CN117560719A

CN117560719A - Load balancing method, base station and OAM

Info

Publication number: CN117560719A
Application number: CN202210922730.1A
Authority: CN
Inventors: 张化; 许森; 熊尚坤; 信金灿
Original assignee: China Telecom Corp Ltd
Current assignee: China Telecom Corp Ltd
Priority date: 2022-08-02
Filing date: 2022-08-02
Publication date: 2024-02-13

Abstract

The disclosure provides a load balancing method, a base station and OAM, and belongs to the technical field of mobile communication. The method comprises the following steps: receiving a third deployment message sent by the OAM and comprising second configuration information of the model; sending a request message of model reasoning data to a second base station; receiving a reply message of the second base station; the request message comprises a requested measurement object, an IP address and a port address of the first base station; the reply message comprises a measurement object, an IP address and a port address of the second base station, which can be provided; the request message and the reply message are used for establishing a transmission channel of model reasoning data between the first base station and the second base station; receiving a report message comprising model reasoning data sent by a second base station; model reasoning is carried out based on the model updated by the second configuration information, model reasoning data and measurement report data of the UE, and model reasoning output for load balancing is determined; and sending the model reasoning output to the second base station, and carrying out load balancing operation on the UE in the coverage area based on the model reasoning output.

Description

Load balancing method, base station and OAM

Technical Field

The disclosure belongs to the technical field of wireless communication, and in particular relates to a load balancing method, a base station and an OAM (Operation Administration and Maintenance, network management equipment).

Background

With the development of wireless communication technology, network traffic is increased, available frequency bands are increased, service quality of a base station can be improved through network load balancing, and network performance is improved.

Traditional network load balancing decisions mainly depend on reporting of current or past cell load states, and then whether operations such as switching are needed or not is determined based on reporting conditions.

However, for the scenario of high mobility and large connection, as network traffic increases network load, resource status changes are accelerated, and the conventional manner cannot perform load balancing timely, so that even if load balancing is performed, the problems of ping-pong handover between different cells, cell overload and user service quality degradation still occur easily, and thus it is difficult to ensure the overall network service quality.

Disclosure of Invention

The embodiment of the disclosure aims to provide a load balancing method, a base station and OAM, which can solve the problem that the traditional load balancing is difficult to ensure the overall network service quality.

In order to solve the above technical problems, the present disclosure is implemented as follows:

in a first aspect, an embodiment of the present disclosure provides a load balancing method, applied to a first base station, where the first base station is an anchor base station, the method includes: receiving a third deployment message sent by the OAM, wherein the third deployment message comprises second configuration information of the model; sending a request message of model reasoning data to a second base station, and receiving a reply message of the second base station; the request message comprises a requested measurement object, an IP address and a port address of the first base station; the reply message comprises a measurement object, an IP address and a port address of the second base station, which can be provided; the request message and the reply message are used for establishing a transmission channel of model reasoning data between the first base station and the second base station, and the second base station is an auxiliary base station of the first base station; receiving a report message comprising model reasoning data sent by a second base station; model reasoning is carried out based on the model updated by the second configuration information, model reasoning data and measurement report data of the UE, and model reasoning output for load balancing is determined; and sending the model reasoning output to the second base station, and carrying out load balancing operation on the UE in the coverage area based on the model reasoning output.

In a second aspect, an embodiment of the present disclosure provides a base station, where the base station is a first base station, and the first base station is an anchor base station, including: the system comprises a receiving module, a sending module, a model reasoning module and a load balancing module; the receiving module is used for receiving a third deployment message sent by the OAM, wherein the third deployment message comprises second configuration information of the model; the sending module is also used for sending a request message of model reasoning data to the second base station; the receiving module is also used for receiving a reply message of the second base station; the request message comprises a requested measurement object, an IP address and a port address of the first base station; the reply message comprises a measurement object, an IP address and a port address of the second base station, which can be provided; the request message and the reply message are used for establishing a transmission channel of model reasoning data between the first base station and the second base station, and the second base station is an auxiliary base station of the first base station; the receiving module is also used for receiving a report message which is sent by the second base station and comprises model reasoning data; the model reasoning module is used for carrying out model reasoning based on the model updated by the second configuration information, model reasoning data and measurement report data of the UE, and determining model reasoning output for load balancing; the sending module is also used for sending model reasoning output to the second base station; and the load balancing module is used for carrying out load balancing operation on the UE in the coverage area based on model reasoning output.

In a third aspect, an embodiment of the present disclosure provides a load balancing method, applied to a second base station, where the method includes: receiving a third deployment message sent by the OAM, wherein the third deployment message comprises second configuration information of the model; receiving a request message of model reasoning data sent by a first base station, and sending a reply message to the first base station; the first base station is an anchor base station, the second base station is an auxiliary base station of the first base station, and the request message comprises a requested measurement object, an IP address and a port address of the first base station; the reply message comprises a measuring object, an IP address and a port address of the second base station, wherein the measuring object, the IP address and the port address of the second base station can be provided, and the request message and the reply message are used for establishing a transmission channel of model reasoning data between the first base station and the second base station; sending a report message comprising model reasoning data to a first base station; and receiving model reasoning output sent by the first base station, and carrying out load balancing operation on the UE in the coverage area based on the model reasoning output.

In a fourth aspect, an embodiment of the present disclosure provides a second base station, including: the device comprises a receiving module, a sending module and a load balancing module; the receiving module is used for receiving a third deployment message sent by the OAM, wherein the third deployment message comprises second configuration information of the model; the receiving module is also used for receiving a request message of model reasoning data sent by the first base station; the sending module is also used for sending a reply message to the first base station; the first base station is an anchor base station, the second base station is an auxiliary base station of the first base station, and the request message comprises a requested measurement object, an IP address and a port address of the first base station; the reply message comprises a measuring object, an IP address and a port address of the second base station, wherein the measuring object, the IP address and the port address of the second base station can be provided, and the request message and the reply message are used for establishing a transmission channel of model reasoning data between the first base station and the second base station; the sending module is also used for sending a report message comprising the model reasoning data to the first base station; the receiving module is also used for receiving the model reasoning output sent by the first base station; and the load balancing module is used for carrying out load balancing operation on the UE in the coverage area based on model reasoning output.

In a fifth aspect, an embodiment of the present disclosure provides a load balancing method, applied to OAM, where the method includes: determining that the first base station is an anchor base station and the second base station is an auxiliary base station of the first base station; and sending a third deployment message to the first base station and the second base station so that the first base station and the second base station perform model reasoning output based on the model updated by the second configuration information and perform load balancing based on the model reasoning output, wherein the third deployment message comprises the second configuration information.

In a sixth aspect, embodiments of the present disclosure provide an OAM including: a determining module and a transmitting module; the determining module is used for determining that the first base station is an anchor base station and the second base station is an auxiliary base station of the first base station; the sending module is further configured to send a third deployment message to the first base station and the second base station, so that the first base station and the second base station perform model reasoning output based on the model updated by the second configuration information, and perform load balancing based on the model reasoning output, where the third deployment message includes the second configuration information.

In a seventh aspect, embodiments of the present disclosure provide a base station comprising a processor, a memory and a program or instruction stored on the memory and executable on the processor, the program or instruction when executed by the processor implementing the steps of the load balancing method according to the first or third aspect.

In an eighth aspect, an embodiment of the present disclosure provides an OAM including a processor, a memory, and a program or instruction stored on the memory and executable on the processor, the program or instruction implementing the steps of the load balancing method according to the fifth aspect when executed by the processor.

In a ninth aspect, embodiments of the present disclosure provide a readable storage medium having stored thereon a program or instructions which, when executed by a processor, implement the steps of the load balancing method according to the first, third or fifth aspects.

In a tenth aspect, embodiments of the present disclosure provide a chip, where the chip includes a processor and a communication interface, where the communication interface is coupled to the processor, and where the processor is configured to execute a program or instructions to implement the load balancing method according to the first aspect, the third aspect, or the fifth aspect.

In an eleventh aspect, embodiments of the present disclosure provide a computer program product comprising instructions which, when run on a computer, cause the computer to perform the steps of the load balancing method according to the first, third or fifth aspects.

In the embodiment of the disclosure, the OAM configures the trained model update to the base station, that is, transmits second configuration information of the model to the first base station and the second base station, so that the first base station and the second base station update the model. The first base station collects measurement data from the UE and the base station and conducts model reasoning, prediction information including network load, terminal track prediction and the like, decision information about terminal switching and the like are output, so that the network is triggered to execute corresponding operations, for example, the first base station conducts load balancing operation on the UE in the coverage area of the first base station, and the second base station conducts load balancing operation on the UE in the coverage area of the second base station, and therefore performance of the network can be optimized. According to the scheme, the signaling flow design is carried out aiming at the processes of data collection, model deployment, model reasoning and the like among network nodes in a load balancing scene, and under the condition that the model is an AI/ML related model, the network can predict information such as user service, track and the like through an AI/ML technology so as to adopt a corresponding load balancing strategy, so that the problems of overlarge air interface signaling overhead and the like caused by network management and optimization in a traditional manual control mode are solved, and the user service experience can be improved while the network performance is ensured more flexibly and intelligently.

Drawings

Fig. 1 is a schematic diagram of a wireless network intelligent network function architecture according to an embodiment of the disclosure;

fig. 2 is a schematic diagram of a load balancing system provided by an embodiment of the present disclosure;

fig. 3 is one of the interactive flow diagrams of the load balancing method provided in the embodiment of the present disclosure;

FIG. 4 is a second schematic diagram of an interaction flow of a load balancing method according to an embodiment of the disclosure;

fig. 5 is a schematic diagram of one possible structure of a base station according to an embodiment of the disclosure;

fig. 6 is a second possible structural diagram of a base station according to an embodiment of the disclosure;

fig. 7 is a schematic diagram of one possible structure of OAM according to an embodiment of the present disclosure;

fig. 8 is a third possible structural diagram of a seed base station according to an embodiment of the disclosure;

fig. 9 is a second schematic diagram of a possible OAM structure according to an embodiment of the present disclosure;

fig. 10 is a hardware schematic of a network entity according to an embodiment of the disclosure.

Detailed Description

The following description of the technical solutions in the embodiments of the present disclosure will be made clearly and completely with reference to the accompanying drawings in the embodiments of the present disclosure, and it is apparent that the described embodiments are some embodiments of the present disclosure, but not all embodiments. All other embodiments, which can be made by one of ordinary skill in the art without inventive effort, based on the embodiments in this disclosure are intended to be within the scope of this disclosure.

The terms first, second and the like in the description and in the claims, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged, where appropriate, such that embodiments of the disclosure may be practiced in sequences other than those illustrated and described herein, and that the objects identified by "first," "second," etc. are generally of the same type and are not limited to the number of objects, e.g., the first object may be one or more. Furthermore, in the description and claims, "and/or" means at least one of the connected objects, and the character "/", generally means that the associated object is an "or" relationship.

It is noted that the techniques described in embodiments of the present disclosure are not limited to LTE (Long Term Evolution )/LTE-a (LTE-Advanced, evolution of LTE) systems, but may also be used in other wireless communication systems, such as CDMA (Code Division Multiple Access ), TDMA (Time Division Multiple Access, time division multiple access), FDMA (Frequency Division Multiple Access ), OFDMA (Orthogonal Frequency Division Multiple Access, orthogonal frequency division multiple access), SC-FDMA (Single-carrier Frequency-Division Multiple Access, single carrier frequency division multiple access), and other systems. The terms "system" and "network" in embodiments of the present application are often used interchangeably, and the techniques described may be used for both the above-mentioned systems and radio technologies, as well as other systems and radio technologies. However, the following description describes an NR system for purposes of example and NR terminology is used in much of the following description, although the techniques may also be applied to applications other than NR system applications, such as 6G (6 th Generation) communication systems.

The load balancing method provided by the embodiment of the disclosure is described in detail below through specific embodiments and application scenarios thereof with reference to the accompanying drawings.

Fig. 1 is a schematic diagram of a wireless network intelligent network function architecture according to an embodiment of the disclosure, as shown in fig. 1, including: a data collection function 101, a model training function 102, a model reasoning function 103, and an execution function 104; wherein the data collection function 101 provides input data to the model training function 102 and the model reasoning function 103; the incoming data contains measurements, feedback from the terminal or different network entities. The model training function 102 performs AI/ML model training, validation and testing functions and is capable of generating model performance metrics as part of a model testing process. The model training function 102 is also responsible for data preprocessing, cleaning, formatting, and conversion based on training data provided by the data collection function. The model inference function 103 provides AI/ML model inference output based on the trained model, which contains prediction and decision information for each node of the network to judge and execute. The model inference function 103 also has data processing capability. The executive function 104 receives the output of the model inference function 103 and triggers or performs a corresponding action.

Fig. 2 is a schematic diagram of a load balancing system according to an embodiment of the disclosure. As shown in fig. 2, the load balancing system includes: OAM 200, base station 201, base station 202, UE accessing base station 201 and UE accessing base station 202. The OAM 200 is configured to trigger a load balancing mechanism, determine that the base station 201 is an anchor base station, and determine that at least one base station 202 within a load balancing area is an auxiliary base station; the OAM 200 learns to determine first configuration information of a model through offline training, sends anchor base station indication information and the first configuration information to the base station 201 through a first deployment message, and sends the first configuration information to at least one base station 202 through a second deployment message, so that the base station 201 and the at least one base station 202 acquire training data required by the online training of the model; the base station 201 and at least one base station 202 respectively determine and send training data required by model online training to the OAM 200 through measurement data reported by UE in respective coverage areas and data measured by the base station; after receiving the training data reported by the base stations in the load balancing area, the OAM 200 performs online training based on the training data, so as to obtain an updated model, obtain second configuration information of the model, send the second configuration information to the base stations in the load balancing area, and after receiving the configuration information, the first base station 201 initiates a request message of model reasoning data to at least one second base station 202, and after receiving the request message, each second base station 202 sends a reply message or a failure message to the first base station 201 according to the self-capability, and the base station 202 sending the reply message sends a report message of the model reasoning data to the first base station 201, so that the first base station 201 performs model reasoning; after the first base station 201 obtains the model inference output, load balancing can be performed on the UEs within the coverage area of the first base station based on the model inference output; the first base station 201 sends a model inference output to the at least one second base station 202, and each second base station 202 performs load balancing on UEs within coverage of each second base station based on the model inference output.

Fig. 3 is an interactive flow chart of a load balancing method according to an embodiment of the present disclosure, as shown in fig. 3, the method includes S301 and S315 described below. In the embodiment of the disclosure, the network management device may directly send the trained model to the base station, so that the cooperative information between the base stations performs model reasoning, and performs load balancing based on model reasoning output; the network management device may also send configuration information for online training of the model to the first base station, instruct the base stations to cooperatively obtain model training input data, and perform online training according to the model training input data obtained by the base stations, so as to obtain a more accurate model, that is, the following S301 to S308 may be executed first, and then the network management device may send the configuration information of the trained model to the base station, so that the base station may perform reasoning and load balancing based on the updated model.

S301, the OAM determines that a first base station is an anchor base station and a second base station is an auxiliary base station of the first base station, and the first configuration information of the model is obtained based on historical data and a pre-configured load balancing model training model.

Illustratively, the OAM may determine the load balancing area range according to at least one of measurement report information and geographical information; according to the calculation power and the storage capacity of the base stations in the load balancing area, determining the first base station with the model reasoning function as an anchor base station, and determining the base stations nearby the first base station as at least one of reasoning or training of the auxiliary base station participation model of the first base station. The anchor base station is used for collecting data required by load balancing model training or reasoning from the adjacent base stations.

Specifically, after the OAM determines the load balancing area, the OAM may further acquire the wireless measurement related history data reported by the OAM-stored load balancing area base station; dividing the wireless measurement related historical data into a training data set, a verification data set and a test data set; and training the model based on the data set, and determining the learning algorithm, parameters, required characteristic input and other information of the related model, thereby obtaining the first configuration information of the model.

The historical data related to the wireless measurement is the data related to the wireless measurement reported by the base station stored in the OAM.

Illustratively, the model referred to in the embodiments of the present disclosure may be an AI (Artificial Intelligence, machine learning) model or an ML (Machine learning artificial intelligence) model.

Optionally, in the load balancing method provided in the embodiment of the present disclosure, the first configuration information includes at least one of the following 1-1 to 1-4:

1-1: a list of secondary base station identities;

the auxiliary base station identification list indicates effective auxiliary base stations in the load balancing area range. The auxiliary base station is used for providing measurement report data of the characteristic input information for the anchor base station.

Illustratively, the auxiliary base station identification list includes one or more base station identifications, and the base station identifications indicate effective auxiliary base stations in the load balancing area range to acquire data required by model training and reasoning so as to facilitate operations such as model updating on subsequent relevant base stations.

1-2: indexing a model;

the model index indicates the applicable use cases of the models to be deployed and the algorithms used by the models.

Optionally, the model index indicates at least one of the following a1 to a3:

a1: model use case

The model use cases indicate the model use cases required for realizing the target use cases.

For example, the target use case includes at least one of: load prediction, UE trajectory prediction, UE mobility prediction, and UE traffic prediction. The indication of the model use case may be of an enumeration type.

In the embodiment of the disclosure, the model use case can be used for improving the accuracy of the target use case or reducing the computational complexity to save network resources.

a2: model class

The model class indicates an adaptive model of the target use case, and the indication of the model class may be an enumeration type.

For example, the model class may include at least one of: linear regression, logistic regression, decision trees, SVM (support vector machines, support vector machine), random forest, etc.

a3: model parameters

The model parameters are configuration variables inside the model, and the configuration variables comprise at least one of the following: weight, bias, learning rate, and number of iterations.

It should be noted that, in the embodiment of the present disclosure, the model parameters may be carried in a model deployment message, or may be carried in a model update message. The model parameters can be used to update parameters of the model in the base station, thereby improving accuracy and confidence of model training.

1-3: a first list of feature input information;

wherein the first list indicates at least one characteristic input information collected by the receiving base station for model training.

Optionally, the first list comprises at least one of: the characteristic input information of the UE and the characteristic input information of the base station.

Specifically, the feature input information of the UE includes at least one of the following b1 to b4:

b1: UE location information;

b2: UE history mobility information;

b3: UE movement speed;

b4: the UE wirelessly measures information.

Illustratively, the UE wireless measurement information includes at least one of: RSRP (Reference Signal Receiving Power, reference signal received power), RSRQ (Reference Signal Receiving Quality, reference signal received quality), and SINR (Signal to Interference plus Noise Ratio, signal-to-interference plus noise ratio).

Specifically, the characteristic input information of the base station includes at least one of the following c1 to c 3:

c1: current UE traffic sum;

it will be appreciated that the current UE traffic sum is the sum of all UEs currently measured to access the base station.

c2: the current base station wireless measurement information;

c3: predicted base station radio measurement information.

Illustratively, the base station wireless measurement information includes at least one of: PRB (Physical Resource Block ) utilization of a cell comprised by a base station, average RRC (Radio Resource Control ) connection number of a cell comprised by a base station, and packet loss rate of a cell comprised by a base station.

1-4: registration request information element of feature input information.

If the registration request information element of the feature input information indicates to start, the receiving base station starts measurement according to the indication in the list of the feature input information; if the registration request information element of the feature input information indicates stopping, the receiving base station stops measuring and reporting; if the registration request information element of the feature input information indicates addition, the receiving base station adds the measured value or the preset value indicated by the registration request information element to the measurement started by the list indication of the feature input information; wherein the receiving base station may ignore the "add" registration request information element if the receiving base station has initiated the measurement indicated by the feature input information addition indication information.

S302, OAM sends a first deployment message to the first base station and sends a second deployment message to the second base station.

The first deployment message comprises anchor base station indication information and first configuration information, the second deployment message comprises the first configuration information, and the second base station is an auxiliary base station of the first base station.

The anchor base station indication information indicates that the receiving base station is an anchor base station, the anchor base station is used for collecting statistical information of surrounding neighbor base stations, and the anchor base station indication information can be of a boolean type or an enumeration type.

S303, the first base station determines that the first base station is an anchor base station based on the first deployment message sent by the OAM.

S304, the first base station measures and transmits UE measurement configuration to the UE in the coverage area based on the first configuration information, and training data required by the training model is determined according to measurement report data of the UE and measurement statistical results of the first base station.

S305, the second base station measures and transmits UE measurement configuration to the UE in the coverage area based on the second deployment message transmitted by the OAM, and determines training data required by the training model according to measurement report data of the UE and measurement statistical results of the second base station.

Wherein the UE measurement configuration described above includes at least one of the following 2-1 to 2-3:

2-1: UE history mobility information;

2-2: RRM (Radio Resource Management ) measurement configuration;

wherein the RRM measurement configuration comprises at least one of: triggering information of periodic measurement and UE wireless measurement information; the UE wireless measurement information includes at least one of: RSRP, RSRQ, and SNIR.

2-3: MDT (Minimization of Drive-tests, minimization of drive tests) measurement configuration.

Wherein the MDT measurement configuration comprises at least one of: trigger information for periodic measurements, UE location information, and UE movement speed.

The triggering information of the periodic measurement comprises the triggering period and the recorded period.

It is understood that after receiving the UE measurement configuration sent by the first base station, the UE in the coverage area of the first base station may perform corresponding measurement based on the indicated measurement configuration. After receiving the UE measurement configuration sent by the second base station, the UE in the coverage area of the second base station may perform corresponding measurement based on the indicated measurement configuration. For example, measurement values such as UE location information, UE RSRP, UE RSRQ, SINR and the like are measured, and corresponding measurement data is collected.

If the MDT measurement configuration is indicated in the UE measurement configuration, the UE can collect and record corresponding measurement quantities according to the MDT measurement configuration.

Optionally, in an embodiment of the present disclosure, the measurement report data of the UE may include measurement report data corresponding to at least one of: UE historical mobility information, RRM measurement configuration, and MDT measurement configuration.

The measurement report data corresponding to the RRM measurement configuration comprises at least one of the following: UE radio measurement information and time stamp.

The measurement report data corresponding to the MDT measurement configuration comprises at least one of the following: at least one of UE location information and UE movement speed, and a time stamp.

S306, the first base station sends a model training input message including training data to the OAM.

S307, the second base station sends a model training input message including training data to the OAM.

Wherein the model training input message comprises: the time stamp and the feature input information indicated in the first configuration information.

It can be understood that if the receiving base station can provide the feature input information indicated in the first configuration information, the receiving base station starts measurement and performs feedback according to the corresponding request in the first configuration information. And then the receiving base station receives and processes the measurement report data from at least one UE, and sends the measurement report data of the UE and the measurement statistics result of the receiving base station to the OAM through a model training input message.

S308, the OAM receives model training input information comprising training data sent by the first base station and the second base station, and based on the training data, the model is trained to obtain second configuration information of the model.

After the OAM receives the model training input message sent from the base station within the load balancing area, the OAM may divide the data carried by the model training input message into a training data set, a verification data set and a test data set; model training is performed based on the first configuration information and the training data set, parameters are adjusted based on the verification data set, and the second configuration information of the updated model is obtained based on the test data set optimizing characteristics, so that a more accurate model can be obtained.

Optionally, in an embodiment of the present disclosure, the second configuration information includes at least one of the following 3-1 to 3-10:

3-1: a list of secondary base station identities;

it may be understood that the base station identifier included in the auxiliary base station identifier list in the second configuration information may be the same or different from the base station identifier included in the auxiliary base station identifier list in the first configuration information, which is not specifically limited in the embodiment of the present disclosure.

For example, the list of secondary base station identities in the second configuration information may include newly added secondary base station identities, and the remaining secondary base station identities after part of the secondary base station identities in the first configuration information are deleted.

3-2: a second list of feature input information;

wherein the second list indicates at least one characteristic input information collected by the base station for model reasoning, the second list comprising at least one of: the characteristic input information of the UE and the characteristic input information of the base station.

3-3: a registration request information element of the feature input information;

3-4: reporting the characteristics;

for example, if the registration request information element of the feature input information indicates that it is started, each location in the feature indication bitmap is reported as a measurement object reported by the requesting second base station.

3-5: reporting the period;

the reporting period indicates a reporting period of the periodic measurement.

3-6: adding indication information of the characteristic input information;

wherein the addition indication information of the feature input information indicates the newly added feature input information.

3-7: adding indication information of the auxiliary base station;

for example, if the second configuration information carries the auxiliary base station adding instruction information, the receiving base station should add the relevant base station to the auxiliary base station identifier list according to the instruction information, and is responsible for collecting and storing the measurement information from the receiving base station as the model training information.

3-8: change indication information of the model use case;

For example, if the second configuration information carries the change instruction information of the model use case, the receiving base station shall change the model use case, increase or decrease the call of the model use case according to the change instruction information of the model use case.

3-9: change indication information of model class;

for example, if the second configuration information carries the modification instruction information of the model class, the receiving base station modifies the model class according to the instruction information of the model class, so that a model with more suitable and accurate algorithm can be selected.

3-10: change indication information of model parameters;

for example, if the second configuration information carries the indication information for changing the model parameter, the receiving base station should change the model parameter information according to the indication information of the model parameter to obtain a more accurate model.

S309, the OAM sends a third deployment message to the first base station and the second base station, so that the first base station and the second base station perform model reasoning output based on the model updated by the second configuration information, and perform load balancing based on the model reasoning output.

Wherein the third deployment message includes second configuration information of the updated model.

It should be noted that, the second base station identifies the base station in the list for the auxiliary base station indicated in the third deployment message.

Further, the first base station receives a third deployment message sent by the OAM; and the second base station receives the third deployment message sent by the OAM.

It should be noted that, in the embodiment of the present disclosure, if S301 to S308 are not executed, in S309, the OAM may send the indication information of the anchor base station to the first base station together with the second configuration information, that is, the indication information of the anchor base station may be carried in the third deployment message, where the second configuration information is the configuration information corresponding to the OAM trained model.

It is understood that after the receiving base station receives the third deployment message, the receiving base station may update the model in the receiving base station according to the second configuration information of the model. Specifically, the first base station updates the model in the first base station, and the second base station updates the model in the second base station.

In the embodiment of the present disclosure, if the feature input information of the model requirement is updated in the third deployment message, the receiving base station may send the UE measurement configuration to the UE again according to the third deployment message, so that the UE may perform periodic reporting of measurement data based on the feature input information of the latest model requirement.

S310, the first base station sends a request message of model reasoning data to the second base station.

Further, the second base station receives a request message for model inference data transmitted by the first base station.

S311, the second base station sends a reply message to the first base station.

Further, the first base station receives a reply message of the second base station.

After receiving the third deployment message, the first base station requests the model reasoning data from the auxiliary base station according to the latest model indication information and the auxiliary base station identification list in the load balancing area range.

In particular, the request message for model inference data instructs the second base station to start, stop or add a measurement procedure based on the third deployment message.

The request message of the model reasoning data comprises a requested measurement object, an IP address and a port address of the first base station; the reply message comprises a measurement object, an IP address and a port address of the second base station, which can be provided; the request message and the reply message are used for the first base station and the second base station to establish a transmission channel of model reasoning data.

Optionally, in the load balancing method provided by the embodiment of the present disclosure, the request message of the model inference data may include, but is not limited to, at least one of the following 4-1 to 4-8:

4-1: a message type;

wherein the message type indicates: the request message for model reasoning data is used to request the data for model reasoning.

4-2: the first base station measures the ID;

4-3: the second base station measures the ID;

4-4: reporting a cell list;

wherein the reported cell list indicates at least one of d1 and d2:

d1: cell reporting information including, but not limited to: cell ID, SSB (Synchronization Signal and PBCH block, synchronization signal and PBCH (Physical Broadcasting Channel, physical broadcast channel) block) report list, and SSB index.

d2: reporting period, reporting period indicates average window length of all measured objects.

4-5: reporting information of characteristic input information;

wherein, the reporting information of characteristic input information indicates: measuring objects requested by a receiving base station; the measurement object includes at least one of: the characteristic input information of the UE and the characteristic input information of the base station.

Specifically, in the case where the measurement object includes the feature input information of the base station, the measurement object includes the feature input information of at least one cell of the second base station; wherein the characteristic input information of each of the at least one cell includes at least one of: the sum of the current traffic of the UEs in the cell and the radio measurement of the cell. Wherein the wireless measurement of the cell comprises at least one of: the PRB utilization of a cell, the average RRC connection number of a cell, and the packet loss rate of a cell.

4-6: model reasoning information list indication information;

wherein the model inference information list indicates information: indicating the data needed for the new model reasoning.

Optionally, in the load balancing method provided by the embodiment of the present disclosure, the model inference information list indication information includes at least one of the following e1 and e2:

e1: the IP address and the port address of the first base station indicate the second base station to feed back data through the user interface;

e2: feature input information of at least one cell of the second base station;

wherein the model inference information list indicates that the characteristic input information of each of at least one cell of the second base station in the information includes at least one of the following e21-e 26:

e21: a cell radio measurement prediction result;

wherein the wireless measurement result of the cell includes at least one of: the PRB utilization rate of the cell, the average RRC connection number of the cell, and the prediction result of the packet loss rate of the cell.

e22: a UE track prediction result in the cell;

e23: and the UE traffic prediction result in the cell indicates the total prediction of all UE traffic in the cell.

e24: neighbor resource state prediction information indicating a prediction of PRB usage for each cell and each SSB region of all traffic in the downlink and uplink.

e25: the prediction result of the wireless measurement of the adjacent cell comprises at least one of the following: the PRB utilization rate of the neighbor cells, the average RRC connection number of the neighbor cells and the prediction result of the packet loss rate of the neighbor cells.

e26: traffic is offloaded to the UE radio measurement information of the cell.

4-7: a registration request information element of the feature input information;

if the registration request information element of the feature input information indicates addition, the receiving base station shall add the measurement quantity or the predicted value indicated in the model reasoning information list to the measurement started by the list of the feature input information indicated by the first configuration information. If the receiving base station has initiated the measurement indicated by the model inference information list indication information, the receiving base station ignores the information.

4-8: and reporting the characteristics.

Optionally, in the load balancing method provided by the embodiment of the present disclosure, the reply message of the model inference data includes at least one of the following 5-1 to 5-5:

5-1: message type, indicating that the reply message is a model reasoning reply message;

5-2: the IP address and the port address of the second base station indicate the second base station to feed back data through the user interface;

5-3: the first base station measures the ID;

5-4: the second base station measures the ID;

5-5: the first critical diagnosis instruction information indicates an unintelligible or lost message in the messages received by the second base station.

S312, the second base station sends a report message comprising the model reasoning data to the first base station.

And the first base station receives a report message which is sent by the second base station and comprises model reasoning data.

The report message of the model reasoning data is used for reporting the requested measurement information to the second base station.

Optionally, in an embodiment of the present disclosure, the report message of the model inference data includes at least one of the following 6-1 to 6-5:

6-1: message type, indicating that model reasoning data acquisition is successful;

6-2: a measurement ID of the first base station;

6-3: a measurement ID of the second base station;

6-4: cell measurements, the measurements of the cell comprising at least one of: the cell ID and the request message of the model reasoning data request the measurement reporting data corresponding to the reported characteristic input information.

The feature input information requested to be reported by the request message of the model reasoning data comprises at least one of the following: the characteristic input information of the UE and the characteristic input information of the base station.

The characteristic input information of the base station requesting reporting in the request message of the model reasoning data comprises at least one of the following: current base station radio measurement information, predicted UE trajectories, predicted UE traffic totals, current UE traffic totals, neighbor resource state prediction information, neighbor radio measurement predictions, and UE radio measurements offloaded to cells.

6-5: a time stamp.

S313, the first base station performs model reasoning based on the model updated by the second configuration information, the model reasoning data and the measurement report data of the UE, and determines model reasoning output for load balancing.

The measurement report data of the UE are measurement results periodically reported by the UE in the coverage area of the first base station.

Optionally, the model inference output includes at least one of predictive information and decision information for load balancing. The model inference output can be used for analysis or execution of actions by the network.

Alternatively, in embodiments of the present disclosure, the model inference output described above may include at least one of the following 7-1 to 7-6:

7-1: consumption prediction information of virtual resources;

for example, consumption prediction information for virtual resources may indicate predicted base station node computational power and average consumption of memory.

7-2: a list of load balancing cells indicating target cells carrying handover traffic;

7-3: radio resource state prediction information of a target cell, indicating a predicted radio resource utilization rate of the target cell;

7-4: a list of predicted congested cells indicating congested cells within a predicted effective time range;

7-5: a list of predicted congestion relief cells indicating cells for which predicted congestion is relieved within a predicted effective range;

7-6: and predicting the UE migrated to the target cell, and indicating the predicted UE switched to the target base station, wherein the target base station is the base station where the target cell is located.

S314, the first base station sends model reasoning output to the second base station, and load balancing operation is carried out on the UE in the coverage area based on the model reasoning output.

The first base station may send the model inference output to the second base station via at least one of a control plane and a user plane of the inter-base station interface, for example. For example, the first base station may send the model inference output to the second base station via at least one of a control plane and a user plane of the Xn interface.

If the model inference output is a handover policy, the first base station selects a cell to be handed over for the UE within the coverage area of the first base station based on the handover policy, and switches the UE to the cell to be handed over corresponding to the handover policy.

S315, the second base station receives the model reasoning output sent by the first base station, and performs load balancing operation on the UE in the coverage area based on the model reasoning output

If the model inference output is a handover policy, the second base station selects a cell to be handed over for the UE within the coverage area of the second base station based on the handover policy, and switches the UE to the cell to be handed over corresponding to the handover policy.

In the embodiment of the disclosure, the AI-based load balancing can make full use of priori data provided by the wireless system, for example, a load balancing strategy can be designed based on prediction of user tracks and prediction of information such as user services, high-quality service experience can be provided for users while network performance is ensured, system capacity is further improved, and manual intervention of network management and optimization tasks can be reduced to the greatest extent.

The embodiment of the disclosure provides a load balancing method, a network management side firstly executes a model training function, sets an anchor base station for a base station in a management range of the model training function to collect, process and report required data, and sends first configuration information of the model to the anchor base station (namely a first base station) and an auxiliary base station (namely a second base station) determined by OAM in a load balancing area range. The first base station is used as an anchor base station to collect measurement data required by model training from the UE and from adjacent base stations, and send relevant data to the OAM for online model training, the OAM updates and configures the trained model to the base stations, namely second configuration information of the model is sent to the first base station and the second base station, so that the first base station and the second base station update the model. The first base station continuously collects measurement data from the UE and the base station and conducts model reasoning, prediction information including network load, terminal track prediction and the like, decision information about terminal switching and the like are output, so that the network is triggered to execute corresponding operations, for example, the first base station conducts load balancing operation on the UE in the coverage area of the first base station, and the second base station conducts load balancing operation on the UE in the coverage area of the second base station, and therefore performance of the network can be optimized. According to the scheme, the signaling flow design is carried out aiming at the processes of data collection, model deployment, model training, model reasoning and the like among network nodes in a load balancing scene, and under the condition that the model is an AI/ML related model, the prediction of the network to information such as user service, track and the like through the AI/ML technology can be realized, so that a corresponding load balancing strategy is adopted, the problems of overlarge air interface signaling overhead and the like caused by network management and optimization in a traditional manual control mode are solved, and the user service experience can be improved while the network performance is ensured more flexibly and intelligently.

Optionally, in the load balancing method provided by the embodiment of the present disclosure, before S301, S300 may further include:

s300, the OAM determines whether the network meets the preset condition, and the preset condition indicates the OAM to trigger a load balancing mechanism.

Wherein the preset conditions include at least one of the following: the access quantity of the cell terminals of the base station is unbalanced, and the ping-pong switching frequency of the terminals accessed to the base station is larger than a preset threshold value.

Based on the scheme, the OAM can perform offline model training through the stored historical data under the condition that the network performance is not good, and send the trained model to each base station node of the network. For example, under the condition that the access quantity of the cell terminal of the base station is unbalanced or the frequency of ping-pong switching of the access base station is greater than a preset threshold value, the load balancing method based on model training and reasoning in the embodiment of the disclosure is triggered, so that future network performance can be accurately predicted in time according to the actual access information of the base station, load balancing can be performed in advance by each base station, the probability of user service quality degradation is reduced, and the overall network service quality is ensured.

Optionally, as shown in fig. 4, in the load balancing method provided in the embodiment of the present disclosure, in a case where the measurement requested in the request message of the model inference data by the second base station cannot be provided, that is, after S310 described above, the method may further include S316:

S316, the second base station sends a failure message to the first base station.

Further, the first base station receives a failure message sent by the second base station.

Wherein the failure message indicates that the second base station cannot provide the measurement of each request indicated in the request message of the model inference data.

Optionally, in an embodiment of the present disclosure, the failure message described above includes, but is not limited to, at least one of the following 8-1 to 8-6:

8-1: message type, indicating failure of model reasoning data acquisition;

8-2: the first base station measures the ID;

8-3: the second base station measures the ID;

8-4: the failure reasons indicate the reasons of the special events of the XnAP protocol (Xn application protocol, xn interface application flow protocol), and at least comprise wireless network layer reasons, transport layer reasons, protocol reasons and the like;

8-5: second critical diagnostic indication information indicating that the received message is not understood, lost or contains information of a logical error;

8-6: waiting for a retransmission time indicating a time at which the first base station re-initiates the request.

Based on the scheme, under the condition that the second base station cannot provide the model reasoning data for the first base station, the second base station can report the failure message to the first base station, so that the first base station can reselect other auxiliary base stations or initiate the request message of the model reasoning data to the second base station.

Optionally, in the load balancing method provided in the embodiment of the present disclosure, in conjunction with fig. 4, after S313 described above, the following S317 may further be included:

and S317, the first base station sends a model performance feedback message to the OAM.

Optionally, in an embodiment of the present disclosure, the model performance feedback information includes at least one of the following 9-1 to 9-4:

9-1: calculating power consumption overhead, wherein the power consumption overhead indicates at least one of the following: the power consumption cost of model training and the power consumption cost of model reasoning;

9-2: model prediction confidence, which indicates the accuracy of the measured data prediction;

9-3: training is time-consuming;

9-4: reasoning is time consuming.

Based on the scheme, the first base station can feed back the performance of the model to the OAM after obtaining the model reasoning output, so that the model and the network service quality of the evaluation model can be optimized, namely, a network management side performs model optimization according to the network performance feedback information of each base station node, and network intelligent load balancing is realized.

Optionally, in the load balancing method provided by the embodiment of the present disclosure, after the base station performs the load balancing mechanism, the network performance of the base station after load balancing may also be reported to the OAM, so that the OAM may determine whether to continue to update the model to perform load balancing or stop load balancing based on the reported network performance. Illustratively, after S314 described above, as shown in fig. 4, the method further includes S318 described below, and after S315 described above, the method may further include S319 described below:

And S318, the first base station sends the measurement feedback message after load balancing to the OAM.

And S319, the second base station sends the measurement feedback message after load balancing to the OAM.

Further, the OAM receives measurement feedback messages sent by the first base station and the second base station.

The measurement feedback message indicates network performance of the base station after load balancing.

Optionally, the measurement feedback message comprises at least one of the following 10-1 to 10-3:

10-1: UE measurement information from the target base station;

under the condition that the measurement feedback message is sent by a first base station, the target base station is the first base station or the base station to which the UE is switched in the coverage area of the first base station; and under the condition that the measurement feedback message is sent by the second base station, the target base station is the second base station or the base station to which the UE is switched in the coverage area of the second base station.

10-2: resource status information from the target base station;

10-3: system KPI (Key Performance Indicator ) of the target system where the target base station is located.

Wherein the system KPI of the target system comprises at least one of the following: throughput, delay, RLF of the target system (Radio Link Failure, radio link failure rate) and RLF of neighbor systems of the target system.

Optionally, in the load balancing method provided by the embodiment of the present disclosure, if the measurement feedback message indicates that the network performance after load balancing does not meet the preset network performance, the above-mentioned processes of model update, reasoning and load balancing are repeatedly performed (i.e. S308 to S315 described above); if the measurement feedback information indicates that the network performance after load balancing meets the preset network performance; illustratively, as shown in fig. 4, the method may further include the following S320 to S322:

and S320, under the condition that the measurement feedback information indicates that the network performance meets the preset condition, the OAM sends a model training pause message to the first base station and the second base station.

Wherein, the network performance meets the preset condition indication: the radio resource utilization rates of different cells in the network reach equilibrium, and the state tends to be stable.

Wherein the model training suspension message indicates that the receiving base station exits the load balancing mechanism.

Optionally, in an embodiment of the present disclosure, the model training pause message includes at least one of the following 11-1 and 11-2:

11-1: load balancing stop indication information;

11-2: the reason for the stop.

Illustratively, the cause of the stopping includes at least one of: network resource saving and UE energy saving.

S321, if the model training pause message sent by the OAM is received, the first base station stops the load balancing operation.

S322, if the model training pause message sent by the OAM is received, the second base station stops the load balancing operation.

Based on the scheme, under the condition that the wireless resource utilization rates of different cells in the network reach equilibrium and the state tends to be stable, the OAM can send a model training pause message to the anchor base station and other base stations within the range of the load balancing effective area so as to instruct the network to temporarily exit from a load balancing mechanism based on a model technology, stop the transmission of signaling and data about model training input, model reasoning output and the like, and further save network resources.

It should be noted that, in the load balancing method provided by the embodiment of the present disclosure, the execution body may also be a load balancing device, or a control module in the load balancing device for executing the load balancing method. In the embodiment of the present disclosure, a method for performing load balancing by using a load balancing device is taken as an example, and the load balancing device provided by the embodiment of the present disclosure is described.

Fig. 5 is a schematic structural diagram of a base station according to an embodiment of the present disclosure, and as shown in fig. 5, a base station 500 is a first base station, including: a receiving module 501, a transmitting module 502, a model reasoning module 503 and a load balancing module 504; a receiving module 501, configured to receive a third deployment message sent by the OAM, where the third deployment message includes second configuration information of the updated model; the sending module 502 is configured to send a request message of model inference data to a second base station; the receiving module 501 is further configured to receive a reply message of the second base station; the request message comprises a requested measurement object, an IP address and a port address of the first base station; the reply message comprises a measurement object which can be provided, an IP address and a port address of the second base station; the request message and the reply message are used for establishing a transmission channel of model reasoning data between the first base station and the second base station, and the second base station is an auxiliary base station of the first base station; the receiving module 501 is further configured to receive a report message including model inference data sent by the second base station; the model reasoning module 503 is configured to perform model reasoning based on the model updated by the second configuration information, the model reasoning data, and measurement report data of the UE, and determine a model reasoning output for load balancing; the sending module 502 is further configured to send the model inference output to the second base station; the load balancing module 504 is configured to perform load balancing operation on UEs within a coverage area based on the model inference output.

Optionally, the base station 500 may further include: a determining module and a measuring module; the determining module is used for determining that the first base station is an anchor base station based on the first deployment message sent by the OAM; the first deployment message comprises anchor point base station indication information and first configuration information of a model; the measurement module is used for measuring based on the first configuration information; the sending module is used for sending UE measurement configuration to the UE in the coverage area based on the first configuration information; the determining module is further configured to determine training data required by a training model according to measurement report data of the UE and measurement statistics results of the first base station; the sending module is further configured to send a model training input message including the training data to the OAM.

Optionally, the first configuration information includes at least one of: the method comprises the steps of a secondary base station identification list, a model index, a first list of feature input information and a registration request information element of the feature input information; the auxiliary base station identification list indicates effective auxiliary base stations in the load balancing area range, and the auxiliary base stations are used for providing measurement report data of characteristic input information for anchor point base stations; a model index indicating the applicable use cases of the model to be deployed and the algorithm used by the model; a first list of feature input information indicating at least one feature input information collected by the receiving base station for model training; the first list of feature input information includes at least one of: the characteristic input information of the UE and the characteristic input information of the base station.

Optionally, the model index indicates at least one of: model use cases, model categories and model parameters; the model use cases indicate the model use cases required for realizing the target use cases; the model class indicates an adaptation model of the target use case; the model parameters are configuration variables inside the model, the configuration variables comprising at least one of: weight, bias, learning rate, and number of iterations.

Optionally, the second configuration information includes at least one of: the method comprises the steps of an auxiliary base station identification list, a second list of feature input information, a registration request information element of the feature input information, a reporting feature, a reporting period, addition indication information of the feature input information, addition indication information of the auxiliary base station, change indication information of a model use case, change indication information of a model class and change indication information of a model parameter; wherein the second list of feature input information indicates at least one feature input information collected by the base station for model reasoning, the second list of feature input information comprising at least one of: characteristic input information of UE and characteristic input information of a base station; the addition instruction information of the feature input information indicates newly added feature input information.

Optionally, the request message includes at least one of: the method comprises the steps of message type, first base station measurement identification ID, second base station measurement ID, report cell list, report information of feature input information, model reasoning information list indication information, registration request information element of feature input information and report feature; wherein the message type indicates: the request message is used for requesting the data of the model reasoning; the reported cell list indicates at least one of: the cell reports information and reporting period; the cell reporting information includes at least one of: cell ID, SSB report list, SSB index; reporting period indicates average window length of all measurement objects; reporting information indication of characteristic input information: measuring objects requested by a receiving base station; the measurement object includes at least one of: characteristic input information of UE and characteristic input information of a base station; model reasoning information list indication information: indicating the data needed for the new model reasoning.

Optionally, in the case of the feature input information of the base station included in the measurement object, the measurement object includes feature input information of at least one cell of the second base station; wherein the characteristic input information of each of the at least one cell comprises at least one of: the sum of the current flow of the UE in the cell and the wireless measurement information of the cell; the radio measurement information of the cell includes at least one of: the PRB utilization of a cell, the average RRC connection number of a cell, and the packet loss rate of a cell.

Optionally, the model inference information list indication information includes at least one of: the IP address and the port address of the first base station indicate the second base station to feed back data through the user interface; feature input information of at least one cell of the second base station; wherein the characteristic input information of each of the at least one cell comprises at least one of: the method comprises the steps of predicting a wireless measurement result of a cell, predicting a UE track in the cell, predicting a UE flow in the cell, predicting information of neighbor cell resource states, predicting a wireless measurement result of a neighbor cell, and unloading the flow to the UE wireless measurement information of the cell.

Optionally, the report message includes at least one of: message type, indicating that model reasoning data acquisition is successful; a measurement ID of the first base station; a measurement ID of the second base station; cell measurement results, the cell measurement results include: the cell ID and the measurement report data corresponding to the characteristic input information requested to report by the request message; the feature input information requested to be reported by the request message comprises at least one of the following: characteristic input information of UE and characteristic input information of a base station; a time stamp.

Optionally, the model training input message comprises: the time stamp and the measurement report data of the feature input information indicated in the first configuration information.

Optionally, the request message instructs the second base station to start, stop or increase the measurement procedure based on the third deployment message.

Optionally, the feature input information of the UE includes at least one of: UE location information, UE historical mobility information, UE movement speed, UE wireless measurement information; wherein the UE wireless measurement information includes at least one of: RSRP, RSRQ, and SINR.

Optionally, the feature input information of the base station includes at least one of: current UE traffic sum, current base station radio measurement information, and predicted base station radio measurement information; wherein the base station wireless measurement information includes at least one of: the PRB utilization of the cells included in the base station, the average RRC connection number of the cells included in the base station, and the cell packet loss rate included in the base station.

Optionally, the feature input information of the base station requested to be reported by the request message includes at least one of the following: current base station wireless measurement information, predicted UE trajectories, predicted UE traffic totals, current UE traffic totals, neighbor resource state prediction information, neighbor wireless measurement prediction results, and UE wireless measurement offloaded to a cell; wherein the base station wireless measurement information includes at least one of: the PRB utilization rate of the cells included in the base station, the average RRC connection number of the cells included in the base station, and the cell packet loss rate included in the base station.

Optionally, if the registration request information element of the feature input information indicates start, the receiving base station starts measurement according to the indication in the list of the feature input information; or if the registration request information element of the feature input information indicates stopping, the receiving base station stops measuring and reporting; or if the registration request information element of the feature input information indicates addition, the receiving base station adds the measured value or the preset value indicated by the registration request information element to the measurement initiated by the list indication of the feature input information.

Optionally, if the registration request information element of the feature input information indicates that the registration request information element is started, each position in the feature indication bitmap is reported as a measurement object reported by the requested second base station.

Optionally, the UE measurement configuration includes at least one of: UE historical mobility information, RRM measurement configuration, and MDT measurement configuration; wherein the RRM measurement configuration comprises at least one of: triggering information of periodic measurement and UE wireless measurement information; the MDT measurement configuration includes at least one of: trigger information of periodic measurement, UE position information and UE moving speed; periodically measured trigger information including a trigger period and a recorded period; the UE wireless measurement information includes at least one of: RSRP, RSRQ, and SNIR.

Optionally, the measurement report data of the UE includes at least one corresponding measurement report data of: UE historical mobility information, RRM measurement configuration, and MDT measurement configuration; the measurement report data corresponding to the RRM measurement configuration comprises: UE wireless measurement information and time stamp; the measurement report data corresponding to the MDT measurement configuration comprises: at least one of UE location information and UE movement speed, and a time stamp.

Optionally, the reply message includes at least one of: message type, indicating that the reply message is a model reasoning reply message; the IP address and the port address of the second base station indicate the second base station to feed back data through the user interface; the first base station measures the ID; the second base station measures the ID; the first critical diagnosis instruction information indicates an unintelligible or lost message in the messages received by the second base station.

Optionally, the receiving module is further configured to receive a failure message sent by the second base station after the sending module sends the request message of the model inference data to the second base station, where the failure message indicates that the second base station cannot provide the measurement of each request indicated by the request message; wherein the failure message includes at least one of: message type, indicating failure of model reasoning data acquisition; the first base station measures the ID; the second base station measures the ID; failure cause, indicating the cause of the XnAP protocol specific event; the failure cause includes at least one of: radio network layer reasons, transport layer reasons, and protocol reasons; second critical diagnostic indication information indicating that the received message is not understood, lost or contains information of a logical error; waiting for a retransmission time indicating a time at which the first base station re-initiates the request.

Optionally, the sending module is further configured to send a measurement feedback message after load balancing to the OAM after the load balancing module performs load balancing operation on the UE in the coverage area based on model reasoning output; wherein the measurement feedback message indicates network performance of the load-balanced base station.

Optionally, the measurement feedback message includes at least one of: UE measurement information from a target base station, resource state information from the target base station and system KPI of a target system in which the target base station is located; the target base station is a first base station or a base station to which the UE is switched in the coverage area of the first base station; the system KPI of the target system comprises at least one of the following: throughput, delay, RLF of the target system, and RLF of neighbor systems of the target system.

Optionally, the load balancing module is configured to stop the load balancing operation if the model training pause message sent by the OAM is received after the sending module sends the measurement feedback message after load balancing to the OAM; the model training suspension message indicates the first base station to exit the load balancing mechanism; the model training pause message includes at least one of: load balancing stop indication information and stop reason; the stopping reasons comprise at least one of the following: network resource saving and UE energy saving.

Optionally, the sending module is further configured to send a model performance feedback message to the OAM after the model inference output module determines the model inference output for load balancing; wherein the model performance feedback message includes at least one of: the power consumption overhead, the model prediction confidence, the training time consumption and the reasoning time consumption are calculated; the power consumption overhead indicates at least one of: the power consumption cost of the model training and the power consumption cost of the model reasoning; the model prediction confidence indicates the accuracy of the measured data predictions.

Optionally, the model inference output includes at least one of predictive information and decision information for load balancing.

Optionally, the sending module is specifically configured to: and sending the model reasoning output to the second base station through at least one of a control plane and a user plane of the interface between the base stations.

Optionally, the model inference output includes at least one of: consumption prediction information of virtual resources, which indicates predicted base station computing power and average consumption of memory; a list of load balancing cells indicating target cells carrying handover traffic; radio resource state prediction information of a target cell, indicating a predicted radio resource utilization rate of the target cell; a list of predicted congestion cells indicating the predicted congestion cells within a predicted effective time range; a list of predicted congestion relief cells indicating cells for which predicted congestion is relieved within a predicted effective range; predicted UE migration to target cell.

Optionally, the load balancing module is specifically configured to: if the model reasoning output is the switching strategy, the first base station selects a cell to be switched for the UE in the coverage area of the first base station based on the switching strategy, and switches the UE to the cell to be switched corresponding to the switching strategy.

The base station 500 provided in the embodiment of the present disclosure can implement each process implemented by the first base station in the method embodiment of fig. 1 to fig. 4, and can achieve the same technical effects, and for avoiding repetition, a detailed description is omitted here.

Fig. 6 is a schematic structural diagram of a base station according to an embodiment of the present disclosure, and as shown in fig. 6, a base station 600 is a second base station, where the second base station includes: a receiving module 601, a transmitting module 602, and a load balancing module 603; the receiving module 601 is configured to receive a third deployment message sent by an OAM, where the third deployment message includes second configuration information of a model; the receiving module 601 is further configured to receive a request message of model inference data sent by the first base station; the sending module 602 is further configured to send a reply message to the first base station; the first base station is an anchor base station, the second base station is an auxiliary base station of the first base station, and the request message comprises a requested measurement object, an IP address and a port address of the first base station; the reply message comprises a measurement object, an IP address and a port address of the second base station, wherein the request message and the reply message are used for establishing a transmission channel of model reasoning data between the first base station and the second base station; the sending module 602 is further configured to send a report message including model inference data to the first base station; the receiving module 601 is further configured to receive a model inference output sent by the first base station; the load balancing module 603 is configured to perform load balancing operation on UEs within a coverage area based on the model inference output.

Optionally, the base station 600 further includes: a measurement module and a determination module; the measurement module is used for measuring based on the second deployment message sent by the OAM; the sending module is further configured to send UE measurement configuration to the UE in the coverage area based on a second deployment message sent by the OAM, where the second deployment message includes first configuration information of the model; the determining module is used for determining training data required by the training model according to measurement report data of the UE and measurement statistical results of the second base station; the sending module is further configured to send a model training input message including the training data to the OAM.

Optionally, the feature input information of the base station requested to be reported by the request message includes at least one of the following: current UE traffic sum, current base station radio measurement information, and predicted base station radio measurement information; wherein the base station wireless measurement information includes at least one of: the PRB utilization of the cells included in the base station, the average RRC connection number of the cells included in the base station, and the packet loss rate of the cells included in the base station.

Optionally, the feature input information of the base station includes at least one of: current base station wireless measurement information, predicted UE trajectories, predicted UE traffic totals, current UE traffic totals, neighbor resource state prediction information, neighbor wireless measurement prediction results, and UE wireless measurement offloaded to a cell; wherein the base station wireless measurement information includes at least one of: the PRB utilization rate of the cells included in the base station, the average RRC connection number of the cells included in the base station, and the cell packet loss rate included in the base station.

Optionally, the sending module is further configured to send a failure message to the first base station if the receiving module receives the request message and fails to provide the measurement of each request indicated in the request message; wherein the failure message includes at least one of: message type, indicating failure of model reasoning data acquisition; the first base station measures the ID; the second base station measures the ID; failure cause, indicating the cause of the XnAP protocol specific event; the failure cause includes at least one of: radio network layer reasons, transport layer reasons, and protocol reasons; second critical diagnostic indication information indicating that the received message is not understood, lost or contains information of a logical error; waiting for a retransmission time indicating a time at which the first base station re-initiates the request.

Optionally, the receiving module is specifically configured to: and receiving the first base station transmission model reasoning output through at least one of a control plane and a user plane of the interface between the base stations.

Optionally, the load balancing module is specifically configured to: if the model reasoning output is the switching strategy, the second base station selects a cell to be switched for the UE in the coverage area of the second base station based on the switching strategy, and switches the UE to the cell to be switched corresponding to the switching strategy.

Optionally, the load balancing module is further configured to stop the load balancing operation if the model training pause message sent by the OAM is received after the sending module sends the load-balanced measurement feedback message to the OAM; the model training suspension message indicates the second base station to exit the load balancing mechanism; the model training pause message includes at least one of: load balancing stop indication information and stop reason; the stopping reasons comprise at least one of the following: network resource saving and UE energy saving.

The base station 600 provided in the embodiment of the present disclosure can implement each process implemented by the second base station in the embodiment of the method of fig. 1 to fig. 4, and can achieve the same technical effects, and for avoiding repetition, a detailed description is omitted here.

Fig. 7 is a schematic structural diagram of an OAM according to an embodiment of the present disclosure, as shown in fig. 7, the OAM 700 includes: a determining module 701 and a transmitting module 702; the determining module 701 is configured to determine that the first base station is an anchor base station and the second base station is an anchor base station of the first base station; the sending module 702 is further configured to send a third deployment message to the first base station and the second base station, so that the first base station and the second base station perform model inference output based on the model updated by the second configuration information, and perform load balancing based on the model inference output, where the third deployment message includes the second configuration information.

Optionally, the OAM 700 may further include: the model training module and the receiving module; the model training module is used for training a model based on historical data and a preconfigured load balancing model to obtain first configuration information of the model, wherein the historical data is wireless measurement related data reported by a base station stored in the OAM; the sending module is used for sending a first deployment message to the first base station and sending a second deployment message to the second base station; the first deployment message comprises anchor base station indication information and the first configuration information, the second deployment message comprises the first configuration information, and the second base station is an auxiliary base station of the first base station; the receiving module is used for receiving model training input information which is sent by the first base station and the second base station and comprises training data; the model training module is further configured to train a model to obtain second configuration information of the model based on the training data.

Optionally, the first configuration information includes at least one of: the method comprises the steps of a secondary base station identification list, a model index, a first list of feature input information and a registration request information element of the feature input information; the auxiliary base station identification list indicates effective auxiliary base stations in the load balancing area range, and the auxiliary base stations are used for providing measurement report data of characteristic input information for anchor point base stations; a model index indicating the applicable use cases of the model to be deployed and the algorithm used by the model; a first list of feature input information indicating at least one feature input information collected by the receiving base station for model training; the list of first characteristic input information includes at least one of: the characteristic input information of the UE and the characteristic input information of the base station.

Optionally, the model training module is specifically configured to: dividing data carried by a model training input message into a training data set, a verification data set and a test data set; model training is conducted based on the first configuration information and the training data set, parameters are adjusted based on the verification data set, and the second configuration information of the updated model is obtained based on the test data set optimizing features.

Optionally, the receiving module is further configured to receive the measurement feedback message sent by the first base station and the second base station after the sending module sends the third deployment message to the first base station and the second base station; the measurement feedback message indicates the network performance of the base station after load balancing; the sending module is further used for sending a model training suspension message to the first base station and the second base station under the condition that the measurement feedback message indicates that the network performance of the base station meets the preset condition; the model training suspension message indicates the first base station and the second base station to exit the load balancing mechanism; wherein the model training pause message comprises at least one of: load balancing stop indication information and stop reason; the stopping reasons comprise at least one of the following: network resource saving and UE energy saving.

Optionally, the receiving module is further configured to receive the model performance feedback message sent by the first base station after the sending module sends the third deployment message to the first base station and the second base station; wherein the model performance feedback message includes at least one of: the power consumption overhead, the model prediction confidence, the training time consumption and the reasoning time consumption are calculated; the power consumption overhead indicates at least one of: the power consumption cost of the model training and the power consumption cost of the model reasoning; the model prediction confidence indicates the accuracy of the measured data predictions.

Optionally, the determining module is further configured to determine, before determining that the first base station is an anchor base station, whether the network meets a preset condition, where the preset condition indicates OAM to trigger a load balancing mechanism; wherein the preset conditions include at least one of the following: the access quantity of the cell terminals of the base station is unbalanced, and the ping-pong switching frequency of the terminals accessed to the base station is larger than a preset threshold value.

Optionally, the determining module is specifically configured to: determining a load balancing area range according to at least one of measurement reporting information and geographic information; determining a first base station with a model reasoning function as an anchor base station according to the calculation power and the storage capacity of the base stations in the load balancing area; the anchor base station is used for collecting data required by load balancing model training from the adjacent base stations.

Optionally, the OAM further includes: a segmentation module; the acquisition module is also used for acquiring wireless measurement related historical data reported by the OAM stored load balancing area range base station after the determination module determines the load balancing area range; the segmentation module is used for segmenting the wireless measurement related historical data into a training data set, a verification data set and a test data set.

The OAM 700 provided in the embodiment of the present disclosure can implement each process implemented by the OAM in the embodiments of the methods of fig. 1 to 4, and can achieve the same technical effects, and in order to avoid repetition, a detailed description is omitted here.

Optionally, as shown in fig. 8, the embodiment of the present disclosure further provides a base station 800, including a processor 801, a memory 802, and a program or an instruction stored in the memory 802 and capable of running on the processor 801, where the program or the instruction implements each process of the embodiment of the load balancing method when executed by the processor 801, and the same technical effects can be achieved, and for avoiding repetition, a description is omitted herein.

Optionally, as shown in fig. 9, the embodiment of the present disclosure further provides an OAM 900, including a processor 901, a memory 902, and a program or an instruction stored in the memory 902 and capable of running on the processor 901, where the program or the instruction implements each process of the embodiment of the load balancing method when executed by the processor 901, and the same technical effects can be achieved, and for avoiding repetition, a description is omitted herein.

It should be noted that, the network entity 1000 shown in fig. 10 is only an example, and should not impose any limitation on the functions and usage scope of the embodiments of the present disclosure. Wherein the network entity 1000 may be the first base station, the second base station, or OAM as described above.

As shown in fig. 10, 1000 includes a central processing unit (Central Processing Unit, CPU) 1001 which can perform various appropriate actions and processes according to a program stored in a ROM (Read Only Memory) 1002 or a program loaded from a storage portion 1008 into a RAM (Random Access Memory ) 1003. In the RAM 1003, various programs and data required for system operation are also stored. The CPU 1001, ROM 1002, and RAM 1003 are connected to each other by a bus 1004. An I/O (Input/Output) interface 1005 is also connected to bus 1004.

The following components are connected to the I/O interface 1005: an input section 1006 including a keyboard, a mouse, and the like; an output portion 1007 including a CRT (Cathode Ray Tube), an LCD (Liquid Crystal Display ), and the like, and a speaker, and the like; a storage portion 1008 including a hard disk or the like; and a communication section 1009 including a network interface card such as a LAN (Local Area Network, wireless network) card, a modem, or the like. The communication section 1009 performs communication processing via a network such as the internet. The drive 1010 is also connected to the I/O interface 1005 as needed. A removable medium 1011, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like, is installed as needed in the drive 1010, so that a computer program read out therefrom is installed as needed in the storage section 1008.

In particular, according to embodiments of the present disclosure, the processes described below with reference to flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program embodied on a computer readable medium, the computer program comprising program code for performing the method shown in the flowcharts. In such an embodiment, the computer program may be downloaded and installed from a network via the communication portion 1009, and/or installed from the removable medium 1011. When the computer program is executed by the central processing unit (CPU 1001), various functions defined in the system of the present application are executed.

The embodiment of the present disclosure further provides a readable storage medium, where a program or an instruction is stored, and when the program or the instruction is executed by a processor, the program or the instruction implements each process of the embodiment of the load balancing method, and the same technical effects can be achieved, so that repetition is avoided, and no further description is given here.

Wherein the processor is a processor in the electronic device described in the above embodiment. The readable storage medium includes computer readable storage medium such as ROM, RAM, magnetic disk or optical disk.

The embodiment of the disclosure further provides a chip, where the chip includes a processor and a communication interface, where the communication interface is coupled to the processor, and the processor is configured to run a program or an instruction, implement each process of the above embodiment of the load balancing method, and achieve the same technical effect, so that repetition is avoided, and no further description is given here.

It should be understood that the chips referred to in the embodiments of the present disclosure may also be referred to as system-on-chip chips, chip systems, or system-on-chip chips, etc.

The embodiments of the present disclosure provide a computer program product including instructions, which when executed on a computer, cause the computer to perform the steps of the load balancing method described above, and achieve the same technical effects, and are not repeated herein.

It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element. Furthermore, it should be noted that the scope of the methods and apparatus in the embodiments of the present disclosure is not limited to performing the functions in the order shown or discussed, but may also include performing the functions in a substantially simultaneous manner or in an opposite order depending on the functions involved, e.g., the described methods may be performed in an order different from that described, and various steps may also be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present disclosure may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g., ROM/RAM, magnetic disk, optical disk), including several instructions for causing a terminal (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to perform the method described in the embodiments of the present disclosure.

The embodiments of the present disclosure have been described above with reference to the accompanying drawings, but the present disclosure is not limited to the above-described embodiments, which are merely illustrative and not restrictive, and many forms may be made by those of ordinary skill in the art without departing from the spirit of the disclosure and the scope of the claims, which are all within the protection of the present disclosure.

Claims

1. The load balancing method is applied to a first base station, wherein the first base station is an anchor base station, and is characterized by comprising the following steps:

receiving a third deployment message sent by network management equipment (OAM), wherein the third deployment message comprises second configuration information of a model;

sending a request message of model reasoning data to a second base station;

receiving a reply message of the second base station; the second base station is an auxiliary base station of the first base station, and the request message comprises a requested measurement object, an IP address and a port address of the first base station; the reply message comprises a measurement object which can be provided, an IP address and a port address of the second base station; the request message and the reply message are used for establishing a transmission channel of model reasoning data between the first base station and the second base station;

Receiving a report message comprising model reasoning data sent by the second base station;

performing model reasoning based on the model updated by the second configuration information, the model reasoning data and the measurement report data of the UE, and determining model reasoning output for load balancing;

and sending the model reasoning output to the second base station, and carrying out load balancing operation on the UE in the coverage area based on the model reasoning output.

2. The method of claim 1, wherein prior to receiving the third deployment message sent by the OAM, the method further comprises:

determining that the first base station is an anchor base station based on a first deployment message sent by the OAM; the first deployment message comprises anchor point base station indication information and first configuration information of a model;

based on the first configuration information, measuring and sending UE measurement configuration to the UE in the coverage area, and determining training data required by a training model according to measurement report data of the UE and measurement statistical results of the first base station;

and sending a model training input message comprising the training data to the OAM.

3. The method of claim 2, wherein the first configuration information comprises at least one of: the method comprises the steps of a secondary base station identification list, a model index, a first list of feature input information and a registration request information element of the feature input information;

The auxiliary base station identification list indicates effective auxiliary base stations in the load balancing area range, and the auxiliary base stations are used for providing measurement report data of characteristic input information for anchor point base stations; the model index indicates the applicable use cases of the model to be deployed and the algorithm used by the model; the first list indicating at least one feature input collected by a receiving base station for model training, the first list comprising at least one of: the characteristic input information of the UE and the characteristic input information of the base station.

4. A method according to claim 3, wherein the model index indicates at least one of: model use cases, model categories and model parameters;

the model use cases indicate the model use cases required for realizing the target use cases;

the model class indicates an adaptation model of the target use case;

5. The method of claim 1, wherein the second configuration information comprises at least one of:

the method comprises the steps of an auxiliary base station identification list, a second list of feature input information, a registration request information element of the feature input information, a reporting feature, a reporting period, addition indication information of the feature input information, addition indication information of the auxiliary base station, change indication information of a model use case, change indication information of a model class and change indication information of a model parameter;

Wherein the second list indicates at least one characteristic input information collected by the base station for model reasoning, the second list comprising at least one of: characteristic input information of UE and characteristic input information of a base station; the addition indication information of the feature input information indicates newly added feature input information.

6. The method of claim 1, wherein the request message comprises at least one of: the method comprises the steps of message type, first base station measurement identification ID, second base station measurement ID, report cell list, report information of feature input information, model reasoning information list indication information, registration request information element of feature input information and report feature;

wherein the message type indicates: the request message is used for requesting data of model reasoning;

the reported cell list indicates at least one of the following: the cell reports information and reporting period; the cell reporting information comprises at least one of the following: cell ID, synchronization signal and physical broadcast channel block SSB report list, SSB index; the reporting period indicates the average window length of all the measurement objects;

the reported information of the characteristic input information indicates: measuring objects requested by a receiving base station; the measurement object includes at least one of: characteristic input information of UE and characteristic input information of a base station;

The model reasoning information list indicates information: indicating the data needed for the new model reasoning.

7. The method according to claim 6, wherein in case of the feature input information of the base station comprised by the measurement object, the measurement object comprises feature input information of at least one cell of the second base station;

wherein the characteristic input information of each of the at least one cell includes at least one of: the sum of the current flow of the UE in the cell and the wireless measurement information of the cell;

the wireless measurement information of the cell includes at least one of: the physical resource block PRB utilization of a cell, the average radio resource control RRC connection number of a cell, and the packet loss rate of a cell.

8. The method of claim 6, wherein the model inference information list indication information comprises at least one of:

the IP address and the port address of the first base station indicate the second base station to feed back data through a user interface;

feature input information of at least one cell of the second base station;

wherein the characteristic input information of each of the at least one cell includes at least one of: the method comprises the steps of predicting a wireless measurement result of a cell, predicting a UE track in the cell, predicting a UE flow in the cell, predicting information of neighbor cell resource states, predicting a wireless measurement result of a neighbor cell, and unloading the flow to the UE wireless measurement information of the cell.

9. The method of claim 6, wherein the reporting message comprises at least one of:

message type, indicating that model reasoning data acquisition is successful;

a measurement ID of the first base station;

a measurement ID of the second base station;

a cell measurement result, the cell measurement result comprising: the cell ID and the measurement report data corresponding to the characteristic input information requested to be reported by the request message; the feature input information requested to be reported by the request message comprises at least one of the following items: characteristic input information of UE and characteristic input information of a base station;

a time stamp.

10. A method according to claim 3, wherein the model training input message comprises: and reporting data by measuring the time stamp and the characteristic input information indicated in the first configuration information.

11. The method of claim 1, wherein the request message instructs a second base station to start, stop or add a measurement procedure based on the third deployment message.

12. The method according to any of claims 3 to 10, wherein the UE's feature input information comprises at least one of: UE location information, UE historical mobility information, UE movement speed, UE wireless measurement information;

Wherein the UE wireless measurement information includes at least one of: reference signal received power RSRP, reference signal received quality RSRQ, and signal to interference plus noise ratio SINR.

13. The method according to any of claims 3 to 8 or 10, wherein the characteristic input information of the base station comprises at least one of: current UE traffic sum, current base station radio measurement information, and predicted base station radio measurement information;

wherein the base station wireless measurement information includes at least one of: the PRB utilization of the cells included in the base station, the average RRC connection number of the cells included in the base station, and the cell packet loss rate included in the base station.

14. The method of claim 9, wherein the characteristic input information of the base station comprises at least one of: current base station wireless measurement information, predicted UE trajectories, predicted UE traffic totals, current UE traffic totals, neighbor resource state prediction information, neighbor wireless measurement prediction results, and UE wireless measurement offloaded to a cell;

wherein the base station wireless measurement information includes at least one of: the PRB utilization rate of the cells included in the base station, the average RRC connection number of the cells included in the base station, and the cell packet loss rate included in the base station.

15. The method according to any one of claims 3 to 10, wherein,

if the registration request information element of the feature input information indicates the start, the receiving base station starts measurement according to the indication in the list of the feature input information; or,

if the registration request information element of the feature input information indicates stopping, the receiving base station stops measuring and reporting; or,

if the registration request information element of the feature input information indicates addition, the receiving base station adds the measured value or the preset value indicated by the registration request information element to the measurement started by the list indication of the feature input information.

16. The method of claim 15, wherein each location in the feature indication bitmap is a measurement object reported by the requesting second base station if the registration request information element of the feature input information indicates a start.

17. The method of claim 2, wherein the UE measurement configuration comprises at least one of: UE historical mobility information, radio resource management RRM measurement configuration, and minimization of drive test MDT measurement configuration;

wherein the RRM measurement configuration comprises at least one of: triggering information of periodic measurement and UE wireless measurement information; the MDT measurement configuration includes at least one of: trigger information of periodic measurement, UE position information and UE moving speed; the periodically measured trigger information comprises a trigger period and a recorded period; the UE wireless measurement information includes at least one of: RSRP, RSRQ, and SNIR.

18. The method of claim 17, wherein the measurement report data for the UE comprises measurement report data corresponding to at least one of:

UE historical mobility information, RRM measurement configuration, and MDT measurement configuration;

wherein, the measurement report data corresponding to the RRM measurement configuration includes: UE wireless measurement information and time stamp; the measurement report data corresponding to the MDT measurement configuration comprises: at least one of UE location information and UE movement speed, and a time stamp.

19. The method of claim 1, wherein the reply message comprises at least one of:

the message type indicates that the reply message is a model reasoning reply message;

the IP address and the port address of the second base station indicate the second base station to feed back data through a user interface;

the first base station measures an ID;

the second base station measures an ID;

and the first critical diagnosis indication information indicates an unintelligible or lost message in the messages received by the second base station.

20. The method of claim 1, wherein after the sending the request message for model inference data to the second base station, the method further comprises:

Receiving a failure message sent by the second base station, wherein the failure message indicates that the second base station cannot provide measurement of each request indicated by the request message;

wherein the failure message includes at least one of:

message type, indicating failure of model reasoning data acquisition;

the first base station measures an ID;

the second base station measures an ID;

failure cause, indicating the cause of the XnAP protocol specific event; the failure cause includes at least one of: radio network layer reasons, transport layer reasons, and protocol reasons;

second critical diagnostic indication information indicating that the received message is not understood, lost or contains information of a logical error;

waiting for a retransmission time indicating a time at which the first base station re-initiates the request.

21. The method of claim 1, wherein after the load balancing operation for the in-coverage UEs based on the model inference output, the method further comprises:

sending a measurement feedback message after load balancing to the OAM;

22. The method of claim 21, wherein the measurement feedback message comprises at least one of:

UE measurement information from a target base station, resource state information from the target base station and system key performance index KPI of a target system in which the target base station is located;

the target base station is the first base station or a base station to which the UE is switched in the coverage area of the first base station; the system KPI of the target system comprises at least one of the following: throughput, delay, radio link failure rate RLF of the target system, and RLF of neighbor systems of the target system.

23. The method according to claim 21 or 22, wherein after the sending of the load-balanced measurement feedback message to the OAM, the method further comprises:

if the model training pause message sent by the OAM is received, stopping the load balancing operation;

wherein the model training suspension message indicates that the first base station exits a load balancing mechanism; the model training pause message includes at least one of: load balancing stop indication information and stop reason; the stopping reasons comprise at least one of the following: network resource saving and UE energy saving.

24. The method of claim 1, wherein after the determining the model inference output for load balancing, the method further comprises:

Sending a model performance feedback message to the OAM;

wherein the model performance feedback message includes at least one of: the power consumption overhead, the model prediction confidence, the training time consumption and the reasoning time consumption are calculated; the computational power consumption overhead indicates at least one of: the power consumption cost of the model training and the power consumption cost of the model reasoning; the model prediction confidence indicates the accuracy of the measured data prediction.

25. The method of claim 1, wherein the model inference output comprises at least one of load-balanced prediction information and decision information.

26. The method of claim 25, wherein the model inference output comprises at least one of:

consumption prediction information of virtual resources, which indicates predicted base station computing power and average consumption of memory;

a list of load balancing cells indicating target cells carrying handover traffic;

the radio resource state prediction information of the target cell indicates the predicted radio resource utilization rate of the target cell;

a list of predicted congestion cells indicating the predicted congestion cells within a predicted effective time range;

a list of predicted congestion relief cells indicating cells for which predicted congestion is relieved within a predicted effective range;

Predicted migration to the UE of the target cell.

27. The method of claim 26, wherein said sending a model inference output to said second base station comprises:

and sending model reasoning output to the second base station through at least one of a control plane and a user plane of an interface between the base stations.

28. The method of claim 26, wherein the load balancing the UEs within the coverage area based on the model inference output comprises:

and if the model reasoning output is a switching strategy, the first base station selects a cell to be switched for the UE in the coverage area of the first base station based on the switching strategy, and switches the UE to the cell to be switched corresponding to the switching strategy.

29. A load balancing method applied to a second base station, the method comprising:

receiving a third deployment message sent by network management equipment (OAM), wherein the third deployment message comprises second configuration information of the updated model;

receiving a request message of model reasoning data sent by a first base station;

sending a reply message to the first base station; the first base station is an anchor base station, the second base station is an auxiliary base station of the first base station, and the request message comprises a requested measurement object, an IP address and a port address of the first base station; the reply message comprises a measurement object, an IP address and a port address of the second base station, wherein the request message and the reply message are used for establishing a transmission channel of model reasoning data between the first base station and the second base station;

Sending a report message comprising model reasoning data to the first base station;

and receiving model reasoning output sent by the first base station, and carrying out load balancing operation on the UE in the coverage area based on the model reasoning output.

30. The method of claim 29, wherein prior to receiving the third deployment message sent by the OAM, the method further comprises:

measuring and transmitting UE measurement configuration to the UE in the coverage area based on a second deployment message transmitted by OAM, wherein the second deployment message comprises first configuration information of a model;

according to the measurement report data of the UE and the measurement statistical result of the second base station, determining training data required by a training model;

31. The method of claim 30, wherein the first configuration information comprises at least one of: the method comprises the steps of a secondary base station identification list, a model index, a first list of feature input information and a registration request information element of the feature input information;

the auxiliary base station identification list indicates effective auxiliary base stations in the load balancing area range, and the auxiliary base stations are used for providing measurement report data of characteristic input information for anchor point base stations; the model index indicates the applicable use cases of the model to be deployed and the algorithm used by the model; the first list indicates at least one characteristic input information collected by a receiving base station for model training; the first list includes at least one of: the characteristic input information of the UE and the characteristic input information of the base station.

32. The method of claim 31, wherein the model index indicates at least one of: model use cases, model categories and model parameters;

the model class indicates an adaptation model of the target use case;

33. The method of claim 29, wherein the second configuration information comprises at least one of:

34. The method of claim 29, wherein the request message comprises at least one of: the method comprises the steps of message type, first base station measurement identification ID, second base station measurement ID, report cell list, report information of feature input information, model reasoning information list indication information, registration request information element of feature input information and report feature;

35. The method according to claim 34, wherein in case of the feature input information of the base station comprised by the measurement object, the measurement object comprises feature input information of at least one cell of the second base station;

36. The method of claim 35, wherein the model inference information list indication information comprises at least one of:

feature input information of at least one cell of the second base station;

37. The method of claim 34, wherein the reporting message comprises at least one of:

Message type, indicating that model reasoning data acquisition is successful;

a measurement ID of the first base station;

a measurement ID of the second base station;

a time stamp.

38. The method of claim 31, wherein the model training input message comprises: and reporting data by measuring the time stamp and the characteristic input information indicated in the first configuration information.

39. The method of claim 29, wherein the request message instructs the second base station to start, stop or add a measurement procedure based on the third deployment message.

40. The method according to any of claims 31 to 38, wherein the UE's feature input information comprises at least one of: UE location information, UE historical mobility information, UE movement speed, UE wireless measurement information;

41. The method according to any of claims 31 to 36 or 38, wherein the characteristic input information of the base station comprises at least one of: current UE traffic sum, current base station radio measurement information, and predicted base station radio measurement information;

wherein the base station wireless measurement information includes at least one of: the PRB utilization of the cells included in the base station, the average RRC connection number of the cells included in the base station, and the packet loss rate of the cells included in the base station.

42. The method of claim 37, wherein the characteristic input information of the base station comprises at least one of: current base station wireless measurement information, predicted UE trajectories, predicted UE traffic totals, current UE traffic totals, neighbor resource state prediction information, neighbor wireless measurement prediction results, and UE wireless measurement offloaded to a cell;

43. The method according to any one of claims 31 to 38, wherein,

44. The method of claim 43, wherein each location in the feature indication bitmap is a measurement object reported by the requesting second base station if the registration request information element of the feature input information indicates a start.

45. The method of claim 30, wherein the UE measurement configuration comprises at least one of: UE historical mobility information, radio resource management RRM measurement configuration, and minimization of drive test MDT measurement configuration;

46. The method of claim 30, wherein the measurement report data for the UE comprises measurement report data corresponding to at least one of:

47. The method of claim 29, wherein the reply message includes at least one of:

the first base station measures an ID;

the second base station measures an ID;

48. The method of claim 29, further comprising:

if the measurement of each request indicated in the request message cannot be provided under the condition that the request message is received, a failure message is sent to the first base station;

Wherein the failure message includes at least one of:

message type, indicating failure of model reasoning data acquisition;

the first base station measures an ID;

the second base station measures an ID;

49. The method of claim 29, wherein the model inference output comprises at least one of load-balanced prediction information and decision information.

50. The method of claim 49, wherein the model inference output comprises at least one of:

predicted migration to the UE of the target cell.

51. The method of claim 50, wherein said receiving the model inference output transmitted by the first base station comprises:

and receiving the first base station transmission model reasoning output through at least one of a control plane and a user plane of an interface between the base stations.

52. The method of claim 49, wherein the load balancing the UEs within the coverage area based on the model inference output comprises:

and if the model reasoning output is a switching strategy, the second base station selects a cell to be switched for the UE in the coverage area of the second base station based on the switching strategy, and switches the UE to the cell to be switched corresponding to the switching strategy.

53. The method of claim 29, wherein after the load balancing operation on the in-coverage UEs based on the model inference output, the method further comprises:

Sending a measurement feedback message after load balancing to the OAM;

54. The method of claim 53, wherein the measurement feedback message comprises at least one of:

55. The method according to claim 53 or 54, wherein after said sending the load-balanced measurement feedback message to the OAM, the method further comprises:

wherein the model training suspension message indicates that the second base station exits a load balancing mechanism; the model training pause message includes at least one of: load balancing stop indication information and stop reason; the stopping reasons comprise at least one of the following: network resource saving and UE energy saving.

56. The load balancing method is applied to network management equipment (OAM), and is characterized by comprising the following steps:

determining a first base station as an anchor base station, and determining a second base station as an auxiliary base station of the first base station;

and sending a third deployment message to the first base station and the second base station so that the first base station and the second base station perform model reasoning output of a model updated based on second configuration information of the model and perform load balancing based on the model reasoning output, wherein the third deployment message comprises the second configuration information.

57. The method of claim 56, wherein prior to said sending a third deployment message to said first base station and said second base station, said method further comprises:

obtaining first configuration information of a model based on historical data and a preconfigured load balancing model training model, wherein the historical data is wireless measurement related data reported by a base station stored in the OAM;

sending a first deployment message to the first base station and sending a second deployment message to a second base station; the first deployment message comprises anchor base station indication information and the first configuration information, the second deployment message comprises the first configuration information, and the second base station is an auxiliary base station of the first base station;

And receiving model training input information which is sent by the first base station and the second base station and comprises training data, and training a model to obtain second configuration information of the model based on the training data.

58. The method of claim 57, wherein the first configuration information comprises at least one of: the auxiliary base station identification list, the model index, the list of the characteristic input information and the registration request information element of the characteristic input information;

the auxiliary base station identification list indicates effective auxiliary base stations in the load balancing area range, and the auxiliary base stations are used for providing measurement report data of characteristic input information for anchor point base stations; the model index indicates the applicable use cases of the model to be deployed and the algorithm used by the model; the list of feature input information indicates at least one feature input information collected by the receiving base station for model training; the list of feature input information includes at least one of: the characteristic input information of the UE and the characteristic input information of the base station.

59. The method of claim 58, wherein the model index indicates at least one of: model use cases, model categories and model parameters;

The model use case indicates a model use case required for realizing a target use case, and the target use case comprises at least one of the following: load prediction, UE trajectory prediction, UE mobility prediction, UE traffic prediction;

the model class indicates an adaptation model of the target use case;

60. The method of claim 56, wherein said second configuration information includes at least one of:

the method comprises the steps of an auxiliary base station identification list, a feature input information list, a registration request information element of feature input information, a reporting feature, a reporting period, addition indication information of the feature input information, addition indication information of an auxiliary base station, change indication information of a model use case, change indication information of a model class and change indication information of a model parameter;

wherein the list of feature input information indicates at least one feature input information collected by the base station for model training, the list of feature input information comprising at least one of: characteristic input information of UE and characteristic input information of a base station; the addition indication information of the feature input information indicates newly added feature input information.

61. The method of claim 57, wherein the model training input message comprises: and reporting data by measuring the time stamp and the characteristic input information indicated in the first configuration information.

62. The method of any of claims 58 to 60, wherein the UE's feature input information comprises at least one of: UE location information, UE historical mobility information, UE movement speed, UE wireless measurement information;

63. The method of any one of claims 58 to 60, wherein the characteristic input information of the base station comprises at least one of: current UE traffic sum, current base station radio measurement information, and predicted base station radio measurement information;

wherein the base station wireless measurement information includes at least one of: the physical resource block PRB utilization of a cell comprised by the base station, the average radio resource control RRC connection number of a cell comprised by the base station, and the cell packet loss rate comprised by the base station.

64. The method of any one of claims 58 to 60, wherein,

65. The method of claim 64, wherein each location in the feature indication bitmap is a measurement object reported by the requesting second base station if a registration request information element of the feature input information indicates a start.

66. The method of claim 57, wherein training the model based on the training data yields second configuration information for the model, comprising:

dividing data carried by the model training input message into a training data set, a verification data set and a test data set;

model training is carried out based on the first configuration information and the training data set, parameters are adjusted based on the verification data set, and second configuration information of the updated model is obtained based on the test data set optimizing characteristics.

67. The method of claim 56, wherein after said sending a third deployment message to said first base station and said second base station, said method further comprises:

receiving measurement feedback messages sent by the first base station and the second base station; the measurement feedback message indicates the network performance of the base station after load balancing;

sending a model training suspension message to the first base station and the second base station under the condition that the measurement feedback message indicates that the network performance of the base station meets the preset condition;

wherein the model training suspension message indicates that the first base station and the second base station exit a load balancing mechanism; wherein the model training pause message comprises at least one of: load balancing stop indication information and stop reason; the stopping reasons comprise at least one of the following: network resource saving and UE energy saving.

68. The method of claim 67, wherein the measurement feedback message comprises at least one of:

69. The method of claim 56, wherein after said sending a third deployment message to said first base station and said second base station, said method further comprises:

receiving a model performance feedback message sent by the first base station;

70. The method of claim 56, wherein prior to said determining that the first base station is an anchor base station, the method further comprises:

determining whether a network meets a preset condition, wherein the preset condition indicates the OAM to trigger a load balancing mechanism;

Wherein the preset conditions include at least one of: the access quantity of the cell terminals of the base station is unbalanced, and the ping-pong switching frequency of the terminals accessed to the base station is larger than a preset threshold value.

71. The method of claim 56, wherein said determining that the first base station is an anchor base station comprises:

determining a load balancing area range according to at least one of measurement reporting information and geographic information;

determining the first base station with a model reasoning function as an anchor base station according to the computing power and the storage capacity of the base stations in the load balancing area;

the anchor base station is used for collecting data required by load balancing model training from adjacent base stations.

72. The method of claim 71, wherein after the determining the load balancing area range, the method further comprises:

acquiring wireless measurement related historical data reported by the load balancing area range base station stored in the OAM;

the wireless measurement related history data is partitioned into a training data set, a validation data set, and a test data set.

73. A base station, wherein the base station is a first base station, the first base station is an anchor base station, the first base station comprising: the system comprises a receiving module, a sending module, a model reasoning module and a load balancing module;

The receiving module is configured to receive a third deployment message sent by the network management equipment OAM, where the third deployment message includes second configuration information of the model;

the sending module is used for sending a request message of model reasoning data to the second base station;

the receiving module is further configured to receive a reply message of the second base station; the request message comprises a requested measurement object, an IP address and a port address of the first base station; the reply message comprises a measurement object which can be provided, an IP address and a port address of the second base station; the request message and the reply message are used for establishing a transmission channel of model reasoning data between the first base station and the second base station, and the second base station is an auxiliary base station of the first base station;

the receiving module is further configured to receive a report message including model reasoning data sent by the second base station;

the model reasoning module is used for carrying out model reasoning based on the model updated by the second configuration information, the model reasoning data and the measurement report data of the UE, and determining model reasoning output for load balancing;

the sending module is further configured to send the model inference output to the second base station;

And the load balancing module is used for carrying out load balancing operation on the UE in the coverage area based on the model reasoning output.

74. A base station, wherein the base station is a second base station, the second base station comprising: the device comprises a receiving module, a sending module and a load balancing module;

the receiving module is further used for receiving a request message of model reasoning data sent by the first base station;

the sending module is used for sending a reply message to the first base station; the first base station is an anchor base station, the second base station is an auxiliary base station of the first base station, and the request message comprises a requested measurement object, an IP address and a port address of the first base station; the reply message comprises a measurement object, an IP address and a port address of the second base station, wherein the request message and the reply message are used for establishing a transmission channel of model reasoning data between the first base station and the second base station;

the sending module is further configured to send a report message including model reasoning data to the first base station;

The receiving module is further used for receiving model reasoning output sent by the first base station;

75. A network management equipment, OAM, the OAM comprising: a determining module and a transmitting module;

the determining module is used for determining that a first base station is an anchor base station and a second base station is an auxiliary base station of the first base station;

the sending module is further configured to send a third deployment message to the first base station and the second base station, so that the first base station and the second base station perform model reasoning output based on the model updated by the second configuration information, and perform load balancing based on the model reasoning output, where the third deployment message includes the second configuration information.

76. A base station comprising a processor, a memory and a program or instruction stored on the memory and executable on the processor, which when executed by the processor, implements the steps of the load balancing method of any one of claims 1 to 28, or 29 to 55.

77. A network management equipment OAM, comprising a processor, a memory, and a program or instruction stored on the memory and executable on the processor, which when executed by the processor, implements the steps of the load balancing method of any one of claims 56 to 72.

78. A readable storage medium having stored thereon a program or instructions which when executed by a processor performs the steps of the load balancing method of any one of claims 1 to 28, or 29 to 55, or 56 to 72.