CN116233981A

CN116233981A - Method and device for adjusting power of Long Term Evolution (LTE) base station

Info

Publication number: CN116233981A
Application number: CN202211693586.5A
Authority: CN
Inventors: 李永恒; 张梦杰
Original assignee: Hangzhou Eastcom Software Technology Co ltd
Current assignee: Hangzhou Eastcom Software Technology Co ltd
Priority date: 2022-12-28
Filing date: 2022-12-28
Publication date: 2023-06-06

Abstract

The invention provides a training method and device of a power prediction model and a method and device for adjusting power of a Long Term Evolution (LTE) base station. The training method comprises the following steps: acquiring historical characteristic data of an LTE base station; calculating target power corresponding to the historical characteristic data by adopting an energy efficiency reasonable rule algorithm; and training a power prediction model based on the historical feature data and the target power. The adjusting method comprises the following steps: acquiring characteristic data of the LTE base station according to a preset trigger period; inputting the characteristic data into a trained power prediction model to obtain predicted power; determining a power list to be adjusted by comparing the predicted power with the actual power of the LTE base station; and dynamically adjusting the power according to the power list to be adjusted. Therefore, accurate prediction of the LTE base station power can be efficiently realized, and the method is used for guiding adjustment of the actual power.

Description

Method and device for adjusting power of Long Term Evolution (LTE) base station

Technical Field

One or more embodiments of the present disclosure relate to the field of mobile communications technologies, and in particular, to a method and apparatus for training a power prediction model, and a method and apparatus for adjusting power of a long term evolution LTE base station.

Background

Currently, a long term evolution (Long Term Evolution, abbreviated as LTE) base station performs a radio transceiver station for transferring information with a mobile phone terminal through a mobile communication switching center in a certain radio coverage area. Can reach tens of times of Time Division synchronous code Division multiple access (TD-Synchronous Code Division Multiple Access, TD-SCDMA) technology, and make ubiquitous high-speed internet surfing possible.

However, when the service is idle, the physical resource block (Physical Resource Block, abbreviated PRB) of the LTE cell is low in utilization rate, and the number of connection users is small, in this case, the operator provides the base station transmitting power equivalent to that in busy hours, which is obviously wasted, so it is currently needed to provide a power adjustment method for monitoring the cell service in real time, and in the condition that the coverage perception of the user is unchanged when the service is idle, the base station transmitting power is reduced as much as possible, so as to realize the overall energy saving goal of the communication operator, thereby realizing sustainable development. In the prior art, the defects of more judging conditions, complex operation, numerical deviation problem and low batch operation efficiency exist.

Disclosure of Invention

The invention describes a training method and a device of a power prediction model, and a method and a device for adjusting the power of a Long Term Evolution (LTE) base station, which can solve the technical problems. The method comprises the steps of collecting and converging LTE base station power and related data, carrying out data analysis, screening out important characteristics, calculating out reasonable power values by using energy efficiency reasonable rules as target data, and transmitting the reasonable power values into an extreme gradient lifting XGBoost (EXtreme Gradient Boosting, abbreviated as XGBoost) algorithm to obtain a power prediction model for predicting the optimal power values of the LTE base station in batches, so that the energy efficiency of the base station is improved efficiently.

According to a first aspect, a method of training a power prediction model is provided. The method comprises the following steps:

acquiring historical characteristic data of a Long Term Evolution (LTE) base station; calculating target power corresponding to the historical characteristic data by adopting an energy efficiency reasonable rule algorithm; and training a power prediction model based on the historical feature data and the target power.

In some embodiments, obtaining historical feature data of the long term evolution, LTE, base station includes at least one of: the method comprises the steps of uplink physical resource module PRB utilization rate, maximum user number, average timing advance TA, switching success rate, number of uplink available physical resource modules PRB, average downlink cell throughput rate, uplink and downlink total throughput, average uplink channel quality indicator CQI, infinite resource control RRC connection establishment completion times, call establishment success rate, uplink interference average value, disconnection rate, average user number and wireless connection rate.

In some embodiments, the energy efficiency rationality rule algorithm includes the following calculation formula:

wherein W represents the target power, W _t Represents maximum power, W _m The minimum power of the coverage perception is represented, e represents a natural constant of 1.4, p represents the average value of the uplink utilization rate and the downlink utilization rate, N represents the number of existing users, N represents the maximum number of users, q represents a natural constant of 0.98, and w represents a buffer power value.

In some embodiments, training a power prediction model based on the historical feature data and a target power comprises:

determining the importance degree of each feature by carrying out correlation analysis on a plurality of features included in the historical feature data; and determining target characteristics with importance degrees within a preset range in the multiple characteristics. Inputting the target characteristics into the power prediction model to obtain predicted power; and updating model parameters of the power prediction model based on the predicted power and the target power.

In some more specific embodiments, determining the importance level of each feature of the historical feature data by performing a correlation analysis on a plurality of features included in the historical feature data includes:

preprocessing the historical characteristic data, wherein the preprocessing mode comprises at least one of the following steps: removing repeated data, undefined data and null data; and carrying out correlation analysis on the preprocessed multiple features, and determining the importance degree of each feature.

In some more specific embodiments, updating model parameters of the power prediction model based on the predicted power and the target power comprises:

based on the predicted power and the target power, updating model parameters of the power prediction model in a grid search mode.

In some embodiments, the power prediction model is implemented using an extreme gradient boost XGBoost algorithm.

According to a second aspect, a method for adjusting power of a long term evolution, LTE, base station is provided. The method comprises the following steps:

acquiring characteristic data of the LTE base station according to a preset trigger period; inputting the characteristic data into a power prediction model trained by the method provided by the first aspect to obtain predicted power; determining a power list to be adjusted by comparing the predicted power with the actual power of the LTE base station; and dynamically adjusting the power according to the power list to be adjusted.

According to a third aspect, there is provided a training apparatus for a power prediction model, the apparatus comprising:

the characteristic data acquisition module is configured to acquire historical characteristic data of the long-term evolution (LTE) base station; the target power calculation module is configured to calculate target power corresponding to the historical characteristic data by adopting an energy efficiency reasonable rule algorithm; and a training module configured to train a power prediction model based on the historical feature data and the target power.

In some embodiments, the training module comprises: a correlation analysis unit configured to determine the importance degrees of the features by performing correlation analysis on the plurality of features included in the history feature data; a target feature determining unit configured to determine a target feature of which importance degrees are within a preset range among the plurality of features; the power prediction unit is configured to input the target characteristics into the power prediction model to obtain predicted power; and a parameter updating unit configured to update model parameters of the power prediction model based on the predicted power and the target power.

In some more specific embodiments, the correlation analysis unit is specifically configured to: preprocessing the historical characteristic data, wherein the preprocessing mode comprises at least one of the following steps: removing repeated data, undefined data and null data; and carrying out correlation analysis on the preprocessed multiple features, and determining the importance degree of each feature.

In some embodiments, the training module is specifically configured to: based on the predicted power and the target power, updating model parameters of the power prediction model in a grid search mode.

In some embodiments, the historical feature data includes at least one of: the method comprises the steps of uplink physical resource module PRB utilization rate, maximum user number, average timing advance TA, switching success rate, number of uplink available physical resource modules PRB, average downlink cell throughput rate, uplink and downlink total throughput, average uplink channel quality indicator CQI, infinite resource control RRC connection establishment completion times, call establishment success rate, uplink interference average value, disconnection rate, average user number and wireless connection rate.

According to a fourth aspect, an embodiment of the present application further provides an apparatus for adjusting long term evolution LTE base station power, where the apparatus includes:

the characteristic data acquisition module is configured to acquire characteristic data of the LTE base station according to a preset trigger period; the power prediction module is configured to input the characteristic data into a power prediction model trained by the device provided by the third aspect to obtain predicted power; the power comparison module is configured to determine a power list to be adjusted by comparing the predicted power with the actual power of the LTE base station; and the power adjustment module is configured to dynamically adjust the power according to the power list to be adjusted.

In the method and the device provided by the embodiment of the specification, the power data is obtained based on the reasonable energy efficiency rule to serve as target data, important features are selected from the feature data such as the industrial parameters, the KPI and the like, and are transmitted into the XGBoost algorithm to train a power prediction model for predicting the power of the LTE base station and guiding the adjustment of the actual power of the LTE base station. The calculation efficiency and the accuracy of the power are effectively improved.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the description of the embodiments below are briefly introduced, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.

Fig. 1 is a schematic flow chart of a training method of a power prediction model according to an embodiment of the present disclosure;

fig. 2 is a flow chart illustrating a method for adjusting LTE base station power according to an embodiment of the present disclosure;

fig. 3 is a schematic structural diagram of a training device of a power prediction model according to an embodiment of the present disclosure;

fig. 4 is a schematic structural diagram of a device for adjusting LTE base station power according to an embodiment of the present disclosure.

Detailed Description

The following describes the scheme provided in the present specification with reference to the drawings.

In order to make the objects, technical solutions and advantages of the embodiments of the present application more clear, the technical solutions of the embodiments of the present application will be described below with reference to the accompanying drawings.

In the description of embodiments of the present application, words such as "exemplary," "such as" or "for example," are used to indicate by way of example, illustration, or description. Any embodiment or design described herein as "exemplary," "such as" or "for example" is not necessarily to be construed as preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary," "such as" or "for example," etc., is intended to present related concepts in a concrete fashion.

In the description of the embodiments of the present application, the term "and/or" is merely an association relationship describing an association object, and indicates that three relationships may exist, for example, a and/or B may indicate: a alone, B alone, and both A and B. In addition, unless otherwise indicated, the term "plurality" means two or more.

Furthermore, the terms "first," "second," and the like, are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating an indicated technical feature. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include one or more such feature. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless expressly specified otherwise.

Fig. 1 shows a flow chart of a training method of a power prediction model according to an embodiment of the present disclosure, where, as shown in fig. 1, the method includes the following steps:

step S110, historical characteristic data of a Long Term Evolution (LTE) base station is obtained.

In one embodiment, a plurality of samples may be obtained in this step, wherein each sample includes historical feature data of a corresponding LTE base station.

And step S120, calculating target power corresponding to the historical characteristic data by adopting an energy efficiency reasonable rule algorithm.

In some embodiments, the energy efficiency rationality rule algorithm of this step is calculated using the following formula (1):

In some embodiments, the historical characteristic data in each sample may be substituted into formula (1) to obtain the corresponding target power value. Thus, a plurality of target power values corresponding to the plurality of samples can be obtained.

Step S130, training a power prediction model based on the historical feature data and the target power.

Specifically, input features are determined based on historical feature data, and then the input features are processed by a power prediction model to obtain corresponding predicted power, so that model parameters in the power prediction model are updated based on the predicted power and the corresponding target power.

For the determination of the input features described above, in some embodiments, the historical feature data may be directly used as the input features of the power prediction model.

In other embodiments, important features may be selected from the historical feature data as inputs to the power prediction model. In some specific embodiments, determining the importance level of each feature by performing correlation analysis on a plurality of features included in the historical feature data; and determining a plurality of characteristics with importance degrees within a preset range as important characteristics. By way of example, the preset range may be ranked top 10%, or a degree of importance greater than 0.6. It is to be understood that several references herein refer to one or more.

In still other embodiments, the historical feature data may be preprocessed, and the feature data obtained after preprocessing may be used as the input feature. Illustratively, preprocessing may include removing duplicate data, removing undefined data, removing values outside the valid range, removing null data, filling in missing values.

It should be noted that the selection and pretreatment of the above important features may be alternatively implemented, or may be implemented in combination.

For the updating of the model parameters described above, in some embodiments, the model parameters of the power prediction model may be updated by means of a grid search. In other embodiments, the implementation of this step includes updating model parameters of the power prediction model using back propagation and gradient descent methods.

It is to be appreciated that the implementation of the power prediction model may be based on a tree model, such as a gradient boost decision tree (Gradient Boosting Decision Tree, GBDT) or an extreme gradient boost tree (eXtreme Gradient Boosting, XGBoost), and may also be based on a deep neural network (Deep Neural Networks, DNN).

The above is a description of the training method of the power prediction model. After the power prediction model is subjected to iterative training for multiple times by adopting the method shown in fig. 1, a trained power prediction model can be obtained. It should be understood that the training of the power prediction model may be performed by using a training set, and a verification set may be used for determining whether the power prediction model is trained, for example, if the accuracy of the trained power prediction model on the verification set reaches a preset threshold (e.g., 0.8), the model is considered to be trained, and the training is ended. In addition, the training set and the validation set are typically mutually exclusive, but where the structure of the samples is identical.

Further, the adjustment of the power of the LTE base station can be guided by using the trained power prediction model. Fig. 2 is a flow chart illustrating a method for adjusting LTE base station power according to an embodiment of the present disclosure. As shown in fig. 2, the method comprises the following steps:

step S210: and acquiring characteristic data of the LTE base station according to a preset trigger period.

In some embodiments, the predetermined trigger period is set manually based on expert experience, and the present trigger period is, for example, one hour.

It is to be understood that the feature data includes feature terms that are consistent with the feature terms of the training phase input power prediction model.

Step S220: and inputting the characteristic data into the trained power prediction model to obtain predicted power.

Step S230: and determining a power list to be adjusted by comparing the predicted power with the actual power of the LTE base station.

In some embodiments, a list of required power adjustments is obtained based on the difference between the base station actual power and the predicted power. It should be understood that no difference between the actual power and the predicted power of the base station will not enter the power list to be adjusted.

Step S240: and dynamically adjusting the power according to the power list to be adjusted.

In some embodiments, dynamic adjustment of power is performed with guaranteed coverage awareness unchanged.

In some specific embodiments, the coverage perception invariance means: under the condition that the antenna angle is unchanged and the power is reduced, the signal strength felt by the user in the coverage area is kept unchanged, and a power minimum value exists, so that the actual power is required to be higher than the minimum value.

The invention also provides a training device corresponding to the training method of the power prediction model. Fig. 3 is a schematic structural diagram of a training device for a power prediction model according to an embodiment of the present disclosure. As shown in fig. 3, the apparatus 300 includes:

the feature data acquisition module 310 is configured to acquire historical feature data of the long term evolution LTE base station.

The target power calculation module 320 is configured to calculate a target power corresponding to the historical feature data using an energy efficiency rationality rule algorithm.

The training module 330 is configured to train a power prediction model based on the historical feature data and the target power.

In some embodiments, training module 330 specifically includes: a correlation analysis unit configured to determine the importance degrees of the features by performing correlation analysis on the plurality of features included in the history feature data; a target feature determining unit configured to determine a target feature of which importance degrees are within a preset range among the plurality of features; the power prediction unit is configured to input the target characteristics into the power prediction model to obtain predicted power; and a parameter updating unit configured to update model parameters of the power prediction model based on the predicted power and the target power.

In some specific embodiments, the correlation analysis unit is specifically configured to: preprocessing the historical characteristic data, wherein the preprocessing mode comprises at least one of the following steps: removing repeated data, undefined data and null data; and carrying out correlation analysis on the preprocessed multiple features, and determining the importance degree of each feature.

In some embodiments, training module 330 is specifically configured to: based on the predicted power and the target power, updating model parameters of the power prediction model in a grid search mode.

In some embodiments, the energy efficiency rationality rule algorithm includes the above-described calculation formula (1).

It should be noted that for the description of the apparatus in fig. 3, reference may also be made to the description of the foregoing method.

The invention also provides a device for adjusting the power of the LTE base station. Fig. 4 is a schematic structural diagram of a device for adjusting LTE base station power according to an embodiment of the present disclosure. As shown in fig. 4, the apparatus 400 includes:

the feature data obtaining module 410 is configured to obtain feature data of the LTE base station according to a preset trigger period.

The power prediction module 420 is configured to input the feature data into a power prediction model trained by the device S300, to obtain a predicted power.

The power comparison module 430 is configured to determine the power list to be adjusted by comparing the predicted power with the actual power of the LTE base station.

The power adjustment module 440 is configured to dynamically adjust the power according to the power list to be adjusted.

It should be noted that for the description of the apparatus in fig. 4, reference may also be made to the description of the foregoing method.

According to an embodiment of another aspect, there is also provided a computer-readable storage medium having stored thereon a computer program which, when executed in a computer, causes the computer to perform the method described in connection with fig. 1 and 2.

According to an embodiment of yet another aspect, there is also provided a computing device including a memory having executable code stored therein and a processor that, when executing the executable code, implements the method described in connection with fig. 1 and 2. Those skilled in the art will appreciate that in one or more of the examples described above, the functions described in the present invention may be implemented in hardware, software, firmware, or any combination thereof. When implemented in software, these functions may be stored on or transmitted over as one or more instructions or code on a computer-readable medium.

The foregoing embodiments have been provided for the purpose of illustrating the general principles of the present invention in further detail, and are not to be construed as limiting the scope of the invention, but are merely intended to cover any modifications, equivalents, improvements, etc. based on the teachings of the invention.

Claims

1. A method of training a power prediction model, the method comprising:

acquiring historical characteristic data of a Long Term Evolution (LTE) base station;

calculating target power corresponding to the historical characteristic data by adopting an energy efficiency reasonable rule algorithm;

and training a power prediction model based on the historical feature data and the target power.

2. The method of claim 1, wherein the historical feature data comprises at least one of:

the method comprises the steps of uplink physical resource module PRB utilization rate, maximum user number, average timing advance TA, switching success rate, number of uplink available physical resource modules PRB, average downlink cell throughput rate, uplink and downlink total throughput, average uplink channel quality indicator CQI, infinite resource control RRC connection establishment completion times, call establishment success rate, uplink interference average value, disconnection rate, average user number and wireless connection rate.

3. The method of claim 1, wherein the energy efficiency rationality rule algorithm comprises the following calculation formula:

wherein W represents the target power, W _t Represents maximum power, W _m The minimum power of the coverage perception is represented, e represents a natural constant of 1.4, p represents the average value of the uplink utilization rate and the downlink utilization rate, N represents the existing user number, N represents the maximum user number, q represents a natural constant of 0.98, and w represents slowAnd (5) punching the power value.

4. The method of claim 1, wherein training a power prediction model based on the historical feature data and a target power comprises:

determining the importance degree of each feature by carrying out correlation analysis on a plurality of features included in the historical feature data;

determining target features with importance degrees within a preset range in the multiple features;

inputting the target characteristics into the power prediction model to obtain predicted power;

and updating model parameters of the power prediction model based on the predicted power and the target power.

5. The method of claim 4, wherein determining the importance of each feature by performing a correlation analysis on a plurality of features included in the historical feature data comprises:

preprocessing the historical characteristic data, wherein the preprocessing mode comprises at least one of the following steps: removing repeated data, undefined data and null data;

and carrying out correlation analysis on the preprocessed multiple features, and determining the importance degree of each feature.

6. The method of claim 4, wherein updating model parameters of the power prediction model based on the predicted power and a target power comprises:

7. The method of any of claims 1-6, wherein the power prediction model is implemented using an extreme gradient boost XGBoost algorithm.

8. A method for adjusting power of a long term evolution LTE base station, the method comprising:

acquiring characteristic data of the LTE base station according to a preset trigger period;

inputting the characteristic data into a power prediction model trained by the method of claim 1 to obtain predicted power;

determining a power list to be adjusted by comparing the predicted power with the actual power of the LTE base station;

and dynamically adjusting the power according to the power list to be adjusted.

9. A training device for a power prediction model, the device comprising:

the characteristic data acquisition module is configured to acquire historical characteristic data of the long-term evolution (LTE) base station;

the target power calculation module is configured to calculate target power corresponding to the historical characteristic data by adopting an energy efficiency reasonable rule algorithm;

and a training module configured to train a power prediction model based on the historical feature data and the target power.

10. An apparatus for adjusting power of a long term evolution LTE base station, the apparatus comprising:

the characteristic data acquisition module is configured to acquire characteristic data of the LTE base station according to a preset trigger period;

a power prediction module configured to input the characteristic data into a power prediction model trained by the device of claim 9 to obtain predicted power;

the power comparison module is configured to determine a power list to be adjusted by comparing the predicted power with the actual power of the LTE base station;

and the power adjustment module is configured to dynamically adjust the power according to the power list to be adjusted.