C5, C6, C7}, the nearest left neighbor cell corresponding to the target base station cell C8 is C7, and receiving the cross-layer neighbor cell load data transmitted by the target base station cell C8 in its left neighbor cell co-directional group means that the nearest left neighbor cell C7 receives the cross-layer neighbor cell load data M0={L(1), L(2), L(3), L(4), L(5), L(6)} of its own left neighbor cell co-directional group {C1, C2, C3, C4, C5, C6}, and the nearest left neighbor cell C7 removes the load L(1) corresponding to the K-th layer primary neighbor cell C1 from its left neighbor cell co-directional group from M0, and then adds the load L(7) that the nearest left neighbor cell C7 transmits to the target base station cell C8 at the current time, thereby forming the corresponding first cross-layer neighbor cell load data, that is, M1={L(2), L(3), L(4), L(5), L(6)}. L(4), L(5), L(6), L(7)} and includes the process of transmitting to the target base station cell C8.

[0100] Step 62: After the nearest right neighboring cell detects the second target load data from the received second neighboring cell group load data, it combines the second target load data, its own corresponding actual load at the current time and its own corresponding historical predicted load at the previous time into second cross-layer neighboring cell load data in a predetermined format and transmits it to the corresponding target base station cell, where the second neighboring cell group load data includes the cross-layer neighboring cell load data corresponding to the nearest right neighboring cell's own right neighboring cell co-directional group, and the second target load data includes the cross-layer neighboring cell load data corresponding to all primary neighboring cells after excluding its Kth layer primary neighboring cell from the nearest right neighboring cell's own right neighboring cell co-directional group.

[0101] In some alternative embodiments, the target base station cell is C8, and the right neighbor cell co-directional group corresponding to the target base station cell C8 is {C9, C 10 , C 11 , C12 , C 13 , C 14}, and the nearest right neighbor cell corresponding to the target base station cell C8 is C9. Receiving the cross-layer neighbor cell load data transmitted by the target base station cell C8 in its right neighbor cell co-directional group means that the nearest right neighbor cell C9 is in its own left neighbor cell co-directional group {C 10 , C 11 , C 12 , C 13 , C 14 , C 15}, and the nearest right neighbor cell C9 receives the cross-layer neighbor cell load data M2={L(10),L(11),L(12),L(13),L(14),L(15)} of the K-th layer primary neighbor cell C 15 The process includes the steps of deleting the corresponding load L(15) from M2, and then adding the corresponding load L(9) that it transmits to the target base station cell C8 at the current time, thereby constructing the corresponding second cross-layer neighboring cell load data, that is, M3={L(9), L(10), L(11), L(12), L(13), L(14)}, and transmitting it to the target base station cell C8.

[0102] In this embodiment, the nearest right neighboring cell and the nearest left neighboring cell group constitute the left and right primary neighboring cells of the corresponding base station cell. In this embodiment, the load information transmitted by the base station cell is its own real-time load at the current time, its historical predicted load corresponding to the previous time, and the load information based on the primary neighboring cell in the corresponding cross-layer unidirectional group.

[0103] In this embodiment, before preparing to transmit corresponding load data to a corresponding target base station cell, the nearest left neighboring cell first receives corresponding cross-layer neighboring cell load data transmitted by its corresponding left neighboring cell co-directional group, at which time, the primary neighboring cell of the Kth layer in the left neighboring cell co-directional group corresponding to the nearest left neighboring cell does not belong to the primary neighboring cell in the left neighboring cell co-directional group corresponding to the target base station cell, so the load data transmitted by the nearest left neighboring cell does not include the load data of the primary neighboring cell of the Kth layer in the left neighboring cell co-directional group corresponding to the nearest left neighboring cell. Before preparing to transmit corresponding load data to a corresponding target base station cell, the nearest right neighboring cell first receives corresponding cross-layer neighboring cell load data transmitted by its corresponding right neighboring cell co-directional group, at which time, the primary neighboring cell of the Kth layer in the right neighboring cell co-directional group corresponding to the nearest right neighboring cell does not belong to the primary neighboring cell in the right neighboring cell co-directional group corresponding to the target base station cell, so the load data transmitted by the target base station cell does not include the load data of the primary neighboring cell of the Kth layer in the right neighboring cell co-directional group corresponding to the nearest right neighboring cell.

[0104] In this embodiment, the loads of the primary neighboring cells in the Kth layer of the left neighboring cell same-direction group corresponding to the nearest left neighboring cell / right neighboring cell same-direction group corresponding to the nearest right neighboring cell are deleted from the neighboring cell group load data stored in a predetermined format, and then the real-time loads of the nearest left neighboring cell / nearest right neighboring cell at the current time and the historical predicted loads corresponding to the previous time are added, and the existing corresponding load data is further generated in a predetermined format and transmitted to the target base station cell.

[0105] Through the above steps, the base station cells can transmit load information in a directional relay manner.

[0106] In some embodiments, determining the cross-layer co-directional group in a plurality of base station cells located on a predetermined load transfer route is achieved by the following steps 71 to 73.

[0107] Step 71: Obtain switching frequency granularity data corresponding to each base station cell on a predetermined load transfer route, and select the neighboring cell with the largest switching frequency granularity data as the primary neighboring cell of the base station cell, where the switching frequency granularity data is used to characterize the number of times a user switches between the base station cell and the neighboring cell corresponding to the base station cell within a predetermined time period.

[0108] Step 72: The left primary adjacent cell and the right primary adjacent cell corresponding to each of the selected base station cells are set as a base station cell co-directional group corresponding to the base station cell.

[0109] Step 73: Starting from the target base station cell, select base station cell co-directional groups corresponding to K consecutive base station cells along the forward and reverse directions of the predetermined load migration route, respectively, and perform a de-duplicating process on the base station cells in the selected base station cell co-directional groups to obtain a cross-layer co-directional group.

[0110] In this embodiment, the base station cell C i is the time-granularity statistical data of the number of "neighbor cell pair" switchings for a certain period of time (for example, 168 hours corresponding to one cycle) and each of its neighboring cells (the number of times switching to a neighboring cell and the number of times the neighboring cell is a base station cell C) i Based on the number of times the neighboring cells have switched to the base station cell C, the neighboring cells are divided into two types according to a clustering algorithm. The neighboring cells with a higher number of switching times are designated as the base station cell C. i Of course, the network planner and optimizer will decide the base station cell C, such as the implementation location and coverage direction. i and the base station cell C based on the network engineering parameter information of the neighboring cells or by further referring to the switching statistical data. iManually screen the nearest neighboring cells on both the left and right sides along the direction of the high-speed railway track, and select base station cell C i The base station cell C can be the primary neighboring cell of i The nearest neighboring cells on both the left and right sides are base station cell C. i Then, a cross-layer co-directional group corresponding to each base station cell is determined based on the relationship between the co-directional groups corresponding to multiple base station cells. For example, if C2 is the right adjacent cell of the C1 co-directional group and C3 is the right adjacent cell of the C2 co-directional group, C3 can be related to C1 as the right adjacent cell crossing the 1st layer, and similarly, C1 is the right adjacent cell crossing the Kth layer. k {C1,C2,...C K} constitutes a cross-layer same direction group in which C1 crosses the K layer.

[0111] In this embodiment, each primary adjacent cell of the base station cell is matched and screened into left primary adjacent cells and right primary adjacent cells according to the load transition time-spatial characteristic pattern on the same direction route, and divided into same direction route groups, i.e., same direction groups.

[0112] Target base station cell C i For the primary neighboring cell, the time period with the most number of handovers is selected from the handover statistical data with a granularity of one hour per cycle. For example, if two time periods with the most number of handovers per day are selected, there will be 14 time periods per cycle. From these time periods, user number statistical data D2 with a granularity of 10 to 60 seconds (for example, 30 seconds) is obtained, and D2 is classified into two categories using a clustering algorithm, with a load with a large number of users being set to 1 and a load with a small number of users being set to 0. Next, the primary neighboring cell 1 and primary neighboring cell 2 with load transition spatiotemporal characteristics (which only need to satisfy one of them) as shown in Table 1 and Table 2 are searched and screened, and the screened primary neighboring cell 1 and primary neighboring cell 2 are used as the target base station cell C. i Let the left and right adjacent cells be the cells of .

[0113] [Table 1]

[0114] [Table 2]

[0115] In this embodiment, the cross-layer co-directional group of each base station cell is determined based on multiple related co-directional groups. For example, if C2 is the right adjacent cell of the C1 co-directional group and C3 is the right adjacent cell of the C2 co-directional group, C3 can be related to C1 as the right adjacent cell crossing the first layer, and so on to obtain the right adjacent cell Ck crossing the Kth layer of C1. {C1, C2, ... C K} constitutes a cross layer co-directional group where C1 crosses the Kth layer. In the embodiment of the present disclosure, the cross layer co-directional group corresponding to the determined target base station cell includes a left and a right cross layer co-directional group, for example, a left adjacent cell co-directional group and a right adjacent cell co-directional group, where the left adjacent cell co-directional group and the right adjacent cell co-directional group both have a Kth layer primary adjacent cell, where K is a default value, for example, the target base station cell C i If it is desired to predict future loads earlier and sense loads over longer distances, K can be set larger, but this increases the overhead and difficulty of cross-layer transmission load information, and also increases random uncertainty. Therefore, the K value must be set taking into account the balance between the target requirements of load forecasting and overhead costs, for example, K = 9. As shown in Figure 2, {C1, C2, C3, C4, C5, C6} are cross-layer left-side neighbor cells with C7 crossing 6 layers, and {C8, C9, C 10 , C 11 ,C 12 ,C 13} is a cross-layer same-direction right adjacent cell to C7 crossing the 6th layer, and load information may pass through the 6th layer to the left or right by relay transmission.

[0116] To illustrate load prediction in the embodiments of the present disclosure, Examples 1 to 6 are provided below.

[0117] TIFF2025527012000020.tif48170

[0118] Example 1 When simulating a single-track high-speed railway moving in the positive direction from left to right, the actual load and predicted load are shown in Table 3 and Table 4, respectively. From the predicted load results, it can be seen that the prediction model can detect and sense the impending load and change trend in advance. By combining the actual load and predicted load, the base station cell can prepare operations such as energy saving or load control in advance.

[0119] [Table 3]

[0120] [Table 4]

[0121] Example 2 A single-track high-speed railway moving in the opposite direction from right to left is simulated. The actual load and predicted load are shown in Table 5 and Table 6, respectively. From the predicted load results, the prediction model can detect and sense the impending load and change trend in advance. By combining the actual load and predicted load, the base station cell can prepare operations such as energy saving or load control in advance.

[0122] [Table 5]

[0123] [Table 6]

[0124] Example 3 A simulation was performed for a double-track high-speed railway traveling from left to right and from right to left. The actual load and predicted load are shown in Tables 7 and 8, respectively. The predicted load results show that the prediction model can detect and sense impending loads and change trends in advance. By combining the actual load and predicted load, the base station cell can prepare for operations such as energy saving or load control in advance.

[0125] [Table 7]

[0126] [Table 8]

[0127] Example 4 A double-track high-speed railway is simulated, with two starting points simultaneously moving in a forward direction from left to right. The actual load and predicted load are shown in Tables 9 and 10, respectively. The predicted load results show that the prediction model can detect and sense impending loads and change trends in advance. By combining the actual load and predicted load, the base station cell can prepare for operations such as energy saving or load control in advance.

[0128] [Table 9]

[0129] [Table 10]

[0130] Example 5 In the process of simulating a single-track high-speed train moving in the forward direction from left to right, a single-track high-speed train moving in the reverse direction is randomly added to a certain cell at a certain time. The actual load and predicted load are shown in Table 11 and Table 12, respectively. Judging from the predicted load results, the prediction model can detect and sense impending load and change trends in advance. By combining the actual load and predicted load, the base station cell can prepare operations such as energy saving or load control in advance.

[0131] [Table 11]

[0132] [Table 12]

[0133] Example 6 During the simulation of a single-track high-speed railway moving in the positive direction from left to right, when it reaches an intermediate cell, the number of users is randomly increased or decreased to simulate the load change in the number of users due to passengers getting on and off at the arrival station. The actual load and predicted load are shown in Tables 13 and 14, respectively. Judging from the predicted load results, the prediction model can detect and sense impending load and change trends in advance. By combining the actual load and predicted load, the base station cell can prepare in advance for operations such as energy saving or load control.

[0134] [Table 13]

[0135] [Table 14]

[0136] The embodiments of the present disclosure further provide a dynamic load prediction device, which is for realizing the above-described embodiments and optional embodiments, and what has already been described will not be described again. The terms "module," "unit," "subunit," and the like used below may refer to a combination of software and / or hardware capable of realizing a predetermined function. While the devices described in the following embodiments are preferably realized by software, they may also be realized by hardware or a combination of software and hardware, and this may be considered.

[0137] FIG. 3 is a block diagram of a dynamic load prediction device provided by an embodiment of the present disclosure. As shown in FIG. 3, the device includes an acquisition module 32, an acquisition module 32, a receiving module 33, and a prediction module 34.

[0138] The determination module 31 is configured to determine a cross-layer unidirectional group for a plurality of base station cells located on a predetermined load transfer route, and the base station cells located at both ends of the cross-layer unidirectional group are both K-th layer primary neighbor cells corresponding to the target base station cell, where K is a default value.

[0139] The acquisition module 32 is coupled to the determination module 31 and configured to acquire a real-time load at a current time and a historical predicted load corresponding to a previous time, the historical predicted load being used to characterize a load transient change that occurs at the current time and that is predicted at the previous time.

[0140] The receiving module 33 is coupled to the acquiring module 32 and configured to receive cross-layer neighboring cell load data transmitted by each of the left neighboring cell neighboring group and the right neighboring cell neighboring group of the cross-layer unidirectional group, and the corresponding cross-layer neighboring cell load data both include historical load data and real-time load data, the historical load data is used to characterize the corresponding predicted load of the Kth layer primary neighboring cell of the corresponding neighboring cell unidirectional group at the previous time, and the real-time load data is used to characterize the actual load of the Kth layer primary neighboring cell of the corresponding neighboring cell unidirectional group at the current time.

[0141] The prediction module 34 is coupled to the receiving module 33 and is configured to use a preset load prediction model to perform load prediction based on real-time load, historical predicted load, and cross-layer adjacent cell load data, obtain forward predicted load and reverse predicted load corresponding to the target base station cell, and determine corresponding load prediction results based on the forward predicted load and reverse predicted load.

[0142] The dynamic load forecasting device of the embodiment of the present disclosure includes the steps of: determining a cross-layer unidirectional group among a plurality of base station cells located on a predetermined load migration route, wherein base station cells located at both ends of the cross-layer unidirectional group are K-th layer primary neighbor cells corresponding to a target base station cell, K is a predetermined value; obtaining a real-time load at a current time and a historical predicted load corresponding to a previous time, wherein the historical predicted load is used to characterize a load migration change occurring at the current time predicted at the previous time; and receiving cross-layer neighbor cell load data transmitted by each of a left neighbor cell unidirectional group and a right neighbor cell unidirectional group of the cross-layer unidirectional group, wherein the corresponding cross-layer neighbor cell load data both include historical load data and real-time load data, wherein the historical load data is used to characterize a corresponding predicted load at the previous time of a K-th layer primary neighbor cell included in the corresponding neighbor cell unidirectional group. The present invention employs the steps of: using real-time load data to characterize the actual load at the current time of the Kth layer primary neighbor cell of the corresponding neighbor cell unidirectional group; and using a preset load prediction model to perform load prediction based on the real-time load, historical predicted load, and cross-layer neighbor cell load data, to obtain a forward direction predicted load and a backward direction predicted load corresponding to the target base station cell, and determining a corresponding load prediction result based on the forward direction predicted load and the backward direction predicted load. This solves the problem in the related art that load prediction methods cannot be applied to high-speed mobile communication scenarios in which users move at high speed, the load is sudden, and there is no precise time period regularity. The present invention enables a base station cell to accurately detect and sense the user load and change trend of users arriving at a neighbor cell that is farther away in advance. The combination of the predicted load and actual load of the base station cell provides a beneficial effect of providing an effective reference and guide for operations such as energy saving, load control, and call guarantee of the base station cell.

[0143] In some embodiments, the prediction module 34 is further configured to use a load prediction model to perform a forward load calculation on the real-time load, the historical predicted load, and the historical load data and real-time load data corresponding to the left adjacent cell same-direction group to obtain a forward predicted load, and to use a load prediction model to perform a reverse load calculation on the real-time load, the historical predicted load, and the historical load data and real-time load data corresponding to the right adjacent cell same-direction group to obtain a reverse predicted load.

[0144] In some embodiments, the device includes: obtaining a first actual load at a current time of each base station cell located on a predetermined load transfer route; determining a nearest left neighboring cell and a nearest right neighboring cell corresponding to each base station cell; respectively obtaining actual loads corresponding to a previous time of the nearest right neighboring cell and the nearest left neighboring cell corresponding to the corresponding base station cell; a positive direction is from the left end to the right end of the predetermined load transfer route; the nearest left neighboring cell is used to characterize a first base station cell located on the left side of the corresponding base station cell in the positive direction; and the nearest right neighboring cell is used to characterize a first base station cell located on the right side of the corresponding base station cell in the positive direction; Further, the method may be used to determine a first transition load that will be transitioned to the base station cell at a first predetermined rate at the previous time of the nearest right neighboring cell based on the actual load corresponding to the nearby right neighboring cell at the previous time, and to determine a second transition load that will be transitioned to the base station cell at a second predetermined rate at the previous time of the nearest left neighboring cell based on the actual load corresponding to the nearest left neighboring cell at the previous time, and to determine a real-time load that will be transitioned to the base station cell when a forward prediction is performed at the current time based on the difference between the first actual load and the first transition load, and to determine a real-time load that will be transitioned to the base station cell when a backward prediction is performed at the current time based on the difference between the first actual load and the second transition load.

[0145] TIFF2025527012000033.tif214170

[0146] In some embodiments, the prediction module 34 is further configured to sum the forward predicted load and the backward predicted load to obtain a first predicted total load corresponding to the target base station on a predetermined load shifting route, and accumulate the first predicted total loads corresponding to multiple predetermined load shifting routes to generate a total predicted load, where the load prediction result includes the total predicted load.

[0147] In some embodiments, after determining the corresponding load forecast result, the device respectively obtains a total predicted load corresponding to the current time and a real-time load corresponding to the current time, and a historical total predicted load corresponding to a previous time and a historical actual load corresponding to the previous time; generates a load forecast variation function based on at least two of the actual load corresponding to the current time, the total predicted load corresponding to the current time, the historical total predicted load corresponding to the previous time, and the historical actual load corresponding to the previous time; determines a calculated value corresponding to the corresponding variation function, where the variation function is used to characterize the prediction performance of the load forecast; determines whether the calculated value of the variation function is greater than a predetermined threshold; and adjusts a weighting coefficient corresponding to the load forecast model based on the determination result, where the weighting coefficient corresponding to the load forecast model is greater than a predetermined threshold. i It is further used to include the weight α of the load of the base station itself, the weight β of the actual load of each base station cell, and the first predetermined ratio γ.

[0148] In some embodiments, the variation function includes one of a differential value between the actual load corresponding to the current time and the historical total predicted load corresponding to the previous time, a differential value between the actual load corresponding to the current time and the historical total predicted load corresponding to the previous time and a ratio of the actual load corresponding to the current time, a differential value between the historical total predicted load corresponding to the previous time and the total predicted load corresponding to the current time, a differential value between the actual load corresponding to the current time and the historical actual load corresponding to the previous time and a differential value between the total predicted load corresponding to the current time and the historical total predicted load corresponding to the previous time, and a load differential value generated by weighting the differential value between the actual load corresponding to the current time and the historical actual load corresponding to the previous time.

[0149] In some embodiments, the acquisition module 31 is further configured to: acquire switching frequency granularity data corresponding to each base station cell on a predetermined load transfer route; select the neighboring cell with the largest switching frequency granularity data as the primary neighboring cell of the base station cell, where the switching frequency granularity data is used to characterize the number of times a user switches between the base station cell and the neighboring cell corresponding to the base station cell within a predetermined time; determine the left primary neighboring cell and the right primary neighboring cell corresponding to each selected base station cell as a base station cell co-directional group corresponding to the base station cell; and, starting from the target base station cell, respectively select base station cell co-directional groups corresponding to K consecutive base station cells along the forward and reverse directions of the predetermined load transfer route, and perform overlap elimination processing on the base station cells in the selected base station cell co-directional groups to obtain a cross-layer co-directional group.

[0150] FIG. 4 is a conceptual diagram of a base station in an embodiment of the present disclosure. As shown in FIG. 4, the embodiment of the present disclosure provides a base station comprising a processor 41, a communication interface 42, a memory 43, and a communication bus 44. The processor 41, the communication interface 42, and the memory 43 communicate with each other via the communication bus 44. The memory 43 stores a computer program. The processor 41 executes the method in FIG. 1 when the program stored in the memory 43 is executed.

[0151] The processing in the base station implements the method steps in FIG. 1, and the technical effects achieved thereby are the same as the technical effects achieved when the dynamic load forecasting method in FIG. 1 is implemented in the above embodiment, and will not be described again here.

[0152] The communication bus mentioned in the above base station may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus may be divided into an address bus, a data bus, a control bus, etc. Although only one thick line is used in Figure 4 for ease of illustration, this does not mean that there is only one bus or only one type of bus.

[0153] The communication interface is used for communication between the terminal and other devices.

[0154] The memory may include random access memory (RAM) and may also include non-volatile memory such as at least one disk memory. Optionally, the memory may be at least one storage device remote from said processor.

[0155] The processor may be a general-purpose processor including a central processing unit (CPU), a network processor (NP), or the like, or may be a digital signal processing device (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPIC), a programmable gate array (FPGA), or other programmable logic device, discrete gate or transistor logic device, or discrete hardware component.

[0156] An embodiment of the present disclosure further provides a computer-readable medium having a computer program stored thereon, the computer program being capable of performing the dynamic load prediction method provided by any one of the method embodiments above when executed by a processor.

[0157] In another embodiment provided by the present disclosure, there is further provided a computer program product including instructions that, when executed on a computer, cause the computer to perform the dynamic load prediction method of any one of the preceding embodiments.

[0158] Compared with the related art, the embodiments of the present disclosure provide a dynamic load forecasting method, device, base station and storage medium, including the steps of: determining a cross-layer unidirectional group for a plurality of base station cells located on a predetermined load migration route, where the base station cells located at both ends of the cross-layer unidirectional group are K-th layer primary neighbor cells corresponding to a target base station cell, K is a predetermined value; obtaining a real-time load at a current time and a historical predicted load corresponding to a previous time, where the historical predicted load is used to characterize a load migration change that occurs at the current time predicted at the previous time; receiving cross-layer neighbor cell load data transmitted by each of the left neighbor cell unidirectional group and the right neighbor cell unidirectional group of the cross-layer unidirectional group, where the corresponding cross-layer neighbor cell load data both include historical load data and real-time load data, and the historical load data is a predicted load data corresponding to the K-th layer primary neighbor cell of the corresponding neighbor cell unidirectional group at the previous time. The real-time load data is used to characterize the load, and the real-time load data is used to characterize the actual load at the current time of the Kth layer primary neighbor cell in the corresponding neighbor cell unidirectional group; and the load prediction method uses a preset load prediction model to perform load prediction based on the real-time load, historical predicted load, and cross-layer neighbor cell load data, to obtain the forward direction predicted load and backward direction predicted load corresponding to the target base station cell, and determine the corresponding load prediction result based on the forward direction predicted load and backward direction predicted load. This solves the problem in the prior art that the load prediction method cannot be applied to high-speed mobile communication scenarios such as users moving at high speed, load being sudden, and no precise time period regularity. The base station cell can detect and sense the user load and change trend arriving at a neighbor cell that is farther away in advance at an accurate time. In addition, by combining the predicted load and actual load of the base station cell, it is possible to achieve the beneficial effect of providing effective reference and guidance for operations such as energy saving, load control, and call guarantee of the base station cell.

[0159] It should be noted that in this disclosure, relational terms such as "first," "second," and the like are used only to distinguish one entity or operation from another and do not necessarily require or imply any actual relationship or order between those entities or operations. Furthermore, the terms "comprise," "contain," and any other variations thereof are intended to cover non-exclusive inclusions, so that a process, method, product, or apparatus comprising a set of elements not only includes those elements, but also other elements not expressly listed or inherent in such process, method, product, or apparatus. Absent further qualification, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, product, or apparatus that includes that element.

[0160] The above is merely a specific embodiment of the present disclosure to enable those skilled in the art to understand or realize the present disclosure. Various modifications to these examples will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other examples without departing from the spirit or scope of the present disclosure. Therefore, the present disclosure is not intended to be limited to the examples shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. 1. A dynamic load prediction method applied to a target base station cell under a high speed private network, comprising: determining a cross-layer unidirectional group in a plurality of base station cells located on a predetermined load transfer route, wherein the base station cells located at both ends of the cross-layer unidirectional group are K-th layer primary neighbor cells corresponding to the target base station cell, where K is a predetermined value; obtaining a real-time load for a current time and a historical forecast load corresponding to a previous time, the historical forecast load being used to characterize a load transient change that occurs at the current time and that was predicted at the previous time; receiving cross-layer neighboring cell load data transmitted by each of the left neighboring cell unidirectional group and the right neighboring cell unidirectional group of the cross-layer unidirectional group, wherein the corresponding cross-layer neighboring cell load data both include historical load data and real-time load data, the historical load data being used to characterize the corresponding predicted load of the Kth layer primary neighboring cell included in the corresponding neighboring cell unidirectional group at a previous time, and the real-time load data being used to characterize the actual load of the Kth layer primary neighboring cell included in the corresponding neighboring cell unidirectional group at a current time; performing load prediction based on the real-time load, the historical predicted load, and the cross-layer neighboring cell load data using a preset load prediction model, obtaining a forward predicted load and a backward predicted load corresponding to the target base station cell, and determining a corresponding load prediction result based on the forward predicted load and the backward predicted load. Dynamic load forecasting methods.

2. 10. The method of claim 1, performing load prediction based on the real-time load, the historical predicted load, and the cross-layer neighboring cell load data using a preset load prediction model, and obtaining a forward direction predicted load and a backward direction predicted load corresponding to the target base station cell, using the load forecasting model to perform a forward load calculation on the real-time load, the historical forecast load, and the historical load data and real-time load data corresponding to the left adjacent cell same-direction group, to obtain the forward forecast load; and performing a reverse load calculation on the real-time load, the historical predicted load, and the historical load data and the real-time load data corresponding to the right neighbor cell same-direction group using the load forecasting model to obtain the reverse predicted load. The method of claim 1.

3. 3. The method of claim 2, acquiring a first actual load at a current time of each of the base station cells located on a predetermined load transfer route; determining a nearest left neighboring cell and a nearest right neighboring cell corresponding to each of the base station cells, and respectively obtaining the nearest right neighboring cell and the actual load of the nearest left neighboring cell corresponding to the corresponding base station cell, corresponding to a previous time, wherein a positive direction is defined as a direction from the left end to the right end of the predetermined load transfer route, the nearest left neighboring cell is used to characterize a first base station cell located to the left of the corresponding base station cell in the positive direction, and the nearest right neighboring cell is used to characterize a first base station cell located to the right of the corresponding base station cell in the positive direction; determining a first transition load of the nearest right neighboring cell to be transitioned to the base station cell at a first predetermined rate at the previous time based on an actual load corresponding to the nearest right neighboring cell at the previous time, and determining a second transition load of the nearest left neighboring cell to be transitioned to the base station cell at a second predetermined rate at the previous time based on an actual load corresponding to the nearest left neighboring cell at the previous time; determining the real-time load corresponding to a case where the base station cell performs forward prediction at the current time based on a difference value between the first actual load and the first transitional load, and determining the real-time load corresponding to a case where the base station cell performs backward prediction at the current time based on a difference value between the first actual load and the second transitional load. The method of claim 2.

4.

5. 5. The method of claim 4, determining a corresponding load forecast result based on the forward predicted load and the backward predicted load, summing the forward predicted load and the backward predicted load to obtain a first predicted total load corresponding to the target base station along the predetermined load transfer route; and accumulating the first total predicted loads corresponding to a plurality of predetermined load transfer routes to generate a total predicted load, wherein the load forecast result includes the total predicted load. The method of claim 4.

6. 6. The method of claim 5, After determining the corresponding load forecast result, the method includes: acquiring the total predicted load corresponding to a current time and an actual load corresponding to the current time, and acquiring a historical total predicted load corresponding to a previous time and a historical actual load corresponding to the previous time; generating a variation function of a load forecast based on at least two of an actual load corresponding to a current time, the total forecasted load corresponding to the current time, a historical total forecasted load corresponding to a previous time, and a historical actual load corresponding to a previous time, and determining a calculated value corresponding to the variation function, wherein the variation function is used to characterize a predictive performance of the load forecast; determining whether a calculated value corresponding to the variation function is greater than a predetermined threshold, and adjusting a weighting factor corresponding to the load prediction model based on the determination result, wherein the weighting factor corresponding to the load prediction model is adjusted based on the target base station cell C ｉ a weight α of the load of the base station itself, a weight β of the actual load of each base station cell, and the first predetermined ratio γ. The method of claim 5.

7. The variation function includes one of a difference value between an actual load corresponding to the current time and a historical total predicted load corresponding to the previous time, a difference value between an actual load corresponding to the current time and a historical total predicted load corresponding to the previous time and a ratio of the actual load corresponding to the current time, a difference value between a historical total predicted load corresponding to the previous time and the total predicted load corresponding to the current time, a difference value between an actual load corresponding to the current time and a historical actual load corresponding to the previous time and a difference value between the total predicted load corresponding to the current time and the historical total predicted load corresponding to the previous time, and a load difference value generated by weighting the difference value between an actual load corresponding to the current time and a historical actual load corresponding to the previous time. The method of claim 6.

8.

9. 4. The method of claim 3, receiving cross-layer neighboring cell load data transmitted by each of the left neighboring cell co-directional group and the right neighboring cell co-directional group of the cross-layer co-directional group; After the nearest left neighboring cell detects first target load data from its received first neighboring cell group load data, the nearest left neighboring cell combines the first target load data, its corresponding actual load at its current time and its corresponding historical predicted load at a previous time into first cross-layer neighboring cell load data in a predetermined format and transmits it to the corresponding target base station cell, wherein the first neighboring cell group load data includes cross-layer neighboring cell load data corresponding to the nearest left neighboring cell's own left neighboring cell co-directional group, and the first target load data includes cross-layer neighboring cell load data corresponding to all the primary neighboring cells after excluding its K-th layer primary neighboring cell from the nearest left neighboring cell's own left neighboring cell co-directional group; and after the nearest right neighboring cell detects second target load data from the received second neighboring cell group load data, synthesizes the second target load data, its corresponding actual load at its own current time and its corresponding historical predicted load at a previous time into second cross-layer neighboring cell load data in a predetermined format, and transmits the second cross-layer neighboring cell load data to the corresponding target base station cell, wherein the second neighboring cell group load data includes cross-layer neighboring cell load data corresponding to the nearest right neighboring cell's own right neighboring cell co-directional group, and the second target load data includes cross-layer neighboring cell load data corresponding to all the primary neighboring cells after excluding the K-th layer primary neighboring cell from the nearest right neighboring cell's own right neighboring cell co-directional group. The method of claim 3.

10. 2. The method according to claim 1, wherein the step of determining cross-layer co-directional groups in a plurality of base station cells located on a predetermined load transfer route comprises: obtaining switching frequency granularity data corresponding to each of the base station cells on a predetermined load transfer route, and selecting a neighboring cell with the largest switching frequency granularity data as a primary neighboring cell of the base station cell, wherein the switching frequency granularity data is used to characterize the number of times a user switches between the base station cell and a neighboring cell corresponding to the base station cell within a predetermined time period; a step of setting a left primary neighboring cell and a right primary neighboring cell corresponding to each selected base station cell as a base station cell co-directional group corresponding to the base station cell; selecting the base station cell co-directional groups corresponding to the K consecutive base station cells along the forward and reverse directions of the predetermined load transfer route, starting from the target base station cell, and performing a process of removing overlaps between the base station cells in the selected base station cell co-directional groups to obtain the cross layer co-directional group. The method of claim 1.

11. A dynamic load prediction apparatus applied to a target base station cell under a high-speed private network, comprising: a determining module, an acquiring module, a receiving module, and a predicting module; The determination module is configured to determine a cross-layer unidirectional group among a plurality of base station cells located on a predetermined load transfer route, and the base station cells located at both ends of the cross-layer unidirectional group are K-th layer primary neighbor cells corresponding to the target base station cell, where K is a predetermined value; The acquisition module is configured to acquire a real-time load at a current time and a historical predicted load corresponding to a previous time, the historical predicted load being used to characterize a load transient change that occurs at the current time and that was predicted at the previous time; The receiving module is configured to receive cross-layer neighboring cell load data transmitted by each of the left neighboring cell unidirectional group and the right neighboring cell unidirectional group of the cross-layer unidirectional group, and the corresponding cross-layer neighboring cell load data both include historical load data and real-time load data, the historical load data is used to characterize the corresponding predicted load of the Kth layer primary neighboring cell included in the corresponding neighboring cell unidirectional group at a previous time, and the real-time load data is used to characterize the actual load of the Kth layer primary neighboring cell included in the corresponding neighboring cell unidirectional group at a current time; The prediction module is configured to perform load prediction based on the real-time load, the historical predicted load, and the cross-layer neighboring cell load data using a preset load prediction model, obtain a forward predicted load and a backward predicted load corresponding to the target base station cell, and determine a corresponding load prediction result based on the forward predicted load and the backward predicted load. Dynamic load forecaster.

12. A base station comprising a processor, a communication interface, a memory, and a communication bus, The processor, the communication interface, and the memory communicate with each other via a communication bus. The memory stores computer programs, The processor executes the dynamic load prediction method according to any one of claims 1 to 10 when the program stored in the memory is executed. Base station.

13. A computer-readable medium having a computer program stored thereon, The computer program, when executed by a processor, performs the dynamic load prediction method according to any one of claims 1 to 10. Computer-readable medium.

Images