CN115243293A - Method and device for determining network optimization model, electronic equipment and storage medium - Google Patents
Method and device for determining network optimization model, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN115243293A CN115243293A CN202210865499.7A CN202210865499A CN115243293A CN 115243293 A CN115243293 A CN 115243293A CN 202210865499 A CN202210865499 A CN 202210865499A CN 115243293 A CN115243293 A CN 115243293A
- Authority
- CN
- China
- Prior art keywords
- base station
- network
- station cluster
- network optimization
- optimization model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/02—Arrangements for optimising operational condition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Computer Networks & Wireless Communication (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The embodiment of the application provides a method and a device for determining a network optimization model, electronic equipment and a storage medium, and relates to the field of network optimization and the field of federal learning. The method comprises the following steps: determining at least one base station cluster of the whole network; determining and sending network parameters and network indexes to each base station cluster server to indicate each base station cluster server to train a corresponding local network optimization model by taking the network parameters of each base station in a corresponding base station cluster as training samples and taking the network indexes as training labels; and receiving the local network optimization models sent by all the base station cluster servers, and performing federal learning according to all the local network optimization models to obtain a global network optimization model. The global network optimization model is obtained based on the federal learning of the base station cluster, and on the premise that large data interaction among base stations of the whole network is avoided and potential safety hazards of data are reduced, training samples required by network optimization modeling are enriched, and accuracy and generalization capability of the network optimization model are improved.
Description
Technical Field
The present application relates to the field of wireless network optimization and the field of federal learning, and in particular, to a method and an apparatus for determining a network optimization model, an electronic device, and a storage medium.
Background
In order to meet the increasing demands of huge mobile communication users on service quality and service diversity, the network standards of communication networks are increasing, the network scale is expanding, and the number of communication cells of a 4G network of a certain operator exceeds 200 ten thousand. The enormous network size presents a huge challenge to the wireless network optimization work. The network optimization mainly optimizes network indexes by reasonably adjusting network parameters, so that the network coverage effect and the user service experience are improved. Since the factors affecting network metrics are not constant, wireless communication network optimization should be continuous with changes in network parameters and environment. Most of the traditional wireless network optimization work is to find an optimization strategy through expert experience, and the optimization strategy of 'one station for one strategy and one time for one strategy' of a future communication network cannot be met.
With the rise of artificial intelligence, finding a wireless network optimization strategy by using a machine learning algorithm becomes a research hotspot, and the research ideas mainly include two types: 1. the method comprises the steps of optimizing a single cell based on machine learning, establishing an optimization model by taking network performance of the single cell under different network environments and network parameter configurations as training samples, and generating a single cell optimization scheme, wherein however, in order to ensure the stability of a wireless communication network, a base station in the single cell cannot frequently adjust wireless network parameters, so that the single cell cannot form enough input data samples, and the generated model is insufficient in accuracy and generalization capability; 2. the method comprises the steps of performing multi-cell optimization of machine learning, performing centralized learning to generate a model by converging multi-cell sample use cases, wherein in the process of generating the model by the multi-cell, additional transmission overhead and influence on data safety are brought by large-scale data interaction.
Disclosure of Invention
The embodiment of the application provides a method and a device for determining a network optimization model, electronic equipment, a computer-readable storage medium and a computer program product, which can solve the problems of additional transmission overhead and influence on data safety caused by large-scale data interaction in the process of generating a model by multiple cells. The technical scheme is as follows:
according to a first aspect of the embodiments of the present application, there is provided a method for determining a network optimization model, which is applied to a central server, and the method includes:
determining at least one base station cluster of the whole network; the base station in each base station cluster is connected with a corresponding base station cluster server;
determining and sending network parameters and network indexes to each base station cluster server to indicate each base station cluster server to train a corresponding local network optimization model by taking the network parameters of each base station in a corresponding base station cluster as training samples and taking the network indexes as training labels;
and receiving the local network optimization models sent by all the base station cluster servers, and performing federal learning according to all the local network optimization models to obtain a global network optimization model.
In one possible implementation, determining at least one base station cluster of the entire network includes:
collecting characteristic information of each base station in the whole network, wherein the characteristic information comprises engineering parameters, configuration information and user distribution characteristics;
and classifying the base stations of the whole network according to the characteristic information of each base station to obtain at least one base station cluster.
In a possible implementation manner, classifying base stations of a whole network according to feature information of each base station to obtain at least one base station cluster, including:
generating a characteristic vector corresponding to the characteristic information of each base station aiming at the characteristic information of each base station;
clustering base stations according to the characteristic vectors corresponding to all the base stations of the whole network to obtain at least one base station class;
and classifying the base stations in the base station class according to the transmission distance between each base station in the base station class and the base station cluster server and the data safety transmission rule to obtain at least one base station cluster.
According to a second aspect of the embodiments of the present application, there is provided a method for determining a network optimization model, which is applied to a central server, the method including:
receiving network parameters and network indexes sent by a central server;
acquiring network parameters and network indexes of each base station in a base station cluster corresponding to a base station cluster server; taking the network parameter of each base station as a training sample, and taking the corresponding network index as a corresponding training label;
and training the initial local network optimization model of the base station cluster according to all training samples and training labels in the corresponding base station cluster to obtain a local network optimization model.
In one possible implementation, the training method of the local network optimization model is as follows:
determining loss functions of network optimization models of all base stations in a base station cluster;
determining an average loss function according to the loss functions of the network optimization models of all the base stations and the number of the base stations in the base station cluster; each loss function is used for representing the difference between the training label and the prediction label;
taking the average loss function as a target loss function corresponding to a local network optimization model of the base station cluster;
and determining a loss value of the target loss function according to the training sample and the training label, if the loss value does not meet the training end condition, adjusting model parameters of the local network optimization model, and training the adjusted local network optimization model according to the training sample and the training label until the loss value meets the training end condition.
In a possible implementation manner, after obtaining the local network optimization model, the method further includes:
and sending the local network optimization model to the central server to instruct the central server to generate a global network model according to the local network optimization model.
In one possible implementation, after sending the local network optimization model to the central server, the method further includes:
receiving a global network model sent by a central server;
determining a network index to be optimized of a base station for any base station in a corresponding base station cluster;
and inputting the network index to be optimized into the global network model to obtain the network parameter to be optimized output by the global network model, sending the network parameter to be optimized to the base station, and instructing the base station to execute the network parameter to be optimized so as to optimize the corresponding network index.
In one possible implementation manner, after sending the network parameter to be optimized to the base station, the method further includes:
and receiving an actual network index which is obtained and sent after the base station executes the corresponding network parameter to be optimized, taking the network parameter to be optimized as a new training sample, and taking the actual network index as a new training label.
According to a third aspect of the embodiments of the present application, there is provided an apparatus for determining a network optimization model, applied to a central server, the apparatus including:
a base station cluster determining module, configured to determine at least one base station cluster of a whole network; the base station in each base station cluster is connected with a corresponding base station cluster server;
the first sending module is used for determining and sending the network parameters and the network indexes to each base station cluster server so as to indicate each base station cluster server to train a corresponding local network optimization model by taking the network parameters of each base station in a corresponding base station cluster as training samples and taking the network indexes as training labels;
and the global network optimization model obtaining module is used for receiving the local network optimization models sent by the base station cluster servers and performing federal learning according to all the local network optimization models to obtain the global network optimization model.
According to a fourth aspect of the embodiments of the present application, there is provided a device for determining a network optimization model, which is applied to a base station cluster server, and includes:
the first receiving module is used for receiving the network parameters and the network indexes sent by the central server;
the acquisition module is used for acquiring network parameters and network indexes of each base station in a base station cluster corresponding to the base station cluster server; taking the network parameter of each base station as a training sample, and taking the corresponding network index as a corresponding training label;
and the training module is used for training the initial local network optimization model of the base station cluster according to all training samples and training labels in the corresponding base station cluster to obtain a local network optimization model.
According to a fifth aspect of embodiments of the present application, there is provided an electronic device comprising a memory, a processor and a computer program stored on the memory, the processor implementing the steps of the method as provided by the first aspect when executing the program.
According to a sixth aspect of embodiments herein, there is provided a computer readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of the method as provided by the first aspect.
According to a seventh aspect of embodiments of the present application, there is provided a computer program product, which includes computer instructions stored in a computer-readable storage medium, and when a processor of a computer device reads the computer instructions from the computer-readable storage medium, the processor executes the computer instructions, so that the computer device executes the steps for implementing the method as provided in the first aspect.
The technical scheme provided by the embodiment of the application has the following beneficial effects:
the method comprises the steps that at least one base station cluster of the whole network is determined; the base station in each base station cluster is connected with a corresponding base station cluster server; determining and sending network parameters and network indexes to each base station cluster server to indicate each base station cluster server to train a corresponding local network optimization model by taking the network parameters of each base station in a corresponding base station cluster as training samples and taking the network indexes as training labels; and receiving the local network optimization models sent by all the base station cluster servers, and performing federal learning according to all the local network optimization models to obtain a global network optimization model. The global network optimization model is obtained based on federal learning of the base station cluster, and training samples required by network optimization modeling are enriched on the premise that large data volume interaction among base stations of the whole network is avoided and potential safety hazards of data are reduced, so that accuracy and generalization capability of the network optimization model are improved.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings used in the description of the embodiments of the present application will be briefly described below.
Fig. 1 is a schematic flowchart of a method for determining a network optimization model according to an embodiment of the present disclosure;
fig. 2 is a schematic diagram of a network optimization topology provided in an embodiment of the present application;
fig. 3 is a schematic flowchart of another method for determining a network optimization model according to an embodiment of the present application;
fig. 4 is a schematic structural diagram of a device for determining a network optimization model according to an embodiment of the present application;
fig. 5 is a schematic structural diagram of a determination apparatus of another network optimization model according to an embodiment of the present application;
fig. 6 is a schematic structural diagram of an electronic device according to an embodiment of the present application.
Detailed Description
Embodiments of the present application are described below in conjunction with the drawings in the present application. It should be understood that the embodiments set forth below in connection with the drawings are exemplary descriptions for explaining technical solutions of the embodiments of the present application, and do not limit the technical solutions of the embodiments of the present application.
As used herein, the singular forms "a", "an", "the" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should be further understood that the terms "comprises" and/or "comprising," when used in this specification in connection with embodiments of the present application, specify the presence of stated features, information, data, steps, operations, elements, and/or components, but do not preclude the presence or addition of other features, information, data, steps, operations, elements, components, and/or groups thereof, as embodied in the art. It will be understood that when an element is referred to as being "connected" or "coupled" to another element, it can be directly connected or coupled to the other element or intervening elements may be present. Further, "connected" or "coupled" as used herein may include wirelessly connected or wirelessly coupled. The term "and/or" as used herein indicates at least one of the items defined by the term, e.g., "a and/or B" can be implemented as "a", or as "B", or as "a and B".
To make the objects, technical solutions and advantages of the present application more clear, embodiments of the present application will be described in further detail below with reference to the accompanying drawings.
The terms referred to in this application will first be introduced and explained:
federal Learning (Federal Learning) is a new artificial intelligence basic technology, which is originally used for solving the problem of local model updating of android mobile phone terminal users, and the design goal of the federal Learning is to carry out efficient machine Learning among multiple participants or multiple computing nodes on the premise of guaranteeing information safety during big data exchange, protecting terminal data and personal data privacy and guaranteeing legal compliance. The technique can accomplish joint modeling without data sharing. Specifically, the own data of each data owner cannot leave the local, a global sharing model is jointly established through a parameter exchange mode under an encryption mechanism in a federal system (namely, under the condition of not violating data privacy regulations), and the established model only serves the local target in each region. Federal learning has the following four advantages:
(1) Data isolation is realized, so that data cannot be leaked to the outside, and the requirements of user privacy protection and data security are met;
(2) The quality of the model is ensured to be lossless, negative migration cannot occur, and the effect of the federal model is better than that of a split independent model;
(3) The participants are equal in status, so that fair cooperation is realized;
(4) And the encryption exchange of information and model parameters is carried out under the condition of ensuring that each participant keeps independence, and the growth is obtained simultaneously.
The present application provides a method, an apparatus, an electronic device, a computer-readable storage medium, and a computer program product for determining a network optimization model, which are intended to solve the above technical problems in the prior art.
The technical solutions of the embodiments of the present application and the technical effects produced by the technical solutions of the present application will be described below through descriptions of several exemplary embodiments. It should be noted that the following embodiments may be referred to, referred to or combined with each other, and the description of the same terms, similar features, similar implementation steps and the like in different embodiments is not repeated.
The embodiment of the present application provides a method for determining a network optimization model, which is applied to a central server, and as shown in fig. 1, the method includes:
step S101, determining at least one base station cluster of the whole network; and the base stations in each base station cluster are connected with the corresponding base station cluster server.
The execution subject of the embodiment of the application is a central server, and the determination method of the network optimization model is executed in the central server.
The base stations of the whole network are clustered by collecting the characteristic information of the base stations of the whole network in a certain geographic area (such as each province operator), generating the characteristic vectors corresponding to the characteristic information, obtaining at least one base station class, classifying each base station class, obtaining at least one base station cluster, and the detailed classification process is continued.
Each base station cluster comprises at least two base stations, the transmission distance between the at least two base stations meets preset conditions, safe transmission can be conducted between the at least two base stations, each base station cluster is provided with a corresponding base station cluster server, the base station clusters are controlled by the base station cluster servers, the base stations in each base station cluster can conduct data interaction with the corresponding base station cluster servers, and the base station cluster servers can conduct interaction with a central server.
As shown in fig. 2, which exemplarily shows a network optimization topology provided in the embodiment of the present application, base stations in a whole network are classified to obtain a plurality of base station clusters, each base station cluster includes at least two base stations and a corresponding base station cluster server, a base station in each base station cluster can perform data interaction with a corresponding base station cluster server, and a base station cluster server can represent that interaction is performed between a corresponding base station cluster server and a central server.
And step S102, determining and sending the network parameters and the network indexes to each base station cluster server to indicate each base station cluster server to train a corresponding local network optimization model by taking the network parameters of each base station in the corresponding base station cluster as training samples and taking the network indexes as training labels.
The central server in the embodiment of the application can control each base station cluster server, the central server can issue a plurality of network optimization topics, each optimization topic has corresponding network parameters and network indexes, the network optimization topics in the embodiment of the application can be optimization topics issued according to actual needs, for example, the network optimization topics can reduce interference, improve signal strength, improve transmission rate and the like, and the embodiment of the application does not limit the network optimization topics.
In the embodiment of the present application, the network parameter may be a configuration related to a network indicator or other parameter information, for example, if the network indicator is a network speed, the network parameter may be a configuration related to increasing the network speed.
In the embodiment of the application, the network parameter x is input of a network optimization model, the network index y is output of the network optimization model, the network optimization model is f, y = f (x), and the input x and the output y are used as the model characteristics of federal learning and are issued to the base station cluster server through the central server.
After receiving the network parameters and the network indexes issued by the central server, each base station cluster server acquires the network parameters and the network indexes of each base station in the base station cluster corresponding to the base station cluster server; taking the network parameter of each base station as a training sample, and taking the corresponding network index as a corresponding training label; and training the initial local network optimization model of the base station cluster according to all training samples and training labels in the corresponding base station cluster to obtain a local network optimization model.
It should be noted that, each base station cluster server trains the initial local network optimization model of the base station cluster based on the network parameters of each base station of the corresponding base station cluster as training samples and the network indexes as training labels, although the training samples and the training labels are the same, the obtained local network optimization models are not necessarily the same, and the model parameters of the local network optimization models are different.
The base station cluster server is used as a client in federal learning and uploads a local network optimization model f to a central server km Respective local network optimization models f km Model parameter ω of km May be different.
And step S103, receiving the local network optimization models sent by all the base station cluster servers, and performing federal learning according to all the local network optimization models to obtain a global network optimization model.
And after receiving the local network optimization models sent by each base station cluster server, the central server performs federated learning according to all the local network optimization models to obtain a global network optimization model.
Specifically, the central server receives a local network optimization model f of each base station cluster server km Then, determining the base station class to which the base station cluster server belongs, and optimizing the model f for each local network km Model parameter ω of (1) km Performing weighting aggregation operation to obtain global model parameter omega of global network optimization model k The global model parameter ω can be calculated from the following formula (1) k ：
Wherein, ω is k Is a global model parameter, M represents the mth base station cluster, M is more than 1 and less than M, M is a positive integer, M is the total amount of the base station cluster, S km For the number of base stations, S, in the corresponding base station cluster k Number of base stations, omega, of the whole network km Model parameters of the model are optimized for the respective local network.
The global network optimization model of the embodiment of the application is generated according to the local network optimization models, and the global model parameters of the global network optimization model are obtained by performing weighted aggregation on the model parameters of the local network optimization models, so that the joint network optimization of the network optimization model of the whole network base station is realized.
And after obtaining the global network optimization model, the central server sends the global network optimization model to each base station cluster server so as to instruct each base station cluster server to calculate the optimal optimization scheme of each base station based on the global network optimization model.
The method comprises the steps that at least one base station cluster of the whole network is determined; the base station in each base station cluster is connected with a corresponding base station cluster server; determining and sending network parameters and network indexes to each base station cluster server to indicate each base station cluster server to train a corresponding local network optimization model by taking the network parameters of each base station in a corresponding base station cluster as training samples and taking the network indexes as training labels; and receiving the local network optimization models sent by all the base station cluster servers, and performing federal learning according to all the local network optimization models to obtain a global network optimization model. The global network optimization model is obtained based on federal learning of the base station cluster, and training samples required by network optimization modeling are enriched on the premise that large data volume interaction among base stations of the whole network is avoided and potential safety hazards of data are reduced, so that accuracy and generalization capability of the network optimization model are improved.
The embodiment of the present application provides a possible implementation manner, and determining at least one base station cluster of a whole network includes:
collecting characteristic information of each base station in the whole network, wherein the characteristic information comprises engineering parameters, configuration information and user distribution characteristics;
and classifying the base stations of the whole network according to the characteristic information of each base station to obtain at least one base station cluster.
The method for determining at least one base station cluster in the whole network comprises the following steps:
firstly, collecting characteristic information of each base station of the whole network, wherein the characteristic information comprises engineering parameters, configuration information and user distribution characteristics.
The engineering parameters may be, for example, parameters such as a base station location, a default configuration (e.g., 4G, 5G), a frequency, etc., the configuration information may be functions already opened in the base station, the user distribution characteristics are generated according to a measurement report, the measurement report records a channel condition, a signal strength of each user, an interference condition, etc., and the distribution density of the users under different signal strengths may be determined according to the measurement report, that is, the user distribution characteristics are obtained.
After the characteristic information of each base station is obtained, the base stations of the whole network are classified according to the characteristic information of each base station to obtain at least one base station cluster, and the detailed process is shown in the subsequent part.
The embodiment of the present application provides a possible implementation manner, where classifying base stations in a whole network according to feature information of each base station to obtain at least one base station cluster, includes:
generating a characteristic vector corresponding to the characteristic information of each base station aiming at the characteristic information of each base station;
clustering base stations according to the characteristic vectors corresponding to all the base stations of the whole network to obtain at least one base station class;
and classifying the base stations in the base station class according to the transmission distance between each base station in the base station class and the base station cluster server and the data safety transmission rule to obtain at least one base station cluster.
After collecting the characteristic information of the base stations of the whole network, the embodiment of the application generates the characteristic vector corresponding to the characteristic information of the base stations, specifically, the engineering parametersDenoted as a, configuration information B, and user distribution characteristics c, each base station can be represented by a characteristic vector { a, B, c }, and assuming n base stations in total, the characteristic vector of base station t is B t ＝{a t ，b t ，c t }，1≤t≤n。
After obtaining the eigenvectors corresponding to all the base stations of the whole network, the embodiments of the present application cluster the base stations according to the eigenvectors of all the base stations to obtain at least one base station class, each base station class includes at least one base station, the clustering method includes, but is not limited to, algorithms such as K-means, DBSCAN, etc., assuming that K base station classes are obtained, which can be expressed as D 1 ...D i ...D k Wherein i represents the ith base station class, i is more than 1 and less than k, and i is a positive integer.
After obtaining each base station class, the embodiment of the application classifies the base stations in the base station class according to the transmission distance between each base station and the base station cluster server in the base station class and the data security transmission rule for each base station class to obtain at least one base station cluster, wherein each base station class can be divided into m base station clusters which are respectively D 11 ...D 1m ...D km 。
Specifically, if the transmission distance between the base station and the base station cluster server conforms to the preset transmission distance, and the data safety output rule between the base station and the base station cluster server is that safe data interaction can be performed between the base station and the base station cluster server, the base station can be divided into base station clusters corresponding to the base station cluster server.
The embodiment of the present application provides a method for determining a network optimization model, which is applied to a base station cluster server, and as shown in fig. 3, the method includes:
step S301, receiving the network parameters and the network indexes sent by the central server.
The base station cluster server can receive a network optimization topic sent by the central server, the network optimization topic comprises corresponding network parameters and network indexes, the network parameters x are input of a network optimization model, the network indexes y are output of the network optimization model, the network optimization model is f, y = f (x), the input x and the output y serve as model characteristics of federal learning, and the model characteristics are issued to the base station cluster server through the central server.
Step S302, collecting network parameters and network indexes of each base station in a base station cluster corresponding to a base station cluster server; and taking the network parameter of each base station as a training sample, and taking the corresponding network index as a corresponding training label.
After receiving the network parameters and the network indexes sent by the central server, the base station cluster server acquires the network parameters and the network indexes of each base station in the base station cluster corresponding to the base station cluster server, takes the network parameters of each base station as a training sample, and takes the corresponding network indexes as corresponding training labels.
Step S303, training the initial local network optimization model of the base station cluster according to all training samples and training labels in the corresponding base station cluster to obtain a local network optimization model.
According to the embodiment of the application, the initial local network optimization model of the base station cluster is trained according to all training samples and training labels in the corresponding base station cluster to obtain the local network optimization model, and a global network optimization model is generated for follow-up laying.
The embodiment of the application provides a possible implementation manner, and the training method of the local network optimization model comprises the following steps:
determining loss functions of network optimization models of all base stations in a base station cluster;
determining an average loss function according to the loss functions of the network optimization models of all the base stations and the number of the base stations in the base station cluster; each loss function is used for representing the difference between the training label and the prediction label;
taking the average loss function as a target loss function corresponding to a local network optimization model of the base station cluster;
and determining a loss value of the target loss function according to the training sample and the training label, if the loss value does not meet the training end condition, adjusting model parameters of the local network optimization model, and training the adjusted local network optimization model according to the training sample and the training label until the loss value meets the training end condition.
If the loss function of the network optimization model corresponding to each base station in the base station cluster server is l i (ω)，l i (ω) a loss function characterizing the network optimization model of the base station with parameter ω, with input x and output y.
According to the loss functions of the network optimization models of all the base stations and the number of the base stations in the base station cluster, the average loss function is determined and serves as the target loss function corresponding to the local network optimization model of the base station cluster.
Specifically, the average loss function can be determined by the following equation (2):
wherein L is km (ω) characterizing the mean loss function, i.e. the target loss function, D km Is any one base station cluster, i is any one base station in the any one base station cluster, l i (omega) a loss function characterizing a network optimization model of any one base station in the cluster of base stations, S km Characterizing the total number of base stations in the cluster, D km The sum of the loss functions generated by all base stations in the cluster is divided by the total number of base stations S km Is the average loss function of the local network optimization model for that cluster of base stations.
The method and the device for determining the loss value of the target loss function determine the loss value of the target loss function according to the training sample and the training label, adjust the model parameter of the local network optimization model if the loss value does not meet the training end condition, and train the adjusted local network optimization model according to the training sample and the training label until the loss value meets the training end condition
In addition, the parameter ω when solving the average loss function minimization of the local network optimization model km Namely, model parameters of the local network optimization model, the solving method includes, but is not limited to, algorithms such as large-batch Stochastic Gradient Descent (SGD).
The embodiment of the present application provides a possible implementation manner, and after obtaining the local network optimization model, the method further includes:
and sending the local network optimization model to the central server to instruct the central server to generate a global network model according to the local network optimization model.
The base station cluster server serves as a client side in federal learning, and after obtaining the local network optimization model, the base station cluster server sends the local network optimization model to the central server so as to instruct the central server to generate a global network model according to the local network optimization model.
The embodiment of the present application provides a possible implementation manner, and after sending the local network optimization model to the central server, the method further includes:
receiving a global network model sent by a central server;
for any base station in the corresponding base station cluster, determining a network index to be optimized of the base station;
and inputting the network index to be optimized into the global network model to obtain the network parameter to be optimized output by the global network model, sending the network parameter to be optimized to the base station, and instructing the base station to execute the network parameter to be optimized so as to optimize the corresponding network index.
After receiving the local network optimization models sent by each base station cluster server, the central server generates a global network model according to all the local network optimization models and sends the global network model to each base station cluster server.
After each base station cluster server receives each global network optimization model, according to the global model f k And network optimization conditions of all base stations in the cluster, calculating an optimal optimization scheme aiming at each base station, and issuing the optimal optimization scheme to the base stations.
The network optimization conditions of the base station represent network indexes which are intended to be optimized by the base station, namely network parameters to be optimized, the optimal optimization scheme represents optimal network parameters to be executed, the network indexes to be optimized can be input into the global network model to obtain the network parameters to be optimized output by the global network model, the network parameters to be optimized are sent to the base station, and the base station is instructed to execute the network parameters to be optimized so as to optimize the corresponding network indexes and meet the optimization strategy of 'one station, one strategy and one strategy' of a future communication network.
The embodiment of the present application provides a possible implementation manner, and after sending the network parameter to be optimized to the base station, the implementation manner further includes:
and receiving an actual network index which is obtained and sent after the base station executes the corresponding network parameter to be optimized, taking the network parameter to be optimized as a new training sample, and taking the actual network index as a new training label.
After the base station executes the corresponding network parameters to be optimized, the base station determines the actual network indexes obtained after the execution of the network parameters to be optimized is finished, and reports the network parameters to be optimized as new training samples and the actual network indexes as new training labels to the base station cluster server for further optimizing the model in the next round of training.
The embodiment of the present application provides a device 40 for determining a network optimization model, which is applied to a central server, and as shown in fig. 4, the device 40 may include:
a base station cluster determining module 410, configured to determine at least one base station cluster of the whole network; the base station in each base station cluster is connected with a corresponding base station cluster server;
a first sending module 420, configured to determine and send the network parameter and the network index to each base station cluster server, so as to instruct each base station cluster server to train a corresponding local network optimization model by using the network parameter of each base station in a corresponding base station cluster as a training sample and using the network index as a training label;
and a global network optimization model obtaining module 430, configured to receive the local network optimization models sent by each base station cluster server, and perform federal learning according to all the local network optimization models to obtain a global network optimization model.
The method comprises the steps that at least one base station cluster of the whole network is determined; the base station in each base station cluster is connected with a corresponding base station cluster server; determining and sending network parameters and network indexes to each base station cluster server to indicate each base station cluster server to train a corresponding local network optimization model by taking the network parameters of each base station in a corresponding base station cluster as training samples and taking the network indexes as training labels; and receiving the local network optimization models sent by all the base station cluster servers, and performing federal learning according to all the local network optimization models to obtain a global network optimization model. The global network optimization model is obtained based on federal learning of the base station cluster, and training samples required by network optimization modeling are enriched on the premise that large data volume interaction among base stations of the whole network is avoided and potential safety hazards of data are reduced, so that accuracy and generalization capability of the network optimization model are improved.
The embodiment of the present application provides a possible implementation manner, and the base station cluster determining module includes:
the characteristic information acquisition submodule is used for acquiring characteristic information of each base station in the whole network, and the characteristic information comprises engineering parameters, configuration information and user distribution characteristics;
and the base station cluster determining submodule is used for classifying the base stations of the whole network according to the characteristic information of each base station to obtain at least one base station cluster.
The embodiment of the present application provides a possible implementation manner, and the base station cluster determining submodule includes:
a feature vector generation unit configured to generate a feature vector corresponding to the feature information of each base station;
the base station class determining unit is used for clustering the base stations according to the characteristic vectors corresponding to all the base stations of the whole network to obtain at least one base station class;
and the base station cluster determining unit is used for classifying the base stations in the base station clusters according to the transmission distance between each base station in the base station clusters and the base station cluster server and the data safety transmission rule aiming at each base station cluster to obtain at least one base station cluster.
An embodiment of the present application provides a device 50 for determining a network optimization model, which is applied to a base station cluster server, as shown in fig. 5, where the device 50 may include:
a first receiving module 510, configured to receive the network parameter and the network indicator sent by the central server;
an acquisition module 520, configured to acquire network parameters and network indexes of each base station in a base station cluster corresponding to the base station cluster server; taking the network parameter of each base station as a training sample, and taking the corresponding network index as a corresponding training label;
the training module 530 is configured to train an initial local network optimization model of a base station cluster according to all training samples and training labels in the corresponding base station cluster, so as to obtain a local network optimization model.
The embodiment of the application provides a possible implementation mode, and the training method of the local network optimization model comprises the following steps:
the loss function determining submodule is used for determining the loss functions of the network optimization models of all the base stations in the base station cluster;
the average loss function determining submodule is used for determining an average loss function according to the loss functions of the network optimization models of all the base stations and the number of the base stations in the base station cluster; each loss function is used for representing the difference between the training label and the prediction label;
the target loss function determining submodule is used for taking the average loss function as a target loss function corresponding to the local network optimization model of the base station cluster;
and the training submodule is used for determining a loss value of the target loss function according to the training sample and the training label, adjusting model parameters of the local network optimization model if the loss value does not meet the training ending condition, and training the adjusted local network optimization model according to the training sample and the training label until the loss value meets the training ending condition.
The embodiment of the present application provides a possible implementation manner, and the apparatus further includes:
and the second sending module is used for sending the local network optimization model to the central server so as to instruct the central server to generate the global network model according to the local network optimization model.
In the embodiment of the present application, a possible implementation manner is provided, and the apparatus further includes:
the global network model receiving module is used for receiving the global network model sent by the central server;
the network index determining module to be optimized is used for determining the network index to be optimized of the base station for any base station in the corresponding base station cluster;
and the network parameter determining module to be optimized is used for inputting the network index to be optimized into the global network model to obtain the network parameter to be optimized output by the global network model, sending the network parameter to be optimized to the base station, and instructing the base station to execute the network parameter to be optimized so as to optimize the corresponding network index.
In the embodiment of the present application, a possible implementation manner is provided, and the apparatus further includes:
and the actual network index receiving module is used for receiving the actual network index which is obtained and sent after the base station executes the corresponding network parameter to be optimized, taking the network parameter to be optimized as a new training sample and taking the actual network index as a new training label.
The apparatus of the embodiment of the present application may execute the method provided by the embodiment of the present application, and the implementation principle is similar, the actions executed by the modules in the apparatus of the embodiments of the present application correspond to the steps in the method of the embodiments of the present application, and for the detailed functional description of the modules of the apparatus, reference may be specifically made to the description in the corresponding method shown in the foregoing, and details are not repeated here.
The embodiment of the present application provides an electronic device, which includes a memory, a processor, and a computer program stored on the memory, where the processor executes the computer program to implement the steps of the method for determining a network optimization model, and compared with the related art, the method can implement: the method comprises the steps that at least one base station cluster of the whole network is determined; the base station in each base station cluster is connected with a corresponding base station cluster server; determining and sending network parameters and network indexes to each base station cluster server to indicate each base station cluster server to train a corresponding local network optimization model by taking the network parameters of each base station in a corresponding base station cluster as training samples and taking the network indexes as training labels; and receiving the local network optimization models sent by all the base station cluster servers, and performing federal learning according to all the local network optimization models to obtain a global network optimization model. The global network optimization model is obtained based on the federal learning of the base station cluster, and on the premise that large data interaction among base stations of the whole network is avoided and potential safety hazards of data are reduced, training samples required by network optimization modeling are enriched, and accuracy and generalization capability of the network optimization model are improved.
In an alternative embodiment, an electronic device is provided, as shown in fig. 6, the electronic device 4000 shown in fig. 6 comprising: a processor 4001 and a memory 4003. Processor 4001 is coupled to memory 4003, such as via bus 4002. Optionally, the electronic device 4000 may further include a transceiver 4004, and the transceiver 4004 may be used for data interaction between the electronic device and other electronic devices, such as transmission of data and/or reception of data. In addition, the transceiver 4004 is not limited to one in practical applications, and the structure of the electronic device 4000 is not limited to the embodiment of the present application.
The Processor 4001 may be a CPU (Central Processing Unit), a general-purpose Processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other Programmable logic device, transistor logic device, hardware component, or any combination thereof. Which may implement or perform the various illustrative logical blocks, modules, and circuits described in connection with the disclosure. The processor 4001 may also be a combination that performs a computational function, including, for example, a combination of one or more microprocessors, a combination of a DSP and a microprocessor, or the like.
The Memory 4003 may be a ROM (Read Only Memory) or other type of static storage device that can store static information and instructions, a RAM (Random Access Memory) or other type of dynamic storage device that can store information and instructions, an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical Disc storage, optical Disc storage (including Compact Disc, laser Disc, optical Disc, digital versatile Disc, blu-ray Disc, etc.), a magnetic disk storage medium, other magnetic storage devices, or any other medium that can be used to carry or store a computer program and that can be Read by a computer, and is not limited herein.
The memory 4003 is used for storing computer programs for executing the embodiments of the present application, and is controlled by the processor 4001 to execute. The processor 4001 is used to execute computer programs stored in the memory 4003 to implement the steps shown in the foregoing method embodiments.
The electronic device package may include, but is not limited to, a mobile terminal such as a mobile phone, a notebook computer, a digital broadcast receiver, a PDA (personal digital assistant), a PAD (tablet computer), a PMP (portable multimedia player), a vehicle-mounted terminal (e.g., a car navigation terminal), etc., and a stationary terminal such as a digital TV, a desktop computer, etc., among others. The electronic device shown in fig. 6 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiments of the present disclosure.
Embodiments of the present application provide a computer-readable storage medium, on which a computer program is stored, and when being executed by a processor, the computer program may implement the steps and corresponding contents of the foregoing method embodiments. Compared with the prior art, the method can realize that: the method comprises the steps that at least one base station cluster of the whole network is determined; the base station in each base station cluster is connected with a corresponding base station cluster server; determining and sending network parameters and network indexes to each base station cluster server to indicate each base station cluster server to train a corresponding local network optimization model by taking the network parameters of each base station in a corresponding base station cluster as training samples and taking the network indexes as training labels; and receiving the local network optimization models sent by all the base station cluster servers, and performing federal learning according to all the local network optimization models to obtain a global network optimization model. The global network optimization model is obtained based on federal learning of the base station cluster, and training samples required by network optimization modeling are enriched on the premise that large data volume interaction among base stations of the whole network is avoided and potential safety hazards of data are reduced, so that accuracy and generalization capability of the network optimization model are improved.
It should be noted that the computer readable medium in the present disclosure can be a computer readable signal medium or a computer readable medium or any combination of the two. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples of the computer readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present disclosure, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In contrast, in the present disclosure, a computer readable signal medium may comprise a propagated data signal with computer readable program code embodied therein, either in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: electrical wires, optical cables, RF (radio frequency), etc., or any suitable combination of the foregoing.
Embodiments of the present application further provide a computer program product, which includes a computer program, and when the computer program is executed by a processor, the steps and corresponding contents of the foregoing method embodiments may be implemented. Compared with the prior art, the method can realize that:
the method comprises the steps that at least one base station cluster of the whole network is determined; the base station in each base station cluster is connected with a corresponding base station cluster server; determining and sending network parameters and network indexes to each base station cluster server to indicate each base station cluster server to train a corresponding local network optimization model by taking the network parameters of each base station in a corresponding base station cluster as training samples and taking the network indexes as training labels; and receiving the local network optimization models sent by all the base station cluster servers, and performing federal learning according to all the local network optimization models to obtain a global network optimization model. The global network optimization model is obtained based on federal learning of the base station cluster, and training samples required by network optimization modeling are enriched on the premise that large data volume interaction among base stations of the whole network is avoided and potential safety hazards of data are reduced, so that accuracy and generalization capability of the network optimization model are improved.
The terms "first," "second," "third," "fourth," "1," "2," and the like in the description and in the claims of the present application and in the above-described drawings (if any) are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It should be understood that the data so used are interchangeable under appropriate circumstances such that the embodiments of the application described herein are capable of operation in other sequences than illustrated or otherwise described herein.
It should be understood that, although each operation step is indicated by an arrow in the flowchart of the embodiment of the present application, the implementation order of the steps is not limited to the order indicated by the arrow. In some implementation scenarios of the embodiments of the present application, the implementation steps in the flowcharts may be performed in other sequences as needed, unless explicitly stated otherwise herein. In addition, some or all of the steps in each flowchart may include multiple sub-steps or multiple stages based on an actual implementation scenario. Some or all of these sub-steps or stages may be performed at the same time, or each of these sub-steps or stages may be performed at different times, respectively. In a scenario where execution times are different, an execution sequence of the sub-steps or the phases may be flexibly configured according to requirements, which is not limited in the embodiment of the present application.
The above are only optional embodiments of partial implementation scenarios in the present application, and it should be noted that, for those skilled in the art, other similar implementation means based on the technical idea of the present application are also within the scope of protection of the embodiments of the present application without departing from the technical idea of the present application.
Claims (12)
1. A method for determining a network optimization model is applied to a central server and comprises the following steps:
determining at least one base station cluster of the whole network; the base station in each base station cluster is connected with a corresponding base station cluster server;
determining and sending network parameters and network indexes to each base station cluster server to indicate each base station cluster server to train a corresponding local network optimization model by taking the network parameters of each base station in a corresponding base station cluster as training samples and taking the network indexes as training labels;
and receiving the local network optimization models sent by all the base station cluster servers, and performing federal learning according to all the local network optimization models to obtain a global network optimization model.
2. The method of claim 1, wherein the determining at least one base station cluster of the entire network comprises:
collecting characteristic information of each base station in the whole network, wherein the characteristic information comprises engineering parameters, configuration information and user distribution characteristics;
and classifying the base stations of the whole network according to the characteristic information of each base station to obtain at least one base station cluster.
3. The method of claim 2, wherein the classifying the base stations of the whole network according to the characteristic information of each base station to obtain at least one base station cluster comprises:
generating a characteristic vector corresponding to the characteristic information of each base station aiming at the characteristic information of each base station;
clustering base stations according to the characteristic vectors corresponding to all the base stations of the whole network to obtain at least one base station class;
and classifying the base stations in the base station classes according to the transmission distance between each base station in the base station classes and a base station cluster server and a data safety transmission rule to obtain at least one base station cluster.
4. A method for determining a network optimization model is applied to a base station cluster server, and comprises the following steps:
receiving network parameters and network indexes sent by a central server;
acquiring network parameters and network indexes of each base station in a base station cluster corresponding to the base station cluster server; taking the network parameter of each base station as a training sample, and taking the corresponding network index as a corresponding training label;
and training the initial local network optimization model of the base station cluster according to all training samples and training labels in the corresponding base station cluster to obtain a local network optimization model.
5. The method of claim 4, wherein the training method of the local network optimization model is as follows:
determining loss functions of network optimization models of all base stations in the base station cluster;
determining an average loss function according to the loss functions of the network optimization models of all the base stations and the number of the base stations in the base station cluster; each loss function is used for representing the difference between a training label and a prediction label;
taking the average loss function as a target loss function corresponding to a local network optimization model of the base station cluster;
and determining a loss value of the target loss function according to the training sample and the training label, if the loss value does not meet a training end condition, adjusting model parameters of the local network optimization model, and training the adjusted local network optimization model according to the training sample and the training label until the loss value meets the training end condition.
6. The method of claim 4, wherein after obtaining the local network optimization model, further comprising:
and sending the local network optimization model to a central server to instruct the central server to generate a global network model according to the local network optimization model.
7. The method of claim 6, wherein after sending the local network optimization model to a central server, further comprising:
receiving a global network model sent by a central server;
determining a network index to be optimized of any base station in a corresponding base station cluster;
inputting the network index to be optimized into the global network model to obtain the network parameter to be optimized output by the global network model, sending the network parameter to be optimized to the base station, and instructing the base station to execute the network parameter to be optimized so as to optimize the corresponding network index.
8. The method of claim 7, wherein after sending the network parameter to be optimized to the base station, further comprising:
and receiving an actual network index which is obtained and sent after the base station executes the corresponding network parameter to be optimized, taking the network parameter to be optimized as a new training sample, and taking the actual network index as a new training label.
9. An apparatus for determining a network optimization model, applied to a central server, includes:
a base station cluster determining module, configured to determine at least one base station cluster of a whole network; the base station in each base station cluster is connected with a corresponding base station cluster server;
the first sending module is used for determining and sending the network parameters and the network indexes to each base station cluster server so as to indicate each base station cluster server to train a corresponding local network optimization model by taking the network parameters of each base station in a corresponding base station cluster as training samples and taking the network indexes as training labels;
and the global network optimization model obtaining module is used for receiving the local network optimization models sent by the base station cluster servers and carrying out federal learning according to all the local network optimization models to obtain the global network optimization model.
10. An apparatus for determining a network optimization model, applied to a base station cluster server, includes:
the first receiving module is used for receiving the network parameters and the network indexes sent by the central server;
the acquisition module is used for acquiring network parameters and network indexes of each base station in the base station cluster corresponding to the base station cluster server; taking the network parameter of each base station as a training sample, and taking the corresponding network index as a corresponding training label;
and the training module is used for training the initial local network optimization model of the base station cluster according to all training samples and training labels in the corresponding base station cluster to obtain a local network optimization model.
11. An electronic device comprising a memory, a processor and a computer program stored on the memory, characterized in that the processor executes the computer program to implement the steps of the method of any of claims 1-8.
12. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method according to any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210865499.7A CN115243293A (en) | 2022-07-21 | 2022-07-21 | Method and device for determining network optimization model, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210865499.7A CN115243293A (en) | 2022-07-21 | 2022-07-21 | Method and device for determining network optimization model, electronic equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115243293A true CN115243293A (en) | 2022-10-25 |
Family
ID=83675867
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210865499.7A Pending CN115243293A (en) | 2022-07-21 | 2022-07-21 | Method and device for determining network optimization model, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115243293A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117724853A (en) * | 2024-02-08 | 2024-03-19 | 亚信科技（中国）有限公司 | Data processing method and device based on artificial intelligence |
-
2022
- 2022-07-21 CN CN202210865499.7A patent/CN115243293A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117724853A (en) * | 2024-02-08 | 2024-03-19 | 亚信科技（中国）有限公司 | Data processing method and device based on artificial intelligence |
| CN117724853B (en) * | 2024-02-08 | 2024-05-07 | 亚信科技（中国）有限公司 | Data processing method and device based on artificial intelligence |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111867049B (en) | Positioning method, positioning device and storage medium | |
| Xia et al. | Federated-learning-based client scheduling for low-latency wireless communications | |
| CN113435472A (en) | Vehicle-mounted computing power network user demand prediction method, system, device and medium | |
| US20240005210A1 (en) | Data protection method, apparatus, medium and device | |
| EP3723306A1 (en) | Frequency spectrum management device, system, method and computer readable storage medium | |
| CN111836321A (en) | Cell switching method based on federal learning and edge calculation | |
| CN112650841A (en) | Information processing method and device and electronic equipment | |
| Alnawayseh et al. | Smart congestion control in 5g/6g networks using hybrid deep learning techniques | |
| CN115243293A (en) | Method and device for determining network optimization model, electronic equipment and storage medium | |
| Li et al. | Dynamic spectrum access for internet-of-things based on federated deep reinforcement learning | |
| Agarwal et al. | Deep learning techniques to improve radio resource management in vehicular communication network | |
| CN115802398A (en) | Interference optimization method and device, storage medium and electronic equipment | |
| CN113590857A (en) | Key value matching method and device, readable medium and electronic equipment | |
| Yang et al. | Blind LTE-U/WiFi coexistence system using convolutional neural network | |
| CN117678172A (en) | Channel data generation method, device, equipment and storage medium | |
| CN116150612A (en) | Model training method and communication device | |
| Farhaoui et al. | Big Data and Smart Digital Environment | |
| Pandey et al. | Load distribution in the typical and zero cells in a PLP-PPP vehicular communication network | |
| CN115834580B (en) | Ocean big data-oriented distributed data processing method, device and equipment | |
| Wang et al. | Application of improved Naive Bayes classification algorithm in 5G signaling analysis | |
| CN117279019B (en) | Method and device for predicting spectrum efficiency, electronic equipment and storage medium | |
| WO2024130519A1 (en) | Artificial intelligence (ai) service scheduling method, and apparatus | |
| Li et al. | Device-to-device relay selection based on comprehensive social attributes in heterogeneous wireless networks | |
| CN114727298B (en) | Sharing method and system of wireless network without honeycomb unit, medium and electronic equipment | |
| Zaki et al. | Wireless Communication Channel Scenarios: Machine-Learning-Based Identification and Performance Enhancement. Electronics 2022, 11, 3253 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |