CN117545019A - Method for load balancing and related equipment - Google Patents
Method for load balancing and related equipment Download PDFInfo
- Publication number
- CN117545019A CN117545019A CN202210893482.2A CN202210893482A CN117545019A CN 117545019 A CN117545019 A CN 117545019A CN 202210893482 A CN202210893482 A CN 202210893482A CN 117545019 A CN117545019 A CN 117545019A
- Authority
- CN
- China
- Prior art keywords
- base station
- information
- model
- terminal
- load balancing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 141
- 238000012549 training Methods 0.000 claims abstract description 331
- 238000005259 measurement Methods 0.000 claims description 345
- 238000012545 processing Methods 0.000 claims description 43
- 238000012360 testing method Methods 0.000 claims description 17
- 230000008859 change Effects 0.000 claims description 15
- 238000004590 computer program Methods 0.000 claims description 15
- 230000000737 periodic effect Effects 0.000 claims description 7
- 238000011156 evaluation Methods 0.000 claims description 6
- 230000007246 mechanism Effects 0.000 claims description 6
- 238000012706 support-vector machine Methods 0.000 claims description 6
- 238000012986 modification Methods 0.000 claims description 5
- 230000004048 modification Effects 0.000 claims description 5
- 230000004044 response Effects 0.000 claims description 4
- 230000006978 adaptation Effects 0.000 claims description 3
- 238000013473 artificial intelligence Methods 0.000 description 77
- 238000010801 machine learning Methods 0.000 description 70
- 238000007726 management method Methods 0.000 description 59
- 238000004891 communication Methods 0.000 description 45
- 230000000875 corresponding effect Effects 0.000 description 45
- 230000006870 function Effects 0.000 description 36
- 238000005516 engineering process Methods 0.000 description 19
- 230000008569 process Effects 0.000 description 15
- 230000005540 biological transmission Effects 0.000 description 11
- 238000004422 calculation algorithm Methods 0.000 description 11
- 238000010586 diagram Methods 0.000 description 11
- 230000011664 signaling Effects 0.000 description 11
- 238000005457 optimization Methods 0.000 description 8
- 238000004458 analytical method Methods 0.000 description 7
- 238000013480 data collection Methods 0.000 description 5
- 238000013461 design Methods 0.000 description 5
- 230000003993 interaction Effects 0.000 description 5
- 230000009286 beneficial effect Effects 0.000 description 4
- 230000004069 differentiation Effects 0.000 description 4
- 238000012417 linear regression Methods 0.000 description 4
- 238000007477 logistic regression Methods 0.000 description 4
- 238000012423 maintenance Methods 0.000 description 4
- 238000012795 verification Methods 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000003066 decision tree Methods 0.000 description 3
- 239000003795 chemical substances by application Substances 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000010354 integration Effects 0.000 description 2
- 208000018910 keratinopathic ichthyosis Diseases 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 238000007637 random forest analysis Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 239000000969 carrier Substances 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000001276 controlling effect Effects 0.000 description 1
- 238000013499 data model Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 230000006698 induction Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000035755 proliferation Effects 0.000 description 1
- 230000002787 reinforcement Effects 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000013526 transfer learning Methods 0.000 description 1
- 238000009827 uniform distribution Methods 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W28/00—Network traffic management; Network resource management
- H04W28/02—Traffic management, e.g. flow control or congestion control
- H04W28/08—Load balancing or load distribution
- H04W28/09—Management thereof
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W28/00—Network traffic management; Network resource management
- H04W28/02—Traffic management, e.g. flow control or congestion control
- H04W28/08—Load balancing or load distribution
- H04W28/09—Management thereof
- H04W28/0925—Management thereof using policies
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W28/00—Network traffic management; Network resource management
- H04W28/02—Traffic management, e.g. flow control or congestion control
- H04W28/08—Load balancing or load distribution
- H04W28/09—Management thereof
- H04W28/0958—Management thereof based on metrics or performance parameters
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W28/00—Network traffic management; Network resource management
- H04W28/02—Traffic management, e.g. flow control or congestion control
- H04W28/08—Load balancing or load distribution
- H04W28/09—Management thereof
- H04W28/0958—Management thereof based on metrics or performance parameters
- H04W28/0967—Quality of Service [QoS] parameters
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Quality & Reliability (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The embodiment of the disclosure provides a method for load balancing and related equipment. The method is performed by a first base station, the method comprising: receiving a first model deployment message sent by a network management system to determine that a first base station is an anchor base station in a load balancing area range and a second base station is an auxiliary base station in the load balancing area range, and deploying an AI/ML-based network load balancing model at the first base station; responding to the first model deployment message to acquire first model training data of the first base station; according to the first model deployment message, sending a model training data request message to the second base station, so as to instruct the second base station to obtain second model training data; receiving a model training data reply message sent by a second base station; receiving a training data reporting message sent by a second base station; training a network load balancing model deployed at the first base station through the first model training data and the second model training data; and sending a model update message to the second base station.
Description
Technical Field
The disclosure relates to the technical field of wireless communication, in particular to a method for load balancing, a first base station, a network management system, a second base station and a computer readable storage medium.
Background
Along with the proliferation of network traffic and the increase of frequency bands, how to better realize the uniform distribution of loads among base stations or cells, improve the service quality of the base stations, and improve the network energy efficiency becomes one of the key research directions in the industry.
Disclosure of Invention
The embodiment of the disclosure provides a method for load balancing, a first base station, a network management system, a second base station and a computer readable storage medium.
Embodiments of the present disclosure provide a method for load balancing, the method being performed by a first base station, the method comprising:
receiving a first model deployment message sent by a network management system to determine that the first base station is an anchor base station in a load balancing area range and the second base station is an auxiliary base station in the load balancing area range, and deploying an AI/ML-based network load balancing model at the first base station;
responding to the first model deployment message to acquire first model training data of the first base station;
according to the first model deployment message, a model training data request message is sent to the second base station and used for indicating the second base station to obtain second model training data, and the model training data request message comprises network address information and port information of the first base station and is used for establishing a channel for transmitting AI/ML data;
Receiving a model training data reply message sent by the second base station;
receiving a training data reporting message sent by the second base station, wherein the training data reporting message comprises the second model training data and network address information and port information of the second base station;
training the network load balancing model deployed at the first base station by the first model training data and the second model training data;
and sending a model update message to the second base station for indicating that the network load balancing model deployed on the second base station is updated with the network load balancing model completed by the training of the first base station.
The embodiment of the disclosure provides a method for load balancing, which is executed by a second base station, wherein the second base station is an auxiliary base station in a load balancing area, the load balancing area also comprises a first base station, and the first base station is an anchor base station in the load balancing area; the first base station is deployed with an AI/ML-based network load balancing model which is trained offline through a network management system;
wherein the method comprises the following steps:
receiving a second model deployment message sent by a network management system, so as to deploy the network load balancing model of the offline training of the network management system to the second base station;
Receiving a model training data request message sent by the first base station, wherein the model training data request message comprises network address information and port information of the first base station and is used for establishing a channel for transmitting AI/ML data;
sending a model training data reply message to the first base station;
obtaining second model training data according to the model training data request message;
and sending a training data report message to the first base station, wherein the training data report message comprises the second model training data, and network address information and port information of the second base station.
The embodiment of the disclosure provides a first base station, which comprises:
the processing unit is used for receiving a first model deployment message sent by the network management system to determine that the first base station is an anchor base station in the load balancing area range and the second base station is an auxiliary base station in the load balancing area range, and deploying an AI/ML-based network load balancing model in the first base station;
the processing unit is further configured to obtain first model training data of the first base station in response to the first model deployment message;
a sending unit, configured to send a model training data request message to the second base station according to the first model deployment message, where the model training data request message is used to instruct the second base station to obtain second model training data, and the model training data request message includes network address information and port information of the first base station, and is used to establish a channel for transmitting AI/ML data;
The receiving unit is used for receiving the model training data reply message sent by the second base station;
the receiving unit is further configured to receive a training data report message sent by the second base station, where the training data report message includes the second model training data and network address information and port information of the second base station;
the processing unit is further configured to train the network load balancing model deployed at the first base station through the first model training data and the second model training data;
the sending unit is further configured to send a model update message to the second base station, to instruct updating of the network load balancing model deployed on the second base station with the network load balancing model for which training of the first base station is completed.
The embodiment of the disclosure provides a network management system, which comprises:
the receiving unit is used for acquiring measurement report information and geographic information sent by the base station;
the processing unit is used for determining a load balancing area range based on the measurement report information and the geographic information;
the processing unit is further configured to determine, according to an evaluation index of the base stations in the load balancing area, that a first base station is an anchor base station in the load balancing area, and that a second base station is an auxiliary base station in the load balancing area;
The processing unit is also used for calling a preconfigured network load balancing model based on AI/ML, and offline training the network load balancing model based on a wireless measurement historical data set reported by a base station in the load balancing area range;
the sending unit is used for sending a first model deployment message to the first base station, used for indicating to deploy the offline trained network load balancing model to the first base station, and used for indicating the first base station to acquire first model training data and second model training data, and training the network load balancing model based on the first model training data and the second model training data;
the sending unit is further configured to send a second model deployment message to the second base station, so as to deploy the offline trained network load balancing model to the second base station.
The embodiment of the disclosure provides a second base station, wherein the second base station is an auxiliary base station in a load balancing area, the load balancing area also comprises a first base station, and the first base station is an anchor base station in the load balancing area; and the first base station is provided with an AI/ML-based network load balancing model which is trained offline through a network management system.
Wherein the second base station includes:
the receiving unit is used for receiving a second model deployment message sent by the network management system so as to deploy the network load balancing model for offline training of the network management system to the second base station;
the receiving unit is further configured to receive a model training data request message sent by the first base station, where the model training data request message includes network address information and port information of the first base station, so as to establish a channel for transmitting AI/ML data;
a sending unit, configured to send a model training data reply message to the first base station;
the processing unit is used for obtaining second model training data according to the model training data request message;
the sending unit is further configured to send a training data report message to the first base station, where the training data report message includes the second model training data and network address information and port information of the second base station.
The disclosed embodiments provide a communication device including at least one processor and a communication interface. The communication interface is configured to interact with other communication devices by the communication device, and when the program instructions are executed in the at least one processor, implement a method according to any one of the possible implementations of the above embodiments.
Optionally, the communication device may further comprise a memory. The memory is used for storing programs and data.
Embodiments of the present disclosure provide a network device including at least one processor and a communication interface. The communication interface is used for information interaction between the network device and other communication devices, and when the program instructions are executed in the at least one processor, the method in any one of the possible implementation manners of the above embodiment is implemented.
Optionally, the network device may further comprise a memory. The memory is used for storing programs and data.
Optionally, the network device comprises a base station.
The disclosed embodiments provide a computer readable storage medium having stored thereon a computer program for execution by a communication device, which when executed by a processor, implements a method in any one of the possible implementations of the embodiments described above.
For example, the computer readable storage medium may have stored therein a computer program for execution by a first base station, which when executed by a processor, implements instructions of the method performed by the first base station as in the above embodiments.
For example, the computer readable storage medium may have stored therein a computer program for execution by a second base station, which when executed by a processor, implements instructions of the method performed by the second base station as in the above embodiments.
For example, the computer readable storage medium may store a computer program for execution by a network management system, which when executed by a processor, implements instructions of the method performed by the network management system as in the above embodiments.
Embodiments of the present disclosure provide a computer program product containing instructions. The computer program product, when run on a communication device, causes the communication device to execute instructions of the method in the above-described parties or any one of the possible implementations of the above-described parties.
For example, the computer program product, when executed on the first base station, causes the first base station to execute instructions of the method in any one of the possible implementations of the embodiments described above.
For example, the computer program product, when executed on the second base station, causes the second base station to execute instructions of the method in any one of the possible implementations of the embodiments described above.
For example, the computer program product, when executed on a network management system, causes the network management system to execute instructions of the method in any one of the possible implementations of the embodiments described above.
The disclosed embodiments provide a system chip comprising an input-output interface and at least one processor for invoking instructions in a memory to perform the operations of the method in any of the above-described possible implementations.
Optionally, the system chip may further include at least one memory for storing instructions for execution by the processor and a bus.
The embodiment of the disclosure provides a wireless communication system, which comprises the first base station, the second base station and a network management system.
Drawings
Fig. 1 schematically illustrates an application scenario diagram of a method for load balancing according to an embodiment of the present disclosure.
Fig. 2 schematically illustrates a flow chart of a method for load balancing according to an embodiment of the present disclosure.
Fig. 3 schematically illustrates a RAN (wireless access network, radio access network/radio network) intelligent network function architecture diagram according to an embodiment of the present disclosure.
Fig. 4 schematically illustrates a wireless network distributed AI (Artificial Intelligence) and ML (Machine Learning) load balancing process diagram according to an embodiment of the present disclosure.
Fig. 5 schematically illustrates a flow chart of a method for load balancing according to another embodiment of the present disclosure.
Fig. 6 schematically illustrates a flow chart of a method for load balancing according to yet another embodiment of the present disclosure.
Fig. 7 schematically illustrates a schematic block diagram of a first base station according to an embodiment of the present disclosure.
Fig. 8 schematically illustrates a schematic block diagram of a network management system according to an embodiment of the present disclosure.
Fig. 9 schematically illustrates a schematic block diagram of a second base station according to an embodiment of the disclosure.
Fig. 10 schematically illustrates a schematic block diagram of a communication device according to an embodiment of the present disclosure.
Detailed Description
Example embodiments will now be described more fully with reference to the accompanying drawings. However, the exemplary embodiments may be embodied in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the example embodiments to those skilled in the art.
In the description of the present disclosure, unless otherwise indicated, "/" means "or" and, for example, a/B may mean a or B. "and/or" herein is merely an association relationship describing an association object, and means that three relationships may exist, for example, a and/or B may mean: a exists alone, A and B exist together, and B exists alone. Furthermore, "at least one" means one or more, and "a plurality" means two or more. The terms "first," "second," and the like do not limit the number and order of execution, and the terms "first," "second," and the like do not necessarily differ.
First, some terms that may be involved in the embodiments of the present disclosure will be explained.
AI: artificial intelligence is a theory, method, technique and application system that utilizes a digital computer or a machine controlled by a digital computer to simulate, extend and expand human intelligence, sense the environment, acquire knowledge and use the knowledge to obtain optimal results. In other words, artificial intelligence is an integrated technology of computer science that attempts to understand the essence of intelligence and to produce a new intelligent machine that can react in a similar way to human intelligence. Artificial intelligence, i.e. research on design principles and implementation methods of various intelligent machines, enables the machines to have functions of sensing, reasoning and decision.
The artificial intelligence technology is a comprehensive subject, and relates to the technology with wide fields, namely the technology with a hardware level and the technology with a software level. Artificial intelligence infrastructure technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation/interaction systems, mechatronics, and the like. The artificial intelligence software technology mainly comprises a computer vision technology, a voice processing technology, a natural language processing technology, machine learning/deep learning, automatic driving, intelligent traffic and other directions.
ML: machine learning is a multi-domain interdisciplinary, involving multiple disciplines such as probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory, and the like. It is specially studied how a computer simulates or implements learning behavior of a human to acquire new knowledge or skills, and reorganizes existing knowledge structures to continuously improve own performance. Machine learning is the core of artificial intelligence, a fundamental approach to letting computers have intelligence, which is applied throughout various areas of artificial intelligence. Machine learning and deep learning typically include techniques such as artificial neural networks, confidence networks, reinforcement learning, transfer learning, induction learning, teaching learning, and the like.
OAM: operation Administration and Maintenance, hereinafter also referred to as a network management system or network management, refers to generally classifying management operations of the network into 3 major categories according to actual needs of network operations of operators: operation (Operation), administration (Administration), maintenance (Maintenance), abbreviated as OAM. The operation mainly completes analysis, prediction, planning and configuration work of daily network and business; maintenance is mainly a daily operation activity performed on testing and fault management of a network and services thereof.
And (3) SVM: support Vector Machine the support vector machine is a generalized linear classifier (generalized linear classifier) for binary classification of data in a supervised learning (supervised learning) manner, and the decision boundary is the maximum margin hyperplane (maximum-margin hyperplane) for solving the learning sample.
RSRP: reference Signal Receiving Power, the reference signal received power, is one of the key parameters and physical layer policy requirements that can represent the radio signal strength in the LTE (Long Term Evolution ) network, and is the average value of the received signal power on all REs (Resource elements) that carry the reference signal in a certain symbol.
RSRQ: reference Signal Receiving Quality, which represents the LTE reference signal reception quality, this measure is mainly to order the different LTE candidate cells according to signal quality.
NG-RAN: next Generation-Radio Access Network,5G radio access network.
NR: english abbreviation of New Radio, new air interface or New Radio.
RRC: radio Resource Control, i.e. radio resource control.
UMTS: universal Mobile Telecommunication System, the generic mobile communication system.
E-UTRAN: evolved UMTS Terrestrial Radio Access Network, evolved UMTS terrestrial radio access network.
eNB: the acronym for E-UTRAN NodeB is the E-UTRAN base station.
SSB: synchronization Signal and PBCH block, i.e. synchronization signal and PBCH block.
PRB: physical Resource Block, i.e. physical resource blocks.
QoS: quality of Service A quality of service, which means that a network can utilize various basic technologies to provide better service capability for specified network communication, is a security mechanism of the network, and is a technology for solving the problems of network delay and blocking
IE: information Elements, information element.
MDT: minimization of drive tests, least squares.
RRM: radio Resource Management radio resource management is to provide service quality guarantee for radio user terminals in a network under the condition of limited bandwidth, and the basic starting point is to flexibly allocate and dynamically adjust available resources of a radio transmission part and the network under the conditions of uneven network traffic distribution, fluctuation of channel characteristics due to channel weakness and interference, and the like, so as to maximally improve the radio spectrum utilization rate, prevent network congestion and keep the signaling load as small as possible.
SINR: signal to Interference plus Noise Ratio, signal to interference plus noise ratio, refers to the ratio of the strength of the received useful signal to the strength of the received interfering signal (noise plus interference).
IP: internet Protocol, internetworking protocol.
XnAP: xn Application Protocol, xn application protocol.
KPI: key Performance Indicator, key performance indicators.
RLF: radio Link Failure radio link failure.
OFDM: orthogonal Frequency Division Multiplexing, i.e., orthogonal frequency division multiplexing.
The scheme provided by the embodiment of the present disclosure can be widely applied to a wireless communication system in order to provide various types of communication contents, such as voice, video, packet data, messaging, broadcasting, and so on. These systems are capable of supporting communication with multiple users by sharing available system resources (e.g., time, frequency, and power). Such multiple access systems include, for example, fourth generation (4G) systems (e.g., long Term Evolution (LTE) systems or LTE-advanced (LTE-a) systems) and fifth generation (5G) systems (which may be referred to as New Radio (NR) systems). These systems may employ techniques such as code division multiple access (Code Division Multiple Access, abbreviated CDMA), time division multiple access (Time division multiple access, TDMA), frequency division multiple access (frequency division multiple access, FDMA), orthogonal frequency division multiple access (Orthogonal Frequency Division Multiple Access, OFDMA), or discrete fourier transform spread OFDM, among others. A wireless multiple-access communication system may include multiple base stations or network access nodes, each supporting communication for multiple communication devices, which may be referred to as User Equipment, UEs, or terminals, simultaneously.
Fig. 1 schematically illustrates an application scenario diagram of a method for load balancing according to an embodiment of the present disclosure.
Fig. 1 illustrates an example of a wireless communication system 100 in accordance with various aspects of the present disclosure. The wireless communication system 100 includes a base station 105, a UE 115, and a core network 130. In some examples, the wireless communication system 100 may be an LTE network, an LTE-a network, or an NR network. In some cases, the wireless communication system 100 may support enhanced broadband communications, ultra-reliable (e.g., mission critical) communications, low latency communications, or communications with low cost and low complexity devices.
Base station 105 may communicate wirelessly with UE 115 via one or more base station antennas. The base stations 105 described herein may include base station transceivers, radio base stations, access points, radio transceivers, node bs, evolved node bs (enbs), next generation node bs or giganode bs (any of which may be referred to as a gNB), home node bs, home evolved node bs, or some other suitable terminology. The wireless communication system 100 may include different types of base stations 105 (e.g., macro base stations or small cells). The UEs 115 described herein are capable of communicating with various types of base stations 105 and network devices (including macro enbs, small cell enbs, gnbs, relay base stations, etc.).
Each base station 105 may be associated with a particular geographic coverage area 110 in which communications with the respective UE 115 are supported. Each base station 105 may provide communication coverage for a respective geographic coverage area 110 via a communication link 125, and the communication link 125 between the base station 105 and the UE 115 may utilize one or more carriers. The communication links 125 shown in the wireless communication system 100 may include: uplink transmissions from UE 115 to base station 105, or downlink transmissions from base station 105 to UE 115. The downlink transmission may also be referred to as a forward link transmission, while the uplink transmission may also be referred to as a reverse link transmission.
The base stations 105 may communicate with the core network 130 and with each other. For example, the base station 105 may interface with the core network 130 through a backhaul link 132. The base stations 105 may communicate with each other directly or indirectly over the backhaul link 134.
In addition, the terminals referred to in the embodiments of the present disclosure are devices that provide voice and/or data connectivity to a user, handheld devices with wireless connection capabilities, or other processing devices connected to a wireless modem. The terminals may be mobile terminals such as mobile telephones (or "cellular" telephones) and computers with mobile terminals, which may be, for example, portable, pocket, hand-held, computer-built-in or car-mounted mobile devices which exchange voice and/or data with radio access networks. A Terminal may also be referred to as a Subscriber Unit (Subscriber Unit), a Subscriber Station (Subscriber Sta tion), a Mobile Station (Mobile Station), a Remote Station (RemoteStation), an AP (Access Point), a Remote Terminal (Remote Terminal), an Access Terminal (Access Terminal), a User Terminal (User Terminal), a User Agent (User Agent), or a UE, without limitation.
Fig. 2 schematically illustrates a flow chart of a method for load balancing according to an embodiment of the present disclosure. The method provided by the embodiment of fig. 2 may be performed by the first base station, but the disclosure is not limited thereto.
As shown in fig. 2, a method provided by an embodiment of the present disclosure may include:
in S210, a first model deployment message sent by a network management system is received, so as to determine that the first base station is an anchor base station in a load balancing area range and the second base station is an auxiliary base station in the load balancing area range, and an AI/ML-based network load balancing model is deployed in the first base station.
The network management system in the embodiment of the disclosure is used for offline training of an AI/ML-based network load balancing model.
In an exemplary embodiment, the first model deployment message may include at least one of:
anchor point base station indication information;
a list of auxiliary base station identifications in the load balancing area;
model index of network load balancing model;
inputting a first characteristic input information list of a network load balancing model;
a first feature input information request information element;
the first characteristic input information adds an information element.
In an exemplary embodiment, the model index of the network load balancing model may include at least one of:
The model use case at least comprises a load prediction model and a terminal track prediction model;
model categories, which may be used to indicate the adaptation model of each model use case;
model parameters, which may include at least one of: weight, bias, learning rate, number of iterations, support vector machine support vector.
In an exemplary embodiment, the first feature input information list of the network load balancing model may include at least one of:
characteristic input information of a first terminal in the range of a first base station;
the first base station characteristic input information;
the first neighbor base station characteristic inputs information.
In an exemplary embodiment, the feature input information of the first terminal within the range of the first base station may include at least one of:
position information of the first terminal;
historical mobility information of the first terminal;
wireless measurement information of the first terminal.
Wherein the first base station characteristic input information may include at least one of:
wireless measurement information of a first base station;
the current terminal traffic of the first base station.
The first neighbor base station characteristic input information may include at least one of the following: .
Wireless measurement information of a first neighboring base station;
And unloading the traffic of the first terminal in the range of the first base station to the first terminal performance measurement information of the first neighbor cell corresponding to the first neighbor base station.
Wherein:
when the first characteristic input information request information element is set to be started, the first base station is instructed to start measurement according to a first characteristic input information list;
when a first feature input information request information element is set to stop, instructing the first base station to stop measurement and reporting;
and when the first characteristic input information request information element is set to be added, the first base station is instructed to start measurement according to the first characteristic input information adding information element.
The first base station may determine, according to the first model deployment message, that the first base station is an anchor base station within a load balancing area range, where the load balancing area range may further include a second base station, and the second base station may be an auxiliary base station within the load balancing area range.
The first base station may deploy the network load balancing model offline trained through the network management system at the first base station based on the first model deployment message, and the second base station may be configured to deploy the network load balancing model offline trained through the network management system.
In S220, first model training data of the first base station is obtained in response to the first model deployment message.
In an exemplary embodiment, the first model training data comprises first terminal model training data.
Wherein, responding to the first model deployment message to obtain the first model training data of the first base station may include: according to the first model deployment message, sending a measurement configuration message to a first terminal in a first base station range so as to instruct the first terminal to measure and obtain the first terminal model training data; and receiving a first terminal sample measurement report message sent by the first terminal, wherein the first terminal sample measurement report message comprises the first terminal model training data.
In an exemplary embodiment, the measurement configuration message may include at least one of:
historical mobility information of the first terminal;
radio resource management measurement configuration information of the first terminal;
the MDT measurement configuration information of the first terminal.
Wherein the radio resource management measurement configuration information of the first terminal includes at least one of:
supporting periodic measurement triggering, including a triggering period and a recording period;
Wireless measurement information of the first terminal.
The minimization of drive test measurement configuration information of the first terminal may include at least one of the following:
supporting periodic measurement triggering, including a triggering period and a recording period;
position information of the first terminal;
the speed of movement of the first terminal.
In an exemplary embodiment, the first model training data comprises first base station model training data.
Wherein, responding to the first model deployment message to obtain the first model training data of the first base station may include:
and triggering measurement of the first base station model training data according to the first model deployment message so as to obtain the first base station model training data.
In S230, according to the first model deployment message, a model training data request message is sent to the second base station, so as to instruct the second base station to obtain second model training data, where the model training data request message includes network address information and port information of the first base station, so as to establish a channel for transmitting AI/ML data.
In an exemplary embodiment, the model training data request message may include at least one of:
A first sample message type;
the first base station measures the identification;
the second base station measures the identification;
the network address information and the port information of the first base station indicate the second base station to perform data feedback through a user interface when the model training data request message comprises the network address information and the port information of the first base station;
reporting a cell list by a sample;
inputting a required characteristic into an information list;
adding a model training information list information element;
inputting information registration request information elements by the required characteristics;
and reporting the characteristics of the sample.
In an exemplary embodiment, the sample report cell list may include at least one of:
the cell reporting information of the second cell corresponding to the second base station;
and a first reporting period.
The required feature input information list indicates feature input information of second cells corresponding to second base stations required for model training, and the feature input information of each second cell may include at least one of the following:
inputting information of characteristics required by samples of a second terminal in the range of a second base station;
inputting information of the required characteristics of the second base station sample;
and inputting information by the needed characteristics of the second neighbor base station sample.
Wherein:
when the required characteristic input information registration request information element is set to be started, the second base station is instructed to start measurement according to the required characteristic input information list;
when the required feature input information registration request information element is set to stop, the second base station is instructed to stop measurement and reporting;
and when the required characteristic input information registration request information element is set to be added, the second base station is instructed to start measurement according to the added model training information list information element.
In an exemplary embodiment, the sample required feature input information of the second terminal within range of the second base station may include at least one of:
second terminal sample position information;
second terminal sample history mobility information;
the second terminal samples wirelessly measure information.
Wherein the second base station sample required feature input information may include at least one of:
the second base station samples wirelessly measuring information;
second sample termination traffic.
The feature input information required by the second neighboring base station sample may include at least one of the following:
the second neighbor base station samples wirelessly measure information;
and unloading the flow to second terminal sample performance measurement information of a second neighbor cell corresponding to a second neighbor base station of the second base station by the second terminal in the range of the second base station.
It should be noted that, the execution order of S220 and S230 is not limited, and both may be executed in parallel, or S220 may be executed before S230 is executed, or S230 may be executed before S220 is executed.
In S240, a model training data reply message sent by the second base station is received.
In an exemplary embodiment, the model training data reply message may include at least one of:
network address information and port information of the second base station;
a second sample message type;
the first base station measures the identification;
the second base station measures the identification;
a first critical diagnostic information element, the first critical diagnostic information element being used to indicate an unintelligible or missing information element in the model training data request message received by the second base station.
In an exemplary embodiment, the method may further include: and receiving a model training data failure message sent by the second base station.
Wherein the model training data failure message may include at least one of:
a third sample message type;
the first base station measures the identification;
the second base station measures the identification;
a first failure cause;
a second critical diagnostic information element, the second critical diagnostic information element being used to indicate an unintelligible or missing information element in the model training data request message received by the second base station;
A first waiting retransmission time.
In S250, a training data report message sent by the second base station is received, where the training data report message includes the second model training data, and network address information and port information of the second base station.
In an exemplary embodiment, the second model training data may include at least one of:
a fourth sample message type;
the first base station measures the identification;
the second base station measures the identification;
the cell measurement result of the second cell corresponding to the second base station comprises a cell measurement item, wherein the cell measurement item comprises a cell identifier and a required characteristic input information list.
Wherein the required feature input information list may include at least one of:
second terminal sample characteristic input information of a second terminal within range of a second base station;
the second base station sample characteristic input information;
and inputting information by the sample characteristics of the second adjacent base station.
In an exemplary embodiment, the second terminal sample feature input information of the second terminal within range of the second base station may include at least one of:
second terminal sample position information;
second terminal sample history mobility information;
The second terminal sample wirelessly measures information;
the first sample time stamp.
Wherein the second base station sample feature input information may include at least one of:
the second base station samples wirelessly measuring information;
second sample terminal traffic;
a second sample timestamp.
The second neighbor base station sample feature input information of the second base station may include at least one of the following:
the second neighbor base station samples wirelessly measure information;
unloading the flow to second terminal sample performance measurement information of a second neighbor cell corresponding to a second neighbor base station by a second terminal in the range of the second base station;
third sample timestamp.
In S260, the network load balancing model deployed at the first base station is trained by the first model training data and the second model training data.
In S270, a model update message is sent to the second base station for indicating to update the network load balancing model deployed on the second base station with the network load balancing model for which the first base station training is completed.
In an exemplary embodiment, the model update message may include at least one of:
a second feature input information list;
A second feature input information request information element;
a first reporting feature;
a second reporting period;
a second feature input information adding information element;
the model use case changes the information element;
model class modification information elements;
the model parameters change the information element.
In an exemplary embodiment, the second feature input information list may include at least one of:
inputting information by the second terminal characteristics;
the second base station characteristic inputs information;
the second neighbor base station characteristics input information.
Wherein:
when the second characteristic input information request information element is set to be started, the second base station is instructed to start measurement according to a second characteristic input information list;
when a second feature input information request information element is set to stop, then instructing the second base station to stop measurement and reporting;
and when the second characteristic input information request information element is set to be added, the second base station is instructed to start measurement according to the second characteristic input information adding information element.
In an exemplary embodiment, the second terminal characteristic input information may include at least one of:
second terminal position information;
the second terminal history mobility information;
The second terminal wirelessly measures information.
Wherein the second base station characteristic input information may include at least one of:
the second base station wirelessly measures information;
and the second current terminal flow.
Wherein the second neighbor base station characteristic input information may include at least one of:
the second adjacent base station wirelessly measures information;
and unloading the flow to second terminal performance measurement information of a second neighbor cell corresponding to the second neighbor base station by the second terminal in the range of the second base station.
In an exemplary embodiment, the method may further include: acquiring first model reasoning data of the first base station; sending a model reasoning data request message to the second base station to instruct the second base station to obtain second model reasoning data, wherein the model reasoning data request message comprises network address information and port information of the first base station and is used for establishing a channel for transmitting AI/ML data; receiving a model reasoning data reply message sent by the second base station; receiving a model reasoning data report message sent by the second base station, wherein the model reasoning data report message comprises the second model reasoning data and network address information and port information of the second base station; processing the first model reasoning data and the second model reasoning data through the network load balancing model deployed at the first base station to obtain model reasoning output information; and sending a model reasoning output message to the second base station, wherein the model reasoning output message comprises model reasoning output information, and the model reasoning output information is used for indicating the first base station and the second base station to execute corresponding network load balancing operation.
In an exemplary embodiment, the first model reasoning data may include first terminal model reasoning data.
The obtaining the first model reasoning data of the first base station may include:
receiving a first terminal current measurement report message sent by a first terminal in a first base station range, wherein the first terminal current measurement report message comprises first terminal model reasoning data;
wherein the first terminal model reasoning data comprises at least one of:
the first terminal current historical mobility information;
first current radio resource management measurement configuration information;
the first current MDT measurement configuration information.
In an exemplary embodiment, the first current radio resource management measurement configuration information may include at least one of:
the first terminal is in wireless measurement information at present;
a fourth current timestamp;
wherein the first current minimization of drive tests measurement configuration information comprises at least one of:
the current position information of the first terminal;
the current moving speed of the first terminal;
and a fifth current timestamp.
In an exemplary embodiment, the first model reasoning data may include first base station model reasoning data.
The first base station model reasoning data may include feature input information of one or more first cells of the first base station, each of which may include at least one of:
predicting a radio resource state, which includes a downlink total physical resource block usage of a corresponding first cell;
predicting terminal traffic, including a sum of terminal traffic in the first cell;
and predicting the wireless resource state of the adjacent base station, wherein the wireless resource state comprises the downlink total physical resource block use condition of the corresponding first cell.
In an exemplary embodiment, the model reasoning data request message may include at least one of:
a first current message type;
the first base station measures the identification;
the second base station measures the identification;
the first base station network address and the port address, and when the model reasoning data request message comprises the first base station network address and the port address, the second base station is instructed to perform data feedback through a user interface;
the cell list is currently reported;
the current feature inputs an information list;
model reasoning information list information element;
a feature input information registration request information element;
the characteristics are currently reported.
In an exemplary embodiment, the currently reported cell list may include at least one of:
the cell reporting information of the second cell corresponding to the second base station;
and a third reporting period.
Wherein the current feature input information list may include at least one of:
inputting information by the current required characteristics of the second terminal in the range of the second base station;
the second base station inputs information on the current required characteristics;
the second adjacent base station inputs information according to the current required characteristics;
wherein the model reasoning information list information element comprises: and inputting information of the current required characteristics of the second cell corresponding to the second base station.
Wherein:
when the feature input information registration request information element is set to be started, the second base station is instructed to start measurement according to the current feature input information list;
when the feature input information registration request information element is set to stop, the second base station is instructed to stop measurement and reporting;
and when the characteristic input information registration request information element is set to be added, the second base station is instructed to start measurement according to the model reasoning information list information element.
In an exemplary embodiment, the second terminal currently required feature input information may include at least one of:
The current position information of the second terminal;
the second terminal current historical mobility information;
the second terminal currently wirelessly measures information.
Wherein the currently required feature input information of the second base station may include at least one of the following:
the second base station measures the information wirelessly at present;
the second current terminal flow;
the current required feature input information of the second neighboring base station comprises at least one of the following:
the second adjacent base station measures the information wirelessly at present;
and unloading the flow to second terminal current performance measurement information of a second neighbor cell corresponding to a second neighbor base station of the second base station by the second terminal in the range of the second base station.
Wherein the currently required feature input information of each second cell may comprise at least one of:
a predicted radio resource status of the second cell;
predicted terminal traffic for the second cell;
and predicting the wireless resource state of the adjacent base station of the second cell.
In an exemplary embodiment, the model reasoning data reply message may include at least one of:
a second base station network address and a port address;
a second current message type;
the first base station measures the identification;
the second base station measures the identification;
A third critical diagnostic information element, the third critical diagnostic information element being used to indicate an unintelligible or missing information element in the model reasoning data request message received by the second base station.
In an exemplary embodiment, it may further include:
and receiving a model reasoning data failure message sent by the second base station.
Wherein the model reasoning data failure message may include at least one of:
a third current message type;
the first base station measures the identification;
the second base station measures the identification;
a second base station network address and a port address;
a second failure cause;
a fourth critical diagnostic information element, the fourth critical diagnostic information element being used to indicate an unintelligible or missing information element in the model reasoning data request message received by the second base station;
and waiting for the retransmission time.
In an exemplary embodiment, the second model reasoning data may include at least one of:
a fourth current message type;
the second base station measures the identification;
a second base station network address and a port address;
a second cell current measurement result of a second cell corresponding to the second base station;
The model inferences the output valid time.
In an exemplary embodiment, the second cell current measurement result may include a second cell current measurement item.
The second cell current measurement item may include a second cell identifier and second cell current feature input information reporting, and the second cell current feature input information may report current feature input information indicating that a second cell of the second base station needs to report.
The current characteristic input information of each second cell may include at least one of:
the current characteristics of the second terminal are input with information;
the second base station inputs information according to the current characteristics;
and the current characteristics of the second adjacent base station input information.
In an exemplary embodiment, the second terminal current feature input information may include at least one of:
the current position information of the second terminal;
the second terminal current historical mobility information;
the second terminal is in wireless measurement information at present;
a first current timestamp.
Wherein the second base station current feature input information indicates feature input information of second cells of the second base station, and the feature input information of each second cell may include at least one of the following:
the second base station wirelessly measures information;
A predicted radio resource status of the second cell;
predicted terminal traffic for the second cell;
predicting the wireless resource state of the neighboring base station of the second cell;
a second current timestamp.
The current feature input information of the second neighboring base station indicates feature input information of a second neighboring cell corresponding to the second neighboring base station of the second base station, and the feature input information of each second neighboring cell may include at least one of the following:
the current/predicted radio resource status of the second neighbor cell;
predicting terminal flow of a second neighbor cell;
unloading the flow to the current performance measurement information of the second terminal of the second neighbor cell corresponding to the second neighbor base station by the second terminal in the range of the second base station;
and a third current timestamp.
In an exemplary embodiment, the model inference output information may include at least one of:
load balancing a target cell list;
predicting the radio resource state of a target cell;
predicting a list of congested cells;
predicting a congestion relief cell list;
predicting a terminal migrated to a target cell;
switching mechanism related parameters.
In an exemplary embodiment, it may further include:
receiving a measurement feedback message sent by the second base station after the second base station executes corresponding network load balancing operation;
Optimizing the network load balancing model based on the measurement feedback message;
wherein the feedback measurement information includes at least one of:
terminal measurement information of the target base station;
resource status information of the target base station;
key performance index of system.
In an exemplary embodiment, it may further include:
receiving an AI/ML model training pause message sent by a network management system;
the AI/ML model training pause message includes AI/ML load balancing stop indication information.
In an exemplary embodiment, the AI/ML model training pause message may further include cause indication information;
the cause indication information includes at least one of:
network resources are saved;
the terminal saves energy.
The method for load balancing provided by the embodiment of the disclosure is a distributed inter-base station information cooperation method, and relates to the technical field of wireless communication.
The integration of the AI technology and the wireless network has become the necessity of wireless communication development, the wireless network utilizes the artificial intelligence technology to better cope with more complex network architecture and various communication scenes, and the AI algorithm can classify, count and infer data based on massive data generated by the complex network, so as to further give out the conclusions of analysis, prediction, recommendation and the like.
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 the prediction of user tracks, the prediction of user services and other information, the network performance is ensured, high-quality service experience can be provided for users, the system capacity is further improved, and the task of human intervention network management and optimization can be reduced to the greatest extent.
The intelligent universal architecture of the wireless network is shown in fig. 3, in which relevant functional bodies and data/information flows are defined.
Data Collection function (Data Collection) 310: may be used to provide input Data to Model Training function (Model Training) 320 and Model Inference function (Model Inference) 330, including Training Data (Training Data, also referred to as Model Training Data) provided to Model Training function 320, and Inference Data (Model Inference Data) provided to Model Inference function 330, which may include measurements and/or feedback from terminals and/or different network entities.
Model training function 320: may be used to perform training, validation and testing functions of an AI/ML model (including AI/ML based network load balancing models) based on received training data, model performance metrics can be generated as part of a model testing process. Model training function 320 may also be responsible for data preprocessing, cleansing, formatting, and conversion based on training data provided by data collection function 310. Model training function 320 may also deploy/Update the trained Model to Model reasoning function 330.
Model reasoning function 330: may be used to provide AI/ML model inference outputs based on the model trained by the model training functionality 320, which may contain prediction and decision information for judgment and execution by the nodes of the network. In addition, the model inference function 330 also has the capability to process data. Optionally, the model inference function 330 may also feed back (Model Performance Feedback) the model performance to the model training function 320.
An executive function (Actor) 340: may be used to receive the Output (Output) of the model inference function 330 and trigger or perform a corresponding action and send Feedback (Feedback) to the data collection function 310.
The network load balancing decision in the related art depends on the report of the current/past cell load state, but as the network flow increases, the network load and the resource state change are accelerated, and for high mobility and large connection scenes, the more passive traditional mode may cause the problems of ping-pong handover between different cells, cell overload, user service quality degradation and the like, so that the whole network service quality is difficult to ensure.
In addition, the network load balancing scheme in the related art is based on measurement reporting of users, so that a large amount of signaling overhead is generated on an air interface, and meanwhile, a huge power consumption problem is caused for a terminal.
Therefore, there is a need for a solution that is capable of actively generating load balancing decisions through load traffic prediction independent of network measurement reporting.
Aiming at the problems in the related art, the embodiment of the disclosure provides an AI/ML-based distributed network load balancing scheme, which enables a base station to perform model training and model reasoning based on measured and collected characteristic input information, and actively adopts a corresponding load balancing strategy through the prediction of information such as user service, track and the like by the AI/ML technology, thereby solving the problems of overlarge air interface resource overhead and the like caused by the network management and optimization based on measurement reporting by the method adopted in the related art. Meanwhile, the realization of the base station distributed model training and reasoning process avoids unnecessary signaling interaction between the base station and the network manager, is beneficial to reducing the data interruption of operation service and improving the network load balancing speed.
The embodiment of the disclosure realizes a distributed network load balancing scheme based on AI/ML, enables a base station to utilize the processing capacity of AI/ML technology to big data, performs model training and model reasoning based on a data set formed by prior data such as network wireless resource state, cell load, UE position information and the like, predicts information such as user service and track and the like through an algorithm, actively adopts a corresponding load balancing strategy, is beneficial to a network to take expected measures to prevent network user experience from being reduced, and ensures network service quality.
The embodiment of the disclosure provides a network distributed load balancing method based on AI/ML, which is characterized in that a network management side sets an anchor base station and a load balancing optimization range according to factors such as the load condition of each base station, the calculation power and the storage capacity of the base station and the like, performs offline model training, and deploys an offline trained model to each node. The base station in the embodiments of the present disclosure may be referred to as a node, or may be referred to as a base station node.
The anchor base station can collect related data of the neighbor base station node AI/ML model training, perform online model training by utilizing self measurement data and collected information of the neighbor base station about wireless resource state, cell load, switching history performance data, UE position and the like, and update and deploy the trained model to the base station in an optimization range.
The anchor base station can also be responsible for continuously collecting model reasoning related data, outputting reasoning analysis comprising prediction information such as network load, terminal track prediction and the like, decision information about terminal switching and the like, and triggering the network to execute corresponding operations.
And the anchor base station further trains the model according to the network performance feedback information of each base station node after executing the operation, thereby improving the accuracy of the model and realizing the load balance of the distributed intelligent network.
As shown in fig. 4, the method provided by the embodiment of the present disclosure may be used to implement AI/ML-based distributed network load balancing, where both a model training function and a model reasoning function are deployed at a first base station, and specifically implement a signaling flow as follows:
s1, OAM determines an anchor base station.
OAM (network management/network management system) triggers a network load balancing mechanism based on AI/ML (advanced technology attachment/management) aiming at the problems of unbalanced access quantity of a base station cell terminal, frequent ping-pong switching of the terminal and the like in a network so as to improve network performance.
The OAM calls a preconfigured load balancing use case model (namely the network load balancing model based on the AI/ML), determines an effective area range (namely the load balancing area range) needing intelligent load balancing operation according to measurement report information, geographical information and the like sent by a base station, determines the first base station as an anchor base station in the effective area range according to evaluation indexes such as calculation power, storage capacity and the like of the base stations (comprising the first base station and the second base station) in the effective area range, and is responsible for collecting data required by model training related to adjacent base stations.
The measurement report information may include measurement information reported by each base station and geographical information of each base station, where the measurement report information aims at the problems of uneven load, ping-pong handover, etc. existing in the network.
For example, if the OAM detects that there is serious unevenness in the cell traffic corresponding to a certain base station, the certain cell is particularly busy, and the certain cell is particularly idle, then the base station has a problem of uneven load. The neighboring base station/neighboring base station of the base station or the base station having the same problem as the base station is within the optimization range of the load balancing, i.e. divided into the load balancing area range by the OAM.
For another example, the OAM may select a base station with the best calculation power, storage capacity, and the like in the load balancing area as the anchor base station, but the disclosure is not limited thereto.
In the embodiment of the present disclosure, the OAM may further use one or more base stations in the load balancing area range other than the first base station as the auxiliary base station in the load balancing area range, which is called a second base station or a second base station node.
In the embodiment of the present disclosure, the second base station node may be any base station node except for an anchor base station in the effective area, where a neighboring base station or a neighboring base station node of the first base station refers to a base station node neighboring the first base station, and for distinguishing, the neighboring base station or the first neighboring base station node will be referred to as a first neighboring base station or a first neighboring base station node. The second base station node may be wider than the first base station's neighbor base station node, and is not limited to the first base station's neighbor base station node, and may include the first base station's neighbor base station, and the second base station node may function to provide the first base station node/anchor base station with data required for model training and/or model reasoning.
In the following embodiments, for the sake of distinction, a terminal in the range of the first base station is referred to as a first terminal, a terminal in the range of the second base station is referred to as a second terminal, a neighboring base station of the second base station is referred to as a second neighboring base station or a second neighboring base station node, one or more cells corresponding to the first base station are referred to as a first cell, one or more cells corresponding to the first neighboring base station are referred to as a first neighboring cell, and one or more cells corresponding to the second neighboring base station are referred to as a second neighboring cell.
And S2, performing model training (offline) by OAM.
The OAM performs offline training on a preconfigured model (i.e. the preconfigured network load balancing model) based on historical data, wherein the historical data is a wireless measurement related historical data set (i.e. the wireless measurement historical data set) reported by a base station in the effective area range of OAM storage, and the wireless measurement historical data set is divided into a training data set, a verification data set and a test data set through data processing and dividing processes. The OAM carries out training of the network load balancing model based on the training data set, the verification data set and the test data set, and determines the related information such as the learning algorithm, parameters, required characteristic input and the like of the network load balancing model.
S3a.oam deploys the model to the first base station.
According to the network load balancing model obtained by offline training, the OAM transmits a model deployment message (for distinction, referred to herein as a first model deployment message) to a first base station (e.g., NG-RAN node 1), where the first model deployment message contains model-related information, such as feature input information for training, to instruct AI/ML model-related configuration information, instruct the base station (here, the first base station, that is, an anchor base station) to start measurement and collection of related feature input information (such as first model training data and second model training data), and instruct the first base station to collect these information for training of the network load balancing model described above.
Wherein the first model deployment message may include, but is not limited to, at least one of the following information:
-anchor base station indication information: the boolean or enumeration type is configured to instruct a base station (herein referred to as a first base station or a receiving base station) that receives the first model deployment message to be an anchor base station, and is responsible for collecting and counting measurement report information of each base station node (where each base station node includes a second base station node and/or a first neighboring base station node and/or a second neighboring base station node), where measurement report information of the second base station node includes measurement report information of the second neighboring base station node, and the second model training data includes measurement report information of the second base station node. The measurement report information of the first neighboring base station node is included in the first model training data.
List of secondary base station IDs (identities) within the load balancing area: one or more base station IDs may be included to indicate valid secondary base stations, i.e., second base stations, within the load balancing area to provide data required for model training and reasoning, where the data required for model training provided by the secondary base stations is referred to as second model training data, and the data required for model reasoning provided by the secondary base stations is referred to as second model reasoning data, which also facilitates subsequent model updating operations for the relevant base stations (secondary base stations/second base stations).
-a model index for indicating applicable use cases of the deployed load balancing model and algorithms used by the load balancing model, which may comprise at least one of:
model use case: the enumeration type may include, for example, at least a load prediction model, a terminal track prediction model, and the like, for indicating various use cases required for realizing the target use case.
Model class: enumeration, for example, may include at least linear regression, logistic regression, decision trees, support vector machines, random forests, etc., for indicating an adapted model of the relevant use case.
In the embodiment of the disclosure, the target use case refers to a total load balancing model, where the total load balancing model may include one or more various use cases such as a load prediction model and a terminal track prediction model, and the model class is used to indicate what kind of adaptation model is adopted by the relevant use case, for example, the load prediction model adopts a linear regression algorithm through the model class, and the terminal track prediction model adopts a logistic regression algorithm, but the disclosure is not limited thereto.
Model parameters: configuration variables inside the model, carried in a model deployment message (including a first model deployment message and may further include a second model deployment message) or a model update message, are used to update model parameters, thereby improving model training accuracy and confidence, and the contained information includes, but is not limited to, at least one of the following:
weight- -weight
Offset
-learning rate
Number of iterations
The SVM support vector is a parameter that may be used, and is determined according to different algorithms used.
Feature input information list (for distinction, herein referred to as first feature input information list): for instructing a base station (referred to herein as a first base station) to collect one or more characteristic inputs required for model training, the required information including, but not limited to, at least one of:
UE feature input information, since UE herein refers to UE within range of the first base station, i.e. the first terminal, may also be referred to as feature input information of the first terminal within range of the first base station, including but not limited to at least one of the following:
UE location information, which may also be referred to herein as first terminal location information, since the UE herein is the first terminal
UE historical mobility information, which may also be referred to herein as the first terminal's historical mobility information, since the UE here is the first terminal
UE radio measurement information: since the UE is the first terminal, the radio measurement information, which may also be referred to herein as the first terminal, may include one or more of RSRP, RSRQ, etc.
-base station characteristic input information, which may also be referred to as first base station since the base station herein refers to the first base station, for instructing the first base station to turn on radio measurement information of the first base station and measurement of traffic of UEs connected to the first base station node, including but not limited to at least one of the following:
base station radio measurement information: the base station herein refers to the first base station, and thus may also be referred to as radio measurement information of the first base station, including but not limited to a cell PRB utilization, an average RRC connection number, a packet loss rate, and the like.
Current terminal traffic, herein referred to as traffic of the terminal currently connected to the first base station, and may thus also be referred to as current terminal traffic of the first base station
-neighbor base station characteristic input information, here, the neighbor base station refers to the first neighbor base station, and thus, the first neighbor base station characteristic input information, which may also be referred to as the first base station, includes, but is not limited to:
Neighboring base station radio measurement information, wherein the neighboring base station refers to the first neighboring base station, and thus may also be referred to as radio measurement information of the first neighboring base station
-traffic offload to UE performance measurements of neighbor cells: terminal related information for indicating traffic offload to a neighboring cell, comprising at least one or more or a combination of the following information: qoS parameter information such as UE position information, packet loss rate, time delay and the like, UE wireless measurement information and the like. Here, the UE performance measurement of offloading traffic to a neighboring cell refers to offloading traffic to, by a first terminal within a range of a first base station, first terminal performance measurement information related to the first terminal and corresponding to the first neighboring cell of the first neighboring base station.
-a feature input information request IE, for distinction, referred to as a first feature input information request information element
If the feature input information request IE is set to "start", then the method is used to instruct the base station (here, the first base station) that measurement should be started according to the instruction in the feature input information list (i.e., the first feature input information list); or alternatively
If the feature input information request IE is set to "stop", then it is used to indicate that the base station (here, the first base station) should stop measurement and reporting; or alternatively
If the feature input information request IE is set to "add", it is used to instruct the base station (here, referred to as the first base station) that the measurement information or the prediction information indicated in the "feature input information addition IE (i.e., the first feature input information addition information element described below)" should be added to the measurement initiated by the previously given feature input information list (i.e., the first feature input information list) to obtain the corresponding measurement value or prediction value. If measurement has been initiated on the measurement information or prediction information indicated in the "feature input information addition IE (i.e. the first feature input information addition information element described below)", this information will be ignored.
Feature input information addition IE (for distinction, referred to as first feature input information addition information element): measurement information or prediction information other than the feature input information (i.e., the first feature input information list) indicating that a new addition is required.
And S3b, the OAM deploys the model to the second base station.
The OAM transmits a model deployment message (for distinction, referred to herein as a second model deployment message) to a second base station (e.g., NG-RAN node 2) according to the network load balancing model obtained by offline training, for indicating AI/ML model-related configuration information and indicating the second base station to turn on related measurements, wherein the indication information includes, but is not limited to, at least one of the following:
-a model index for indicating applicable use cases for deploying the model and algorithms used by the model:
model use case: the enumeration type may include, for example, at least a load prediction model, a terminal track prediction model, and the like, for indicating various use cases required for realizing the target use case.
Model class: enumeration, for example, may include at least linear regression, logistic regression, decision trees, support vector machines, random forests, etc., for indicating an adapted model of the relevant use case.
Model parameters: configuration variables inside the model, carried in a model deployment message or a model update message, are used for updating model parameters, thereby improving model training accuracy and confidence, and the contained information includes but is not limited to at least one of the following:
weight- -weight
Offset
-learning rate
Number of iterations
Support vector of SVM
Feature input information list (for distinction, herein referred to as third feature input information list): for instructing the base station (herein referred to as the second base station) to collect one or more characteristic inputs required for model training, the required information including, but not limited to, at least one of:
UE feature input information, here UE refers to the second terminal in the range of the second base station, and thus may also be referred to as feature input information of the second terminal in the range of the second base station, including but not limited to at least one of the following:
UE location information, where UE refers to the second terminal and may therefore be referred to as location information of the second terminal
UE history mobility information, where UE refers to the second terminal, and thus may be referred to as UE history mobility information of the second terminal
UE radio measurement information: the UE herein refers to the second terminal, and thus may be referred to as radio measurement information of the second terminal, may include one or more of RSRP, RSRQ, and the like.
-base station characteristic input information, herein the base station is referred to as a second base station, and thus may be referred to as second base station characteristic input information, which may include, but is not limited to, at least one of the following:
base station radio measurement information: the base station refers to the second base station, and thus may be referred to as radio measurement information of the second base station, and may include, but is not limited to, PRB utilization rate of a cell (where the cell refers to a second cell corresponding to the second base station), average RRC connection number, packet loss rate, and the like
Current terminal traffic, herein referred to as traffic of the terminal connected to the second base station, and may thus be referred to as current terminal traffic of the second base station
-neighbor base station characteristic input information, here, the neighbor base station refers to the second neighbor base station, and thus may be referred to as second neighbor base station characteristic input information, which may include, but is not limited to, at least one of the following:
Neighbor station radio measurement information, wherein the neighbor station refers to the second neighbor station, and thus can be referred to as radio measurement information of the second neighbor station
-traffic offload to UE performance measurements of neighbor cells: terminal related information indicating traffic offload to neighboring cells, comprising at least one or more or a combination of the following information: qoS parameter information such as UE position information, packet loss rate, time delay and the like, UE wireless measurement information and the like. Here, the UE performance measurement of offloading traffic to a neighboring cell refers to offloading traffic to, by a second terminal within a range of a second base station, second terminal performance measurement information related to the second terminal and corresponding to the second neighboring cell of the second neighboring base station.
Feature input information request IE (for distinction, herein referred to as third feature input information request information element)
If the third characteristic input information request IE is set to "start", the base station (here, the second base station) should start measurement according to the indication in the above third characteristic input information list; or alternatively
If the third characteristic input information request IE is set to "stop", the base station (here, the second base station) should stop measurement and reporting; or alternatively
If the third characteristic input information request IE is set to "add", the base station (referred to herein as the second base station) should add the measurement information or prediction information indicated in the "third characteristic input information add IE" to the previously given measurement initiated by the third characteristic input information list to obtain the corresponding measurement value or prediction value. If measurement has been initiated with the measurement information or prediction information indicated in the "third feature input information addition IE", this information will be ignored.
-feature input information addition IE: for distinction, an information element is added to the third characteristic input information, which indicates that measurement information or prediction information other than the above-described third characteristic input information list needs to be newly added.
It should be noted that S3b is model deployment performed by OAM to the second base station node, where the second base station has some model training capability, may perform simple prediction based on data, and may provide the required prediction information to the second base station according to the instruction of the first base station in the subsequent process.
S4, the first base station performs measurement configuration on the terminal.
A first base station (for example, NG-RAN node 1) receives a first model deployment message from the OAM, determines itself as an anchor base station based on the first model deployment message, invokes a preconfigured network load balancing model and updates model parameters according to model related indication information carried in the first model deployment message, and collects related data of a terminal (here, the first terminal) according to characteristic input information required by a use case model.
And the first base station performs measurement configuration on the terminals (namely the first terminals) in the range of the first base station according to the model related indication information carried in the first model deployment message, for example, sends a measurement configuration message to the first terminals, wherein part of position information and the like adopt MDT measurement configuration.
Wherein the information contained in the measurement configuration message may include, but is not limited to, at least one of the following:
UE historical mobility information, here UE refers to the first terminal and may therefore be referred to as the first terminal's historical mobility information;
RRM measurement configuration, here UE refers to the first terminal, and thus may be referred to as radio resource management measurement configuration information of the first terminal, may include, but is not limited to, at least one of:
supporting periodic measurement triggers: may include a trigger period and a recording period
UE radio measurement information: the UE herein refers to the first terminal, and thus may be referred to as radio measurement information of the first terminal, may include one or more of RSRP, RSRQ, etc
-MDT measurement configuration, here UE refers to the first terminal, and thus may be referred to as minimization of drive test measurement configuration information of the first terminal, which may include, but is not limited to, at least one of the following:
supporting periodic measurement triggers: the method comprises a triggering period and a recording period;
UE location information, here UE is the first terminal and may therefore be referred to as the location information of the first terminal;
UE movement speed, where UE is the first terminal, and thus may be referred to as the first terminal movement speed
S5, the terminal performs UE measurement.
The terminal (here, first terminal) performs corresponding measurement according to the configuration information carried in the measurement configuration message sent by the first base station (for example, NG-RAN node 1), collects indicated measurement information, such as RSRP, RSRQ, SINR measurement value of the UE, and the like, and completes recording and collection of the UE position information, the UE moving speed, and the like according to MDT measurement, so as to serve as first terminal model training data, where the first model training data includes first terminal model training data.
S6, the terminal reports UE measurement to the first base station.
The terminal (here, first terminal) transmits measurement information such as historical mobility information, location information, RSRP, RSRQ and the like required for model training to the first base station (for distinguishing, referred to as a first terminal sample measurement report message herein) through a UE measurement report message (for example, NG-RAN node 1), where the UE measurement report message includes the first terminal model training data, and the information included in the message may include, but is not limited to, at least one of the following:
UE historical mobility information, here UE refers to the first terminal and may therefore be referred to as the first terminal's historical mobility information
RRM measurement configuration, here UE refers to the first terminal, and thus may be referred to as radio resource management measurement configuration information of the first terminal, may include, but is not limited to, at least one of:
UE radio measurement information: the UE herein refers to the first terminal, and thus may be referred to as radio measurement information of the first terminal, may include one or more of RSRP, RSRQ, etc
-time stamp, herein means time stamp for obtaining the radio resource management measurement configuration information of the first terminal
-MDT measurement configuration, here UE refers to the first terminal, and thus may be referred to as minimization of drive test measurement configuration information of the first terminal, which may include, but is not limited to, at least one of the following:
UE location information, here UE is the first terminal and may therefore be referred to as the location information of the first terminal;
UE movement speed, where UE is the first terminal, and thus may be referred to as the first terminal movement speed
-a time stamp, here a time stamp for obtaining the minimization of drive tests configuration information of the first terminal
S7, the first base station sends a model training data request message to the second base station.
The first base station (e.g., NG-RAN node 1) receives and processes measurement report data from one or more terminals (first terminals) as first terminal model training data, and starts measurement according to feature input information indicated in the first model deployment message to obtain first base station model training data, the first model training data including the measured first base station model training data, and performs statistics of the measurement data.
In addition, the first base station (e.g., NG-RAN node 1) sends a model training data request message to the second base station (e.g., NG-RAN node 2) instructing the second base station to start, stop or add a measurement procedure based on the first model deployment message, based on the characteristic input information indicated in the first model deployment message.
The model training data request message may be fed back through the control plane and/or the user plane, and the information contained in the model training data request message may include, but is not limited to, at least one of the following:
message types, called first sample message types for distinction
-first base station measurement ID
-second base station measurement ID
-first base station IP address and port address (also referred to as network address information and port information of the first base station): if the model training data request message contains the IE of the IP address and the port address of the first base station, the second base station is instructed to feed back data through the user interface;
reporting cell list, for distinction, referred to as sample reporting cell list, may include, but is not limited to:
cell reporting information: the cells herein refer to the second cells corresponding to the second base station, and may include, but are not limited to, cell IDs (the cells herein refer to all cells corresponding to the second base station node, and may be referred to as second cells for distinction, the number of the second cells may be one or more), SSB report lists, SSB indexes, and the like.
Reporting period: for distinction, it may be referred to as a first reporting period, which may be used to indicate the average window length of all measurement objects.
-a list of required feature input information indicating the feature input information of one or more cells (herein referred to as second cells) of the base station (herein referred to as second base station) required for model training, the feature input information of each cell including, but not limited to, at least one of:
UE feature input information, here UE means the second terminal, which is used for model training, and thus may be referred to as sample required feature input information of the second terminal, which may include, but is not limited to, at least one of the following:
UE location information, here UE means the second terminal, which is used for model training and may therefore be referred to as second terminal sample location information
UE historical mobility information, here UE means the second terminal, which is used for model training and thus may be referred to as second terminal sample historical mobility information
UE radio measurement information: the UE herein refers to a second terminal, which is used for model training, and thus may be referred to as second terminal sample radio measurement information, which may include one or more of RSRP, RSRQ, etc.
-base station characteristic input information, herein referred to as second base station, and used for model training, and thus may be referred to as second base station sample required characteristic input information, which may include, but is not limited to, at least one of:
base station radio measurement information: the base station refers to the second base station and is used for model training, so that the second base station sample wireless measurement information can be referred to as second base station sample wireless measurement information, and can include, but is not limited to, cell PRB utilization, average RRC connection number, packet loss rate, and the like
Current terminal traffic, here the traffic of the terminal connected to the second base station, here the terminal is the second terminal and used for model training, thus can be referred to as second sample terminal traffic
-neighbor characteristic input information, here, the neighbor refers to the second neighbor and is used for model training, so that the characteristic input information required for the second neighbor sample may be referred to as the second neighbor sample, and may include, but is not limited to, at least one of the following:
neighbor cell radio measurement information, wherein the neighbor cell refers to the second neighbor cell and is used for model training, and thus can be referred to as second neighbor cell sample radio measurement information
-traffic offload to UE performance measurements of neighbor cells: terminal related information indicating traffic offload to neighboring cells, comprising at least one or more or a combination of the following information: qoS parameter information such as UE position information, packet loss rate, time delay and the like, UE wireless measurement information and the like. Here, the UE performance measurement for offloading traffic to the neighboring cell refers to offloading traffic to second terminal sample performance measurement information of a second neighboring cell corresponding to the second neighboring cell by the second terminal within the range of the second base station.
-adding a model training information list IE: model training required data for indicating new addition of required feature input information list
-required feature input information registration request IE:
if the required feature input information registration request IE is set to "start", the receiving base station (here referred to as the second base station) should initiate measurements according to the indication in the required feature input information list; or alternatively
If the required feature input information registration request IE is set to "stop", the receiving base station (here, the second base station) should stop measurement and reporting; or alternatively
If the required feature input information registration request IE is set to "add", the base station (referred to herein as the second base station) should add the measured or predicted value indicated in the "add model training information list IE" to the previously given required feature input information list initiated measurement. If a measurement has been initiated on the information indicated in the "Add model training information List IE", this information will be ignored.
A reporting feature, herein referred to as a sample reporting feature, for distinction, to indicate the measurement object reported by the second base station requested for each position in the bitmap if the required feature input information registration request IE indicates "start".
It should be noted that, the bitmap is carried in the model training data request message, that is, the bitmap is used to indicate the reporting feature. For example, assuming that the bitmap has 3 bits in total, if the 1 st bit is 1, the reported measurement object is indicated to include UE feature input information, and if the value is 0, the reported measurement object is indicated to not include UE feature input information. For another example, if the 2 nd bit is 1, the reported measurement object is indicated to include the base station feature input information, and if the value is 0, the reported measurement object is indicated to not include the base station feature input information. For another example, if the 3 rd bit is 1, the reported measurement object is indicated to include the neighbor base station feature input information, and if the value is 0, the reported measurement object is indicated to not include the neighbor base station feature input information. The foregoing is for illustration only, and the disclosure is not limited thereto.
In an embodiment of the present disclosure, the model training data request message may further include network address information and port information of the first base station, so as to establish a channel for transmitting AI/ML data between the first base station and the second base station, where the AI/ML data refers to any relevant data for training an AI/ML-based network load balancing model.
Optionally, s8a, the second base station sends a model training data reply message to the first base station.
The second base station (e.g., NG-RAN node 2) receives the model training data request message sent from the first base station (e.g., NG-RAN node 1) and indicates, through a model training data reply message, measurement object information for which measurement objects requested by the model training data request message can be successfully started, where the information included in the model training data reply message includes at least one of the following:
-a second base station IP address and port address (also referred to as network address information and port information of the second base station): if the model training data reply message contains the IE of the IP address and the port address of the second base station, the second base station is instructed to perform data feedback through a user interface;
message types, called second sample message types for differentiation
-first base station measurement ID
-second base station measurement ID
Critical diagnostic IE: for distinction, referred to as the first critical diagnostic information element, the model training data reply message carries this critical diagnostic IE to indicate which IEs are not understood or lost in the received model training data request message.
Optionally, s8b, the second base station sends a model training data failure message to the first base station.
If the second base station is unable to provide any requested measurements from the first base station, the second base station may send a model training data failure message with appropriate cause values to the first base station, including but not limited to at least one of the following:
message types, called third sample message types for differentiation
-first base station measurement ID
-second base station measurement ID
Failure cause: for distinguishing, the first failure reason is the reason of the XnAP protocol specific event, for example, at least any one of the wireless network layer reason, the transport layer reason, the protocol reason, etc. may be included;
-critical diagnostics: for distinction, referred to as a second critical diagnostic information element, when the received model training data request message part is not understood or lost, or the model training data request message contains a logical error, the IE is used for indicating which IEs in the model training data request message are not understood or lost;
waiting for a retransmission time, called first waiting for retransmission time, for distinguishing, for indicating the time of reinitiation of the request (i.e. the first base station re-sends the model training data request message to the second base station).
Optionally, S9, the second base station sends a model training data reporting message to the first base station.
In the embodiment of the disclosure, the model training data report message may include network address information and port information of the second base station, so as to establish a channel for transmitting AI/ML data between the first base station and the second base station.
The second base station (e.g., NG-RAN node 2) sends a model training data reporting message to the first base station (e.g., NG-RAN node 1) for reporting measurement information requested by the first base station, i.e., the second model training data, including but not limited to at least one of the following:
message type, for differentiation, referred to as fourth sample message type
-first base station measurement ID
-second base station measurement ID
Cell measurement, here referred to as cell measurement of a second cell corresponding to the second base station:
-cell measurement items, which may include, but are not limited to, at least one of:
cell ID, herein referred to as the cell ID of the second cell
-a list of required feature input information including measurement information configured in a model deployment message/model update message and measurement information required for newly added model training, which may include, but is not limited to, at least one of:
UE feature input information, herein UE refers to the second terminal and is used for model training, and may be referred to as second terminal sample feature input information, which may include, but is not limited to, at least one of the following:
UE location information, where UE refers to the second terminal and is used for model training, may be referred to as second terminal sample location information
UE historical mobility information, here UE refers to the second terminal and is used for model training, which may be referred to as second terminal sample historical mobility information
-UE radio measurement information: the UE herein refers to a second terminal, and is used for model training, and may be referred to as second terminal sample radio measurement information, which may include one or more of RSRP, RSRQ, etc
A time stamp, which may be referred to as a first sample time stamp for distinguishing, i.e. a time stamp for obtaining the input information of the second terminal sample feature
-base station characteristic input information, herein base station refers to a second base station, and for model training, may be referred to as second base station sample characteristic input information, which may include, but is not limited to, at least one of the following:
-base station radio measurement information: the base station refers to the second base station and is used for model training, and may be referred to as second base station sample wireless measurement information, which may include, but is not limited to, cell (herein referred to as second cell) PRB utilization, average RRC connection number, packet loss rate, etc
Current terminal traffic, herein referred to as traffic of the terminal connected to the second base station, and used for model training, may be referred to as second sample terminal traffic
A time stamp, which for distinguishing may be referred to as a second sample time stamp, i.e. a time stamp for obtaining the input information of the second base station sample characteristics
-neighbor (second neighbor) feature input information, wherein the neighbor refers to the second neighbor and is used for model training, and may be referred to as second neighbor sample feature input information, including but not limited to at least one of the following:
neighbor cell radio measurement information, wherein the neighbor cell refers to a second neighbor cell and is used for model training, and can be called as second neighbor cell sample radio measurement information
-traffic offload to UE performance measurements of neighbor cells: terminal related information indicating traffic offload to neighboring cells, comprising at least one or more or a combination of the following information: qoS parameter information such as UE position information, packet loss rate, time delay and the like, UE wireless measurement information and the like. Here, it means that the second terminal in the range of the second base station offloads the traffic to the second terminal sample performance measurement information of the second neighboring cell corresponding to the second neighboring base station
The time stamp may be referred to as a third sample time stamp for distinguishing, i.e. the time stamp of acquiring the sample characteristic input information of the second neighboring base station.
S10, the first base station performs model training.
The first base station receives data information counted by neighbor base stations (here, first neighbor base stations) in a load balance optimization range (namely, a load balance area range), receives first terminal model training data reported by a first terminal in the first base station range, and forms first model training data through the first base station model training data obtained through self measurement.
The first base station performs online model training based on the training data set, the verification data set and the test data set, and performs parameter adjustment and feature optimization through model training, verification and test processes so as to obtain a more accurate network load balancing model.
S11, the first base station deploys/updates the network load balancing model after online training to the second base station.
The first base station sends a model update message to the second base station to update the network load balancing model deployed on the second base station by using the network load balancing model trained by the first base station.
The first base station updates the trained AI/ML-based network load balancing model to the second base station to indicate the latest relevant configuration information of the AI/ML-based network load balancing model, wherein the information contained in the model update message comprises at least one of the following information:
-a list of feature input information, for the sake of distinction, referred to as a second list of feature input information:
UE feature input information, here the UE is a second terminal, which may be referred to as second terminal feature input information, including but not limited to at least one of:
UE location information, where UE means the second terminal, may be referred to as second terminal location information
UE historical mobility information, herein UE refers to the second terminal, and may be referred to as second terminal historical mobility information
UE radio measurement information: the UE herein refers to the second terminal, which may be referred to as second terminal radio measurement information, and may include one or more of RSRP, RSRQ, and the like.
-base station characteristic input information, herein referred to as second base station, which may be referred to as second base station characteristic input information, including but not limited to at least one of the following:
base station radio measurement information: the base station refers to the second base station, which may be referred to as second base station radio measurement information, and may include, but is not limited to, cell (herein referred to as second cell) PRB utilization, average RRC connection number, packet loss rate, etc
Current terminal traffic, where a terminal refers to a terminal connected to a second base station, may be referred to as a second current terminal traffic
-neighbor base station characteristic input information, herein, the neighbor base station refers to a second neighbor base station, which may be referred to as second neighbor base station characteristic input information, including but not limited to at least one of the following:
neighboring base station wireless measurement information, wherein the neighboring base station refers to a second neighboring base station, which can be called second neighboring base station wireless measurement information
-traffic offload to UE performance measurements of neighbor cells: terminal related information indicating traffic offload to neighboring cells, comprising at least one or more or a combination of the following information: qoS parameter information such as UE position information, packet loss rate, time delay and the like, UE wireless measurement information and the like. Here, it means that the second terminal in the range of the second base station offloads the traffic to the second terminal performance measurement information of the second neighboring cell corresponding to the second neighboring base station
-a feature input information request IE, for distinction, referred to as a second feature input information request information element
If the second characteristic input information request IE is set to "start", the receiving base station (here referred to as the second base station) should initiate measurement according to the indication in the second characteristic input information list; or alternatively
If the second characteristic input information request IE is set to "stop", the receiving base station (here, the second base station) should stop the measurement and reporting; or alternatively
If the second characteristic input information request IE is set to "add", the receiving base station (referred to herein as the second base station) should add the measured or predicted value indicated in the "second characteristic input information add IE" to the previously given second characteristic input information list-initiated measurement. If the measurement has been initiated with the information indicated in the "second feature input information addition IE", this information will be ignored.
A reporting feature, called first reporting feature, for distinction, to indicate the measurement object requested by the second base station for each position in the bitmap if the second feature input information request IE indicates "start".
Reporting period, called second reporting period for distinction, reporting period indicating periodic measurement
-feature input information addition IE: for distinction, information elements are added, referred to as second feature input information, indicating newly added desired feature input measures
Model use case modification IE
If the model update message carries the model use case change IE, the second base station shall change the model use case according to the indication of the model use case change IE, and increase or decrease the call of the use case model if the second base station supports the model use case change IE. The model use case may be represented by enumerated values, which may include at least UE mobility prediction, UE traffic prediction, etc., for example, to improve accuracy of the target use case or reduce computational complexity to save network resources.
Model class change IE
If the model update message carries a model category change IE, the second base station changes the model category according to the indication of the model category change IE, and selects a more appropriate and accurate algorithm to construct the model. Model classes, which may be represented by enumerated values, may include at least linear regression, logistic regression, decision trees, and the like, for example.
Model parameter Change IE
If the model update message carries a parameter change IE, the second base station should change the model parameter information according to the indication of the model parameter change IE to obtain a more accurate model.
It should be noted that, after the model is trained online by the first base station, the auxiliary base station may change, at this time, the network management system may send a message to the first base station, where the message includes an auxiliary base station adding IE, and if supported, the first base station should add Guan Jizhan to the auxiliary base station ID list according to the indication information of the auxiliary base station adding IE, and is responsible for collecting and storing measurement information from the added relevant base station as model training information.
S12, the first base station sends a model reasoning data request message to the second base station.
In the embodiment of the disclosure, the first base station may further acquire first model inference data of the first base station, where the first model inference data may include first base station model inference data obtained from measurement of the first base station itself, and receive first terminal model inference data from the first terminal.
The first base station sends a model reasoning data request message to the second base station, so as to instruct the second base station to start, stop or increase the measurement process in the model deployment message/model update message, and obtain second model reasoning data.
In the embodiment of the disclosure, the model reasoning data request message may include network address information and port information of the first base station, so as to establish a transmission channel of AI/ML data between the first base station and the second base station, where the AI/ML data refers to data required for model reasoning.
The model reasoning data request message is fed back through the control plane and/or the user plane, and when the fed back data volume is large, the message is transmitted through the user plane, and the information contained in the model reasoning data request message comprises at least one of the following, but is not limited to:
The Message Type, for the sake of distinction, may be referred to as the first current Message Type, which may be, for example, a Message Type such as Produce Code, type of Message
-a first base station measurement ID, it being noted that in the embodiments of the disclosure the measurement ID is different from the base station ID, the base station ID being used to uniquely identify each base station, the measurement ID being used to indicate whose measurement information is
-second base station measurement ID
-a first base station IP address and port address: if the model reasoning data request message contains the IE, the second base station is instructed to feed back data through the user interface;
-a list of reported cells, for the sake of distinction, referred to as a current reported cell list, may comprise at least one of the following:
cell reporting information: here, the cell reporting information of the second cell corresponding to the second base station may include, but is not limited to, a cell ID, an SSB reporting list, an SSB index, and the like.
Reporting period: for distinction, referred to herein as the third reporting period, may be used to indicate the average window length of all measurement objects.
A list of feature input information, called a current feature input information list for distinction, for indicating the feature input information of one or more cells (here referred to as second cells) of the base station (here referred to as second base station) required for model reasoning, the feature input information of each cell including, but not limited to, at least one of the following:
UE feature input information, herein UE refers to the second terminal, which may be referred to as the current desired feature input information of the second terminal, and may include, but is not limited to, at least one of the following:
UE location information, where UE refers to the second terminal and is used for model reasoning, which may be referred to as the second terminal's current location information
UE historical mobility information, where UE refers to the second terminal and is used for model reasoning, which may be referred to as the second terminal's current historical mobility information
UE radio measurement information: the UE herein refers to the second terminal and is used for model reasoning, which may be referred to as second terminal current radio measurement information, may include one or more of RSRP, RSRQ, etc.
-base station characteristic input information, here the base station is the second base station, and for model reasoning, which may be referred to as the second base station currently required characteristic input information, may include, but is not limited to, at least one of the following:
base station radio measurement information: the base station is a second base station, and is used for model reasoning, and may be referred to as current radio measurement information of the second base station, which may include, but is not limited to, cell (herein, the second cell) PRB utilization, average RRC connection number, packet loss rate, etc
Current terminal traffic, herein referred to as traffic of the terminal currently connected to the second base station, may be referred to as second current terminal traffic for distinction
-neighbor base station feature input information, herein, the neighbor base station refers to a second neighbor base station, which may be referred to as feature input information currently required by the second neighbor base station, and may include, but is not limited to, at least one of the following:
neighboring base station wireless measurement information, wherein the neighboring base station refers to a second neighboring base station, which can be called the current wireless measurement information of the second neighboring base station
-traffic offload to UE performance measurements of neighbor cells: terminal related information indicating traffic offload to neighboring cells, comprising at least one or more or a combination of the following information: qoS parameter information such as UE position information, packet loss rate, time delay and the like, UE wireless measurement information and the like. Here, it means that the second terminal in the range of the second base station offloads the traffic to the second terminal current performance measurement information of the second neighboring cell corresponding to the second neighboring base station
Model inference information list IE: the data required to indicate the newly added model reasoning may include, but is not limited to, the following:
-the second base station has one or more cells of feature input information (i.e. the second base station corresponds to the second cell's current desired feature input information), each cell of feature input information comprising at least one or a combination of the following:
-predicting radio resource status: the downlink total PRB usage of a cell (here referred to as a second cell);
-predicting terminal traffic: sum of terminal traffic in cell (here referred to as second cell)
-predicting the neighbor base station radio resource status: downlink total PRB usage of a cell (here referred to as a second cell)
-a feature input information registration request IE:
if the feature input information registration request IE is set to "start", the receiving base station (here, the second base station) should start measurement according to the indication in the above-mentioned current feature input information list; or alternatively
If the feature input information registration request IE is set to "stop", the receiving base station (here, the second base station) should stop measurement and reporting; or alternatively
If the feature input information registration request IE is set to "add", the receiving base station (referred to herein as the second base station) should add the measured or predicted value indicated in the "model inference information list IE" to the previously given measurement initiated by the current feature input information reporting list. If a measurement has been initiated on the information indicated in the "model inference information list IE", this information will be ignored.
Reporting a feature, herein referred to as a current reporting feature for distinction, if the feature input information registration request IE indicates "start", then it is used to indicate the measurement object reported by the second base station requested for each position in the bitmap.
Optionally, s13a, the second base station sends a model reasoning data reply message to the first base station.
The second base station receives the model reasoning data request message from the first base station and indicates the measurement object information for which the measurement can be successfully started for the measurement object requested by the model reasoning data request message through the model reasoning data reply message, wherein the information comprises at least one of the following information:
-a second base station IP address and port address: the second base station is instructed to feed back data through the user interface;
message type, which for differentiation may be referred to as second current message type
-first base station measurement ID
-second base station measurement ID
Critical diagnostic IE: in order to distinguish what is referred to as the third critical diagnostic information element, if the model inference data reply message carries this IE, it is used to indicate which IEs are not understood or lost by the second base station in the received model inference data request message.
Optionally, s13b, the second base station sends a model reasoning data failure message to the first base station.
If the second base station cannot provide any requested measurements from the first base station, the second base station should send a model reasoning data fail message with appropriate cause values, including but not limited to at least one of the following:
Message type, in order to distinguish what may be referred to as a third current message type
-first base station measurement ID
-second base station measurement ID
-a second base station IP address and port address: the second base station is instructed to feed back data through the user interface;
failure cause: in order to distinguish what may be referred to as a second failure cause, it may be a cause of an XnAP protocol specific event, for example, at least one of a radio network layer cause, a transport layer cause, a protocol cause, etc.;
-critical diagnostics: to distinguish what may be referred to as a fourth critical diagnostic information element, when the received portion of the model reasoning data request message is not understood or lost, or the model reasoning data request message contains a logical error, the IE is used to indicate which IEs in the model reasoning data request message are not understood or lost;
waiting for a retransmission time, in order to distinguish what may be referred to as a second waiting retransmission time, the first base station is instructed to re-initiate a request (i.e. send a new model reasoning data request message) to the second base station.
Optionally, S14, the second base station sends a model reasoning data reporting message to the first base station.
In the embodiment of the disclosure, the model inference data report message may include the second model inference data and network address information and port information of the second base station, where the network address information and port information of the second base station may be used to establish a channel for transmitting AI/ML data between the first base station and the second base station.
The second base station sends a model reasoning data reporting message to the first base station for reporting the requested measurement information including, but not limited to, at least one of the following:
message type, for distinguishing what may be referred to as a fourth current message type
-second base station measurement ID
-a second base station IP address and port address: the second base station is instructed to feed back data through the user interface;
-cell measurements, here second cell current measurements of a second cell corresponding to the second base station, may include, but are not limited to, at least one of:
-a cell measurement item, herein referred to as a second cell current measurement item, may include, but is not limited to, at least one of:
cell ID, herein referred to as second cell identity, i.e. identity of the second cell
-cell characteristic input information reporting: the present feature input information of the second cell is reported, the feature input information (in order to distinguish reasoning from training, which may be referred to as present feature input information) needed to be reported by one or more cells (in this case, the second cell) of the base station (in this case, the second cell) includes measurement information configured in a model deployment message/model update message and measurement information needed by newly added model reasoning, and the feature input information of each cell at least includes one or a combination of the following:
UE feature input information, herein UE refers to the second terminal, which may be referred to as the second terminal current feature input information, may include, but is not limited to, at least one of the following:
UE location information, where UE refers to the second terminal, may be referred to as the second terminal's current location information
Historical mobility information of UE, herein UE refers to the second terminal, and may be referred to as the second terminal's current historical mobility information
-UE radio measurement information: the UE herein refers to the second terminal, which may be referred to as the current radio measurement information of the second terminal, and may include one or more of RSRP, RSRQ, and the like
Time stamp, which may be referred to as a first current time stamp for distinguishing
-base station characteristic input information, wherein the base station refers to a second base station, which may be referred to as current characteristic input information of the second base station, refers to characteristic input information of one or more cells of the second base station, and the characteristic input information of each cell at least comprises one or a combination of the following:
-base station radio measurement information: the base station refers to the second base station, which may be referred to as second base station radio measurement information, and may include, but is not limited to, cell (herein referred to as second cell) PRB utilization, average RRC connection number, packet loss rate, etc
-predicting radio resource status: the downlink total PRB usage of a cell (here referred to as a second cell);
-predicting terminal traffic: sum of terminal traffic in a cell (herein referred to as a second cell);
-predicting the neighbor base station radio resource status: the downlink total PRB usage of a cell (here referred to as a second cell);
time stamp, which may be referred to as a second current time stamp for distinguishing
-neighbor base station characteristic input information: the neighboring base station refers to a second neighboring base station, which can be called as current feature input information of the second neighboring base station, and refers to feature input information of one or more cells of the second neighboring base station, wherein the feature input information of each cell at least comprises one or a combination of the following:
-current/predicted radio resource status: the downlink total PRB usage of a cell (here referred to as a second neighbor cell);
-predicting terminal traffic: sum of terminal traffic in a cell (here, a second neighbor cell);
-traffic offload to UE performance measurements of neighbor cells: terminal related information indicating traffic offload to neighboring cells, comprising at least one or more or a combination of the following information: qoS parameter information such as UE position information, packet loss rate, time delay and the like, UE wireless measurement information and the like. Here, it means that the second terminal in the range of the second base station offloads the traffic to the second terminal current performance measurement information of the second neighboring cell corresponding to the second neighboring base station
Time stamp, which may be referred to as a third current time stamp for distinguishing
-model reasoning output valid time: the effective time of the output information is inferred for each model.
S15, the terminal reports UE measurement to the first base station.
In an embodiment of the present disclosure, the first model reasoning data may include first terminal model reasoning data. A terminal (first terminal) within the range of the first base station continuously periodically reports measurement information to the anchor base station according to the measurement configuration, for example, the first terminal sends a first terminal current measurement report message to the first base station, wherein the first terminal current measurement report message comprises first terminal model reasoning data, and the information comprises at least one of the following information but not limited to the following information:
UE historical mobility information, here UE means the first terminal, which may be referred to as the first terminal's current historical mobility information
-RRM measurement configuration, herein UE refers to a first terminal, which may be referred to as first current radio resource management measurement configuration information, which may include, but is not limited to, at least one of the following:
UE radio measurement information: the UE herein refers to the first terminal, which may be referred to as the current radio measurement information of the first terminal, and may include one or more of RSRP, RSRQ, and the like
-time stamp, which may be referred to as fourth current time stamp for distinguishing
-MDT measurement configuration, here UE refers to a first terminal, which may be referred to as first current minimization of drive tests measurement configuration information, which may include, but is not limited to, at least one of the following:
UE location information, where UE refers to the first terminal and may be referred to as the current location information of the first terminal;
UE movement speed, where UE refers to the first terminal and may be referred to as the current movement speed of the first terminal
Time stamp, which may be referred to as a fifth current time stamp for distinction.
S16, the first base station performs model reasoning.
In the embodiment of the disclosure, the first model inference data may further include first base station model inference data, that is, data for model inference obtained by measurement of the first base station itself, which may include, but is not limited to, at least one of the following:
-characteristic input information of one or more cells of the first base station (i.e. the first cell), the characteristic input information of each cell comprising at least one or a combination of the following:
-predicting radio resource status: downlink total PRB usage of a cell (i.e., a first cell);
-predicting terminal traffic: sum of terminal traffic in a cell (i.e., a first cell);
-predicting the neighbor base station radio resource status: downlink total PRB usage of a cell (i.e., a first cell).
The first base station obtains second model reasoning data through a model reasoning data report message, and periodically reports a measurement result of a terminal in an area to obtain first terminal model reasoning data, and obtains the first base station model reasoning data through self measurement, so that required model reasoning related data is obtained, a model reasoning function is executed, and model reasoning output information of an AI/ML-based network load balancing model including prediction information and decision information (such as switching decision information) is generated for analysis or execution operation of a network.
S17, the first base station performs model reasoning output to the second base station.
After the first base station executes the model reasoning function, a model reasoning output message comprising the prediction information and the decision information is sent to the second base station through an Xn interface control plane and/or a user plane, wherein the model reasoning output message is used for indicating the model reasoning output containing the prediction information or the switching decision information, and the information contained in the model reasoning output message comprises at least one of the following components but is not limited to the following components:
load balancing target cell list: indicating a target cell suitable for use as a bearer handover traffic;
-target cell radio resource status prediction: the method comprises the steps of representing prediction of wireless resource utilization rate of a target cell;
-predicting a list of congested cells: indicating a congested cell within a predicted effective time range;
-predicting a congestion relief cell list: indicating cells in which congestion is relieved within a predicted effective time range;
-predicting terminals migrating to the target cell: UE for indicating the predicted handover to the target base station;
-handover mechanism related parameters: indicating handover-related parameter information including, but not limited to, handover compensation range, handover trigger point, etc.
S18, the first base station and the second base station execute network load balancing related operation based on model reasoning analysis.
S19, the second base station sends a base station feedback message (also called a measurement feedback message) to the first base station.
After the second base station performs the load balancing operation, a measurement feedback message is sent to the first base station to provide data needed to monitor the network performance to further optimize the model, where the information contained in the measurement feedback message includes, but is not limited to, at least one of the following:
-feedback measurement information, including but not limited to at least one of:
-UE measurement information of a target base station, where the target base station refers to a base station to which the UE is handed over;
-resource status information of the target base station;
system KPIs (such as throughput, delay, RLF for current and neighbor base stations).
The above steps S7 to S19 are repeated.
S20, the network manager sends an AI/ML model training pause message to the first base station.
When the radio resource utilization rates of different cells of the network reach equilibrium and the state tends to be stable, in order to save network resources, the OAM sends an AI/ML model training pause message to the first base station and the second base station to instruct the network to exit the distributed load balancing mechanism based on the AI/ML technology, and stops the transmission of signaling and data about model training input, model reasoning output and the like, wherein the information contained in the AI/ML model training pause message comprises at least one of the following components but not limited to:
-AI/ML load balancing stop indication
-cause, the cause indication information including, but not limited to, at least one of:
network resource conservation
UE power saving.
In the related art, network load balancing is based on measurement reporting of a base station and a terminal, so that the problems of increased air interface signaling overhead and energy consumption are caused, and the method is passive, so that the network service quality is difficult to ensure. The embodiment of the disclosure provides an AI/ML-based distributed network load balancing scheme, wherein model training and model reasoning functions are deployed on a base station, and signaling design is carried out through the transmission process of AI/ML model training data and reasoning data among interfaces of the base station, so that the distributed data collection and model training of a network are realized to generate corresponding load prediction, and the load balancing scheme is actively generated through load flow prediction, thereby improving the service quality of the network and improving the system capacity.
The embodiment of the disclosure designs a signaling flow for a wireless network distributed load balancing use case based on AI/ML, realizes that a base station side collects network wireless resource states of itself and neighbor base stations, cell load, measurement data such as switching history performance and the like, UE wireless measurement, history mobility information and position information, and the base station side can perform model training and execute model reasoning functions based on the data sets to generate prediction information and decision information related reasoning analysis for a network to execute corresponding operations, thereby realizing distributed network load balancing, being beneficial to the network to take expected measures to prevent network user experience from being reduced and ensuring network service quality.
The embodiment of the disclosure carries out the design of signaling flow aiming at the AI/ML data cooperation process between Xn interfaces, and enhances the interaction of the related AI/ML data between the source base station and the candidate target base station. The related signaling flow between base stations comprises interaction of measurement information and prediction information of network performance and network resource states between base stations, deployment and update of an AI/ML model between base stations, transmission of model reasoning information including mobility management optimization and switching decision between base stations, and signaling cooperation of feedback information between base stations.
Fig. 5 schematically illustrates a flow chart of a method for load balancing according to another embodiment of the present disclosure. The method provided by the embodiment of fig. 5 may be performed by a network management system, but the disclosure is not limited thereto.
As shown in fig. 5, a method provided by an embodiment of the present disclosure may include:
in S510, measurement report information and geographical information transmitted by the base station are acquired.
In S520, a load balancing area range is determined based on the measurement report information and the geographic information.
In S530, according to the evaluation index of the base stations in the load balancing area, determining that the first base station is an anchor base station in the load balancing area, and the second base station is an auxiliary base station in the load balancing area.
In S540, a preconfigured AI/ML-based network load balancing model is invoked, and the network load balancing model is trained offline based on a wireless measurement history data set reported by base stations within the load balancing area.
In S550, a first model deployment message is sent to the first base station, to instruct to deploy the network load balancing model for offline training to the first base station, and to instruct the first base station to acquire first model training data and second model training data, and to train the network load balancing model deployed at the first base station based on the first model training data and the second model training data.
In S560, a second model deployment message is sent to the second base station to deploy the offline trained network load balancing model to the second base station.
In an exemplary embodiment, the second model deployment message may include at least one of:
model index of network load balancing model;
inputting a third characteristic input information list of the network load balancing model;
a third feature input information request information element;
the third characteristic input information adds an information element.
In an exemplary embodiment, the third feature input information list of the network load balancing model may include at least one of:
characteristic input information of the second terminal;
the second base station characteristic inputs information;
the second neighbor base station characteristics input information.
In an exemplary embodiment, the method may further include:
sending an AI/ML model training pause message to the first base station;
the AI/ML model training pause message includes AI/ML load balancing stop indication information.
The execution order of S550 and S560 is not limited, and both may be executed in parallel, or S550 may be executed first and S560 may be executed second, or S560 may be executed first and S550 may be executed second.
Other content of embodiments of the present disclosure may be referenced to the other embodiments described above.
Fig. 6 schematically illustrates a flow chart of a method for load balancing according to yet another embodiment of the present disclosure. The method provided in the embodiment of fig. 6 may be performed by a second base station, where the second base station may be a secondary base station within a load balancing area, and the load balancing area may further include a first base station, and the first base station may be an anchor base station within the load balancing area; and the first base station is provided with an AI/ML-based network load balancing model which is trained offline through a network management system.
As shown in fig. 6, a method provided by an embodiment of the present disclosure may include:
in S610, a second model deployment message sent by a network management system is received, so as to deploy the network load balancing model trained offline by the network management system to the second base station.
In S620, a model training data request message transmitted by the first base station is received, the model training data request message including network address information and port information of the first base station for establishing a channel for transmitting AI/ML data.
In S630, a model training data reply message is sent to the first base station.
In S640, second model training data is obtained according to the model training data request message.
In S650, a training data report message is sent to the first base station, where the training data report message includes the second model training data and network address information and port information of the second base station.
Other content of embodiments of the present disclosure may be referenced to the other embodiments described above.
It should also be understood that the above is only intended to assist those skilled in the art in better understanding the embodiments of the present disclosure, and is not intended to limit the scope of the embodiments of the present disclosure. It will be apparent to those skilled in the art from the foregoing examples that various equivalent modifications or variations can be made, for example, some steps of the methods described above may not be necessary, or some steps may be newly added, etc. Or a combination of any two or more of the above. Such modifications, variations, or combinations thereof are also within the scope of the embodiments of the present disclosure.
It should also be understood that the foregoing description of the embodiments of the present disclosure focuses on highlighting differences between the various embodiments and that the same or similar elements not mentioned may be referred to each other and are not repeated here for brevity.
It should also be understood that the sequence numbers of the above processes do not mean the order of execution, and the order of execution of the processes should be determined by the functions and internal logic thereof, and should not constitute any limitation on the implementation process of the embodiments of the present disclosure.
It is also to be understood that in the various embodiments of the disclosure, terms and/or descriptions of the various embodiments are consistent and may be referenced to one another in the absence of a particular explanation or logic conflict, and that the features of the various embodiments may be combined to form new embodiments in accordance with their inherent logic relationships.
Examples of methods provided by the present disclosure for load balancing are described above in detail. It will be appreciated that the terminals and network devices, in order to implement the above-described functions, include corresponding hardware structures and/or software modules that perform the respective functions. Those of skill in the art will readily appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as hardware or combinations of hardware and computer software. Whether a function is implemented as hardware or computer software driven hardware depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present disclosure.
Fig. 7 schematically illustrates a schematic block diagram of a first base station according to an embodiment of the present disclosure. The first base station 700 provided in the embodiment of fig. 7 may include a processing unit 710, a transmitting unit 720, and a receiving unit 730.
The processing unit 710 may be configured to receive a first model deployment message sent by a network management system, so as to determine that the first base station is an anchor base station within a load balancing area range and the second base station is an auxiliary base station within the load balancing area range, and deploy an AI/ML-based network load balancing model at the first base station.
The processing unit 710 may be further configured to obtain first model training data for the first base station in response to the first model deployment message.
The sending unit 720 may be configured to send a model training data request message to the second base station according to the first model deployment message, for instructing the second base station to obtain second model training data, where the model training data request message may include network address information and port information of the first base station to establish a channel for transmitting AI/ML data.
The receiving unit 730 may be further configured to receive a model training data reply message sent by the second base station.
The receiving unit 730 may be further configured to receive a training data report message sent by the second base station, where the training data report message may include the second model training data and network address information and port information of the second base station.
The processing unit 710 may be further configured to train the network load balancing model deployed at the first base station by the first model training data and the second model training data.
The sending unit 720 may be further configured to send a model update message to the second base station, for indicating to update the network load balancing model deployed on the second base station with the network load balancing model for which the training of the first base station is completed.
Other content of the first base station provided by the embodiments of the present disclosure may refer to the other embodiments described above.
Fig. 8 schematically illustrates a schematic block diagram of a network management system according to an embodiment of the present disclosure. The network management system 800 shown in fig. 8 may include a receiving unit 810, a processing unit 820, and a transmitting unit 830.
The receiving unit 810 may be configured to obtain measurement report information and geographic information sent by a base station.
Processing unit 820 may be configured to determine a load balancing area range based on the measurement report information and the geographic information.
The processing unit 820 may be further configured to determine, according to the evaluation index of the base stations within the load balancing area, that the first base station is an anchor base station within the load balancing area, and that the second base station is a secondary base station within the load balancing area.
The processing unit 820 may also be configured to invoke a preconfigured AI/ML-based network load balancing model, and to train the network load balancing model offline based on a wireless measurement history data set reported by base stations within the load balancing area.
The sending unit 830 may be configured to send a first model deployment message to the first base station, to instruct deployment of the network load balancing model for offline training to the first base station, and to instruct the first base station to acquire first model training data and second model training data, and to train the network load balancing model based on the first model training data and the second model training data.
The sending unit 830 may be further configured to send a second model deployment message to the second base station to deploy the offline trained network load balancing model to the second base station.
Other content of the network management system provided by the embodiments of the present disclosure may refer to the other embodiments described above.
Fig. 9 schematically illustrates a schematic block diagram of a second base station according to an embodiment of the disclosure. The second base station in the embodiment of the present disclosure may be an auxiliary base station within a load balancing area, where the load balancing area may further include a first base station, and the first base station may be an anchor base station within the load balancing area; an AI/ML-based network load balancing model trained offline through a network management system can be deployed in the first base station.
The second base station 900 provided in the embodiment of fig. 9 may include a receiving unit 910, a transmitting unit 920, and a processing unit 930.
The receiving unit 910 is configured to receive a second model deployment message sent by a network management system, so as to deploy the network load balancing model trained offline by the network management system to the second base station.
The receiving unit 910 is further configured to receive a model training data request message sent by the first base station, where the model training data request message includes network address information and port information of the first base station, so as to establish a channel for transmitting AI/ML data.
The sending unit 920 is configured to send a model training data reply message to the first base station.
The processing unit 930 is configured to obtain second model training data according to the model training data request message.
The sending unit 920 is further configured to send a training data report message to the first base station, where the training data report message includes the second model training data and network address information and port information of the second base station.
Other content of the second base station provided by the embodiments of the present disclosure may refer to the other embodiments described above.
The communication device 1000 (which may be a terminal or a network device or a network management system) as shown in fig. 10 may include a processor 1010, a memory 1020, and a transceiver 1030.
Further, the embodiment of the disclosure also provides a wireless communication system, which comprises network equipment, a terminal and a network management system.
It can be clearly understood by those skilled in the art that, when the steps executed by the network device, the terminal and the network management system and the corresponding beneficial effects can refer to the related descriptions of the network device, the terminal and the network management system in the above method, the details are not repeated here for brevity.
It should be understood that the above division of the units is only a functional division, and other division methods are possible in practical implementation.
The embodiment of the disclosure also provides a processing device, which comprises a processor and an interface; the processor is configured to perform the method for load balancing in any of the method embodiments described above.
It should be understood that the processing means may be a chip. For example, the processing device may be a Field programmable gate array (Field-Programmable Gate Array, FPGA), an application specific integrated Chip (Application Specific Integrated Circuit, ASIC), a System on Chip (SoC), a central processing unit (Central Processor Unit, CPU), a network processor (Network Processor, NP), a digital signal processing circuit (Digital Signal Processor, DSP), a microcontroller (Micro Controller Unit, MCU), a programmable controller (Programmable Logic Device, PLD) or other integrated Chip.
In implementation, the steps of the above method may be performed by integrated logic circuits of hardware in a processor or by instructions in the form of software. The steps of a method disclosed in connection with the embodiments of the present disclosure may be embodied directly in a hardware processor for execution, or in a combination of hardware and software modules in a processor for execution. The software modules may be located in a random access memory, flash memory, read only memory, programmable read only memory, or electrically erasable programmable memory, registers, etc. as well known in the art. The storage medium is located in a memory, and the processor reads the information in the memory and, in combination with its hardware, performs the steps of the above method. To avoid repetition, a detailed description is not provided herein.
It should be noted that the processor in the embodiments of the present disclosure may be an integrated circuit chip with signal processing capabilities. In implementation, the steps of the above method embodiments may be implemented by integrated logic circuits of hardware in a processor or instructions in software form. The processor may be a general purpose processor, a digital signal processor (digitalsignal processor, DSP), an application specific integrated circuit (application specific integrated crcuit, ASIC), an off-the-shelf programmable gate array (field programmable gate array, FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware components. The various methods, steps and logic blocks of the disclosure in the embodiments of the disclosure may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of a method disclosed in connection with the embodiments of the present disclosure may be embodied directly in hardware, in a decoded processor, or in a combination of hardware and software modules in a decoded processor. The software modules may be located in a random access memory, flash memory, read only memory, programmable read only memory, or electrically erasable programmable memory, registers, etc. as well known in the art. The storage medium is located in a memory, and the processor reads the information in the memory and, in combination with its hardware, performs the steps of the above method.
The disclosed embodiments also provide a computer readable medium having stored thereon a computer program which, when executed by a computer, implements the method for load balancing in any of the method embodiments described above.
The disclosed embodiments also provide a computer program product which, when executed by a computer, implements the method for load balancing in any of the method embodiments described above.
The embodiment of the disclosure also provides a system chip, which comprises: a processing unit, which may be, for example, a processor, and a communication unit, which may be, for example, an input/output interface, pins or circuitry, etc. The processing unit may execute computer instructions to cause the chips within the terminal, the primary node, and the secondary node to perform any of the methods for load balancing provided by the embodiments of the present disclosure described above.
Optionally, the computer instructions are stored in a storage unit.
Alternatively, the storage unit is a storage unit in the chip, such as a register, a cache, etc., and the storage unit may also be a storage unit in the terminal located outside the chip, such as a read-only memory (ROM) or other type of static storage device that can store static information and instructions, a random access memory (random access memory, RAM), etc. The processor mentioned in any of the above may be a CPU, microprocessor, ASIC, or one or more integrated circuits for controlling the execution of the programs of the method for load balancing described above. The processing unit and the storage unit may be decoupled and respectively disposed on different physical devices, and the respective functions of the processing unit and the storage unit are implemented by wired or wireless connection, so as to support the system chip to implement the various functions in the foregoing embodiments. Alternatively, the processing unit and the memory may be coupled to the same device.
In the above embodiments, it may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer instructions are loaded and executed on a computer, the processes or functions in accordance with embodiments of the present disclosure are produced in whole or in part. The computer may be a general purpose computer, a special purpose computer, a computer network, or other programmable apparatus. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by a wired (e.g., coaxial cable, fiber optic, digital subscriber line (digital subscriber line, DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. that contains an integration of one or more available media. The usable medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a magnetic tape), an optical medium (e.g., a high-density digital video disc (digital video disc, DVD)), or a semiconductor medium (e.g., a Solid State Disk (SSD)), or the like.
In the several embodiments provided in the present disclosure, it should be understood that the disclosed systems, devices, and methods may be implemented in other manners. For example, the apparatus embodiments described above are merely illustrative, e.g., the division of the units is merely a logical function division, and there may be additional divisions when actually implemented, e.g., multiple units or components may be combined or integrated into another system, or some features may be omitted or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or units, which may be in electrical, mechanical or other form.
The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in each embodiment of the present disclosure may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit.
Various objects such as various messages/information/devices/network elements/systems/devices/actions/operations/processes/concepts may be named in the present disclosure, and it should be understood that these specific names do not constitute limitations on related objects, and that the named names may be changed according to the scenario, context, or usage habit, etc., and understanding of technical meaning of technical terms in the present disclosure should be mainly determined from functions and technical effects that are embodied/performed in the technical solution.
In various embodiments of the disclosure, where no special description or logic conflict exists, terms and/or descriptions between the various embodiments are consistent and may reference each other, and features of the various embodiments may be combined to form new embodiments in accordance with their inherent logic relationships.
The foregoing is merely specific embodiments of the disclosure, but the protection scope of the disclosure is not limited thereto, and any person skilled in the art can easily think about changes or substitutions within the technical scope of the disclosure, and it is intended to cover the scope of the disclosure. Therefore, the protection scope of the present disclosure shall be subject to the protection scope of the claims.
Claims (44)
1. A method for load balancing, the method performed by a first base station, the method comprising:
receiving a first model deployment message sent by a network management system to determine that the first base station is an anchor base station in a load balancing area range and the second base station is an auxiliary base station in the load balancing area range, and deploying an AI/ML-based network load balancing model at the first base station;
responding to the first model deployment message to acquire first model training data of the first base station;
according to the first model deployment message, a model training data request message is sent to the second base station and used for indicating the second base station to obtain second model training data, and the model training data request message comprises network address information and port information of the first base station and is used for establishing a channel for transmitting AI/ML data;
receiving a model training data reply message sent by the second base station;
receiving a training data reporting message sent by the second base station, wherein the training data reporting message comprises the second model training data and network address information and port information of the second base station;
Training the network load balancing model deployed at the first base station by the first model training data and the second model training data;
and sending a model update message to the second base station for indicating that the network load balancing model deployed on the second base station is updated with the network load balancing model completed by the training of the first base station.
2. The method of claim 1, wherein the first model deployment message comprises at least one of:
anchor point base station indication information;
a list of auxiliary base station identifications in the load balancing area;
model index of network load balancing model;
inputting a first characteristic input information list of a network load balancing model;
a first feature input information request information element;
the first characteristic input information adds an information element.
3. The method of claim 2, wherein the model index of the network load balancing model comprises at least one of:
the model use case at least comprises a load prediction model and a terminal track prediction model;
model class, is used for pointing out the adaptation model of each model use case;
model parameters, including at least one of: weight, bias, learning rate, number of iterations, support vector machine support vector.
4. The method of claim 2, wherein the first list of feature input information for the network load balancing model comprises at least one of:
characteristic input information of a first terminal in the range of a first base station;
the first base station characteristic input information;
the first neighbor base station characteristic inputs information.
5. The method of claim 4, wherein the characteristic input information of the first terminal within range of the first base station comprises at least one of:
position information of the first terminal;
historical mobility information of the first terminal;
wireless measurement information of the first terminal;
wherein the first base station characteristic input information comprises at least one of:
wireless measurement information of a first base station;
the current terminal flow of the first base station;
wherein the first neighbor base station characteristic input information includes at least one of: .
Wireless measurement information of a first neighboring base station;
and unloading the traffic of the first terminal in the range of the first base station to the first terminal performance measurement information of the first neighbor cell corresponding to the first neighbor base station.
6. The method according to claim 2, wherein:
when the first characteristic input information request information element is set to be started, the first base station is instructed to start measurement according to a first characteristic input information list;
When a first feature input information request information element is set to stop, instructing the first base station to stop measurement and reporting;
and when the first characteristic input information request information element is set to be added, the first base station is instructed to start measurement according to the first characteristic input information adding information element.
7. The method of claim 1, wherein the first model training data comprises first terminal model training data;
wherein, responding to the first model deployment message to obtain first model training data of the first base station comprises the following steps:
according to the first model deployment message, sending a measurement configuration message to a first terminal in a first base station range so as to instruct the first terminal to measure and obtain the first terminal model training data;
and receiving a first terminal sample measurement report message sent by the first terminal, wherein the first terminal sample measurement report message comprises the first terminal model training data.
8. The method of claim 7, wherein the measurement configuration message comprises at least one of:
historical mobility information of the first terminal;
radio resource management measurement configuration information of the first terminal;
The method comprises the steps of measuring configuration information of the minimization drive tests of a first terminal;
wherein the radio resource management measurement configuration information of the first terminal includes at least one of:
supporting periodic measurement triggering, including a triggering period and a recording period;
wireless measurement information of the first terminal;
the minimization of drive test measurement configuration information of the first terminal comprises at least one of the following:
supporting periodic measurement triggering, including a triggering period and a recording period;
position information of the first terminal;
the speed of movement of the first terminal.
9. The method of claim 1, wherein the first model training data comprises first base station model training data;
wherein, responding to the first model deployment message to obtain first model training data of the first base station comprises the following steps:
and triggering measurement of the first base station model training data according to the first model deployment message so as to obtain the first base station model training data.
10. The method of claim 1, wherein the model training data request message comprises at least one of:
a first sample message type;
the first base station measures the identification;
The second base station measures the identification;
the network address information and the port information of the first base station indicate the second base station to perform data feedback through a user interface when the model training data request message comprises the network address information and the port information of the first base station;
reporting a cell list by a sample;
inputting a required characteristic into an information list;
adding a model training information list information element;
inputting information registration request information elements by the required characteristics;
and reporting the characteristics of the sample.
11. The method of claim 10, wherein the sample reporting cell list comprises at least one of:
the cell reporting information of the second cell corresponding to the second base station;
a first reporting period;
the required feature input information list indicates feature input information of second cells corresponding to second base stations required for model training, and the feature input information of each second cell comprises at least one of the following:
inputting information of characteristics required by samples of a second terminal in the range of a second base station;
inputting information of the required characteristics of the second base station sample;
inputting information of the required characteristics of the second neighbor base station sample;
wherein:
when the required characteristic input information registration request information element is set to be started, the second base station is instructed to start measurement according to the required characteristic input information list;
When the required feature input information registration request information element is set to stop, the second base station is instructed to stop measurement and reporting;
and when the required characteristic input information registration request information element is set to be added, the second base station is instructed to start measurement according to the added model training information list information element.
12. The method of claim 11, wherein the sample required feature input information for the second terminal within range of the second base station comprises at least one of:
second terminal sample position information;
second terminal sample history mobility information;
the second terminal sample wirelessly measures information;
wherein the second base station sample required feature input information comprises at least one of:
the second base station samples wirelessly measuring information;
second sample terminal traffic;
the characteristic input information required by the second neighbor base station sample comprises at least one of the following:
the second neighbor base station samples wirelessly measure information;
and unloading the flow to second terminal sample performance measurement information of a second neighbor cell corresponding to a second neighbor base station of the second base station by the second terminal in the range of the second base station.
13. The method of claim 1, wherein the second model training data comprises at least one of:
A fourth sample message type;
the first base station measures the identification;
the second base station measures the identification;
the cell measurement result of the second cell corresponding to the second base station comprises a cell measurement item, wherein the cell measurement item comprises a cell identifier and a required characteristic input information list;
wherein the required feature input information list includes at least one of:
second terminal sample characteristic input information of a second terminal within range of a second base station;
the second base station sample characteristic input information;
and inputting information by the sample characteristics of the second adjacent base station.
14. The method of claim 13, wherein the second terminal sample feature input information for the second terminal within range of the second base station comprises at least one of:
second terminal sample position information;
second terminal sample history mobility information;
the second terminal sample wirelessly measures information;
a first sample timestamp;
wherein the second base station sample feature input information comprises at least one of:
the second base station samples wirelessly measuring information;
second sample terminal traffic;
a second sample timestamp;
the second neighbor base station sample characteristic input information of the second base station comprises at least one of the following:
The second neighbor base station samples wirelessly measure information;
unloading the flow to second terminal sample performance measurement information of a second neighbor cell corresponding to a second neighbor base station by a second terminal in the range of the second base station;
third sample timestamp.
15. The method of claim 1, wherein the model training data reply message comprises at least one of:
network address information and port information of the second base station;
a second sample message type;
the first base station measures the identification;
the second base station measures the identification;
a first critical diagnostic information element, the first critical diagnostic information element being used to indicate an unintelligible or missing information element in the model training data request message received by the second base station.
16. The method as recited in claim 1, further comprising:
receiving a model training data failure message sent by the second base station;
wherein the model training data failure message includes at least one of:
a third sample message type;
the first base station measures the identification;
the second base station measures the identification;
a first failure cause;
a second critical diagnostic information element, the second critical diagnostic information element being used to indicate an unintelligible or missing information element in the model training data request message received by the second base station;
A first waiting retransmission time.
17. The method of claim 1, wherein the model update message comprises at least one of:
a second feature input information list;
a second feature input information request information element;
a first reporting feature;
a second reporting period;
a second feature input information adding information element;
the model use case changes the information element;
model class modification information elements;
the model parameters change the information element.
18. The method of claim 17, wherein the second list of feature input information comprises at least one of:
inputting information by the second terminal characteristics;
the second base station characteristic inputs information;
the second adjacent base station characteristic input information;
wherein:
when the second characteristic input information request information element is set to be started, the second base station is instructed to start measurement according to a second characteristic input information list;
when a second feature input information request information element is set to stop, then instructing the second base station to stop measurement and reporting;
and when the second characteristic input information request information element is set to be added, the second base station is instructed to start measurement according to the second characteristic input information adding information element.
19. The method of claim 18, wherein the second terminal characteristic input information comprises at least one of:
second terminal position information;
the second terminal history mobility information;
the second terminal wirelessly measures information;
wherein the second base station characteristic input information comprises at least one of:
the second base station wirelessly measures information;
the second current terminal flow;
wherein the second neighbor base station characteristic input information includes at least one of:
the second adjacent base station wirelessly measures information;
and unloading the flow to second terminal performance measurement information of a second neighbor cell corresponding to the second neighbor base station by the second terminal in the range of the second base station.
20. The method as recited in claim 1, further comprising:
acquiring first model reasoning data of the first base station;
sending a model reasoning data request message to the second base station to instruct the second base station to obtain second model reasoning data, wherein the model reasoning data request message comprises network address information and port information of the first base station and is used for establishing a channel for transmitting AI/ML data;
receiving a model reasoning data reply message sent by the second base station;
Receiving a model reasoning data report message sent by the second base station, wherein the model reasoning data report message comprises the second model reasoning data and network address information and port information of the second base station;
processing the first model reasoning data and the second model reasoning data through the network load balancing model deployed at the first base station to obtain model reasoning output information;
and sending a model reasoning output message to the second base station, wherein the model reasoning output message comprises model reasoning output information, and the model reasoning output information is used for indicating the first base station and the second base station to execute corresponding network load balancing operation.
21. The method of claim 20, wherein the first model inference data comprises first terminal model inference data;
the method for acquiring the first model reasoning data of the first base station comprises the following steps:
receiving a first terminal current measurement report message sent by a first terminal in a first base station range, wherein the first terminal current measurement report message comprises first terminal model reasoning data;
wherein the first terminal model reasoning data comprises at least one of:
The first terminal current historical mobility information;
first current radio resource management measurement configuration information;
the first current MDT measurement configuration information.
22. The method of claim 21, wherein the first current radio resource management measurement configuration information comprises at least one of:
the first terminal is in wireless measurement information at present;
a fourth current timestamp;
wherein the first current minimization of drive tests measurement configuration information comprises at least one of:
the current position information of the first terminal;
the current moving speed of the first terminal;
and a fifth current timestamp.
23. The method of claim 20, wherein the first model reasoning data comprises first base station model reasoning data;
the first base station model reasoning data includes feature input information of one or more first cells of the first base station, each feature input information of the first cell including at least one of:
predicting a radio resource state, which includes a downlink total physical resource block usage of a corresponding first cell;
predicting terminal traffic, including a sum of terminal traffic in the first cell;
and predicting the wireless resource state of the adjacent base station, wherein the wireless resource state comprises the downlink total physical resource block use condition of the corresponding first cell.
24. The method of claim 20, wherein the model inference data request message comprises at least one of:
a first current message type;
the first base station measures the identification;
the second base station measures the identification;
the first base station network address and the port address, and when the model reasoning data request message comprises the first base station network address and the port address, the second base station is instructed to perform data feedback through a user interface;
the cell list is currently reported;
the current feature inputs an information list;
model reasoning information list information element;
a feature input information registration request information element;
the characteristics are currently reported.
25. The method of claim 24, wherein the currently reported list of cells comprises at least one of:
the cell reporting information of the second cell corresponding to the second base station;
a third reporting period;
wherein the current feature input information list includes at least one of:
inputting information by the current required characteristics of the second terminal in the range of the second base station;
the second base station inputs information on the current required characteristics;
the second adjacent base station inputs information according to the current required characteristics;
Wherein the model reasoning information list information element comprises: the current required characteristic input information of a second cell corresponding to the second base station;
wherein:
when the feature input information registration request information element is set to be started, the second base station is instructed to start measurement according to the current feature input information list;
when the feature input information registration request information element is set to stop, the second base station is instructed to stop measurement and reporting;
and when the characteristic input information registration request information element is set to be added, the second base station is instructed to start measurement according to the model reasoning information list information element.
26. The method of claim 25, wherein the second terminal currently desired feature input information comprises at least one of:
the current position information of the second terminal;
the second terminal current historical mobility information;
the second terminal is in wireless measurement information at present;
wherein the second base station currently required feature input information includes at least one of:
the second base station measures the information wirelessly at present;
the second current terminal flow;
the current required feature input information of the second neighboring base station comprises at least one of the following:
The second adjacent base station measures the information wirelessly at present;
unloading the flow to second terminal current performance measurement information of a second neighbor cell corresponding to a second neighbor base station of the second base station by a second terminal in the range of the second base station;
wherein the currently required feature input information for each second cell comprises at least one of:
a predicted radio resource status of the second cell;
predicted terminal traffic for the second cell;
and predicting the wireless resource state of the adjacent base station of the second cell.
27. The method of claim 20, wherein the model inference data reply message includes at least one of:
a second base station network address and a port address;
a second current message type;
the first base station measures the identification;
the second base station measures the identification;
a third critical diagnostic information element, the third critical diagnostic information element being used to indicate an unintelligible or missing information element in the model reasoning data request message received by the second base station.
28. The method as recited in claim 20, further comprising:
receiving a model reasoning data failure message sent by the second base station;
wherein the model reasoning data failure message includes at least one of:
A third current message type;
the first base station measures the identification;
the second base station measures the identification;
a second base station network address and a port address;
a second failure cause;
a fourth critical diagnostic information element, the fourth critical diagnostic information element being used to indicate an unintelligible or missing information element in the model reasoning data request message received by the second base station;
and waiting for the retransmission time.
29. The method of claim 20, wherein the second model inference data comprises at least one of:
a fourth current message type;
the second base station measures the identification;
a second base station network address and a port address;
a second cell current measurement result of a second cell corresponding to the second base station;
the model inferences the output valid time.
30. The method of claim 29, wherein the second cell current measurement comprises a second cell current measurement item;
the second cell current measurement item comprises a second cell identifier and second cell current characteristic input information reporting, and the second cell current characteristic input information reporting indicates current characteristic input information to be reported by a second cell of the second base station;
The current characteristic input information of each second cell includes at least one of:
the current characteristics of the second terminal are input with information;
the second base station inputs information according to the current characteristics;
and the current characteristics of the second adjacent base station input information.
31. The method of claim 30, wherein the second terminal current feature input information comprises at least one of:
the current position information of the second terminal;
the second terminal current historical mobility information;
the second terminal is in wireless measurement information at present;
a first current timestamp;
wherein the second base station current feature input information indicates feature input information of second cells of the second base station, and the feature input information of each second cell includes at least one of the following:
the second base station wirelessly measures information;
a predicted radio resource status of the second cell;
predicted terminal traffic for the second cell;
predicting the wireless resource state of the neighboring base station of the second cell;
a second current timestamp;
the current feature input information of the second adjacent base station indicates feature input information of a second adjacent cell corresponding to the second adjacent base station of the second base station, and the feature input information of each second adjacent cell comprises at least one of the following:
The current/predicted radio resource status of the second neighbor cell;
predicting terminal flow of a second neighbor cell;
unloading the flow to the current performance measurement information of the second terminal of the second neighbor cell corresponding to the second neighbor base station by the second terminal in the range of the second base station;
and a third current timestamp.
32. The method of claim 20, wherein the model inference output information comprises at least one of:
load balancing a target cell list;
predicting the radio resource state of a target cell;
predicting a list of congested cells;
predicting a congestion relief cell list;
predicting a terminal migrated to a target cell;
switching mechanism related parameters.
33. The method as recited in claim 20, further comprising:
receiving a measurement feedback message sent by the second base station after the second base station executes corresponding network load balancing operation;
optimizing the network load balancing model based on the measurement feedback message;
wherein the feedback measurement information includes at least one of:
terminal measurement information of the target base station;
resource status information of the target base station;
key performance index of system.
34. The method as recited in claim 20, further comprising:
Receiving an AI/ML model training pause message sent by a network management system;
the AI/ML model training pause message includes AI/ML load balancing stop indication information.
35. The method of claim 34, wherein the AI/ML model training pause message further comprises cause indication information;
the cause indication information includes at least one of:
network resources are saved;
the terminal saves energy.
36. A method for load balancing, the method performed by a network management system, the method comprising:
acquiring measurement report information and geographic information sent by a base station;
determining a load balancing area range based on the measurement report information and the geographic information;
determining that a first base station is an anchor base station in the load balancing area according to the evaluation index of the base stations in the load balancing area, and a second base station is an auxiliary base station in the load balancing area;
invoking a preconfigured network load balancing model based on AI/ML, and training the network load balancing model offline based on a wireless measurement historical data set reported by a base station in the load balancing area range;
a first model deployment message is sent to the first base station, and is used for indicating that the network load balancing model which is trained offline is deployed to the first base station, and is used for indicating the first base station to acquire first model training data and second model training data, and the network load balancing model which is deployed to the first base station is trained based on the first model training data and the second model training data;
And sending a second model deployment message to the second base station to deploy the offline trained network load balancing model to the second base station.
37. The method of claim 36, wherein the second model deployment message comprises at least one of:
model index of network load balancing model;
inputting a third characteristic input information list of the network load balancing model;
a third feature input information request information element;
the third characteristic input information adds an information element.
38. The method of claim 37, wherein the third list of characteristic input information for the network load balancing model comprises at least one of:
characteristic input information of the second terminal;
the second base station characteristic inputs information;
the second neighbor base station characteristics input information.
39. The method as recited in claim 36, further comprising:
sending an AI/ML model training pause message to the first base station;
the AI/ML model training pause message includes AI/ML load balancing stop indication information.
40. A method for load balancing, wherein the method is performed by a second base station, the second base station is an auxiliary base station in a load balancing area, the load balancing area also comprises a first base station, and the first base station is an anchor base station in the load balancing area; the first base station is deployed with an AI/ML-based network load balancing model which is trained offline through a network management system;
Wherein the method comprises the following steps:
receiving a second model deployment message sent by a network management system, so as to deploy the network load balancing model of the offline training of the network management system to the second base station;
receiving a model training data request message sent by the first base station, wherein the model training data request message comprises network address information and port information of the first base station and is used for establishing a channel for transmitting AI/ML data;
sending a model training data reply message to the first base station;
obtaining second model training data according to the model training data request message;
and sending a training data report message to the first base station, wherein the training data report message comprises the second model training data, and network address information and port information of the second base station.
41. A first base station, comprising:
the processing unit is used for receiving a first model deployment message sent by the network management system to determine that the first base station is an anchor base station in the load balancing area range and the second base station is an auxiliary base station in the load balancing area range, and deploying an AI/ML-based network load balancing model in the first base station;
The processing unit is further configured to obtain first model training data of the first base station in response to the first model deployment message;
a sending unit, configured to send a model training data request message to the second base station according to the first model deployment message, where the model training data request message is used to instruct the second base station to obtain second model training data, and the model training data request message includes network address information and port information of the first base station, and is used to establish a channel for transmitting AI/ML data;
the receiving unit is used for receiving the model training data reply message sent by the second base station;
the receiving unit is further configured to receive a training data report message sent by the second base station, where the training data report message includes the second model training data and network address information and port information of the second base station;
the processing unit is further configured to train the network load balancing model deployed at the first base station through the first model training data and the second model training data;
the sending unit is further configured to send a model update message to the second base station, to instruct updating of the network load balancing model deployed on the second base station with the network load balancing model for which training of the first base station is completed.
42. A network management system, comprising:
the receiving unit is used for acquiring measurement report information and geographic information sent by the base station;
the processing unit is used for determining a load balancing area range based on the measurement report information and the geographic information;
the processing unit is further configured to determine, according to an evaluation index of the base stations in the load balancing area, that a first base station is an anchor base station in the load balancing area, and that a second base station is an auxiliary base station in the load balancing area;
the processing unit is also used for calling a preconfigured network load balancing model based on AI/ML, and offline training the network load balancing model based on a wireless measurement historical data set reported by a base station in the load balancing area range;
the sending unit is used for sending a first model deployment message to the first base station, used for indicating to deploy the offline trained network load balancing model to the first base station, and used for indicating the first base station to acquire first model training data and second model training data, and training the network load balancing model based on the first model training data and the second model training data;
The sending unit is further configured to send a second model deployment message to the second base station, so as to deploy the offline trained network load balancing model to the second base station.
43. The second base station is an auxiliary base station in a load balancing area, and further comprises a first base station in the load balancing area, wherein the first base station is an anchor base station in the load balancing area; the first base station is deployed with an AI/ML-based network load balancing model which is trained offline through a network management system;
wherein the second base station includes:
the receiving unit is used for receiving a second model deployment message sent by the network management system so as to deploy the network load balancing model for offline training of the network management system to the second base station;
the receiving unit is further configured to receive a model training data request message sent by the first base station, where the model training data request message includes network address information and port information of the first base station, so as to establish a channel for transmitting AI/ML data;
a sending unit, configured to send a model training data reply message to the first base station;
The processing unit is used for obtaining second model training data according to the model training data request message;
the sending unit is further configured to send a training data report message to the first base station, where the training data report message includes the second model training data and network address information and port information of the second base station.
44. A computer readable storage medium storing a computer program, characterized in that the computer program when executed by a processor implements the method of any one of claims 1 to 35 or the method of any one of claims 36 to 39 or the method of claim 40.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210893482.2A CN117545019A (en) | 2022-07-27 | 2022-07-27 | Method for load balancing and related equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210893482.2A CN117545019A (en) | 2022-07-27 | 2022-07-27 | Method for load balancing and related equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117545019A true CN117545019A (en) | 2024-02-09 |
Family
ID=89792400
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210893482.2A Pending CN117545019A (en) | 2022-07-27 | 2022-07-27 | Method for load balancing and related equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117545019A (en) |
-
2022
- 2022-07-27 CN CN202210893482.2A patent/CN117545019A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2022022334A1 (en) | Artificial intelligence-based communication method and communication device | |
| US20240040461A1 (en) | Ue, network node and methods for handling mobility information in a communications network | |
| US10966108B2 (en) | Optimizing radio cell quality for capacity and quality of service using machine learning techniques | |
| US11496933B2 (en) | Method and apparatus for updating handover parameters in open-radio access network (O-RAN) environment | |
| US20230362758A1 (en) | Methods and apparatuses for handover procedures | |
| CN109495935A (en) | A kind of switching method, base station and user terminal | |
| EP3962171A1 (en) | User equipment centric wide area optimization method and system thereof | |
| US20220210706A1 (en) | Method and apparatus for identifying target neighbor cell for handover of user equipment (ue) | |
| WO2020092541A2 (en) | Optimization of 5g (fifth generation) beam coverage and capacity and nsi (network slice instance) resource allocation | |
| US20230269606A1 (en) | Measurement configuration for local area machine learning radio resource management | |
| WO2022118083A1 (en) | Dynamic multi-access policy generation | |
| US20230198640A1 (en) | Channel state information values-based estimation of reference signal received power values for wireless networks | |
| US9844060B2 (en) | Self-tuning base station for a wireless network | |
| Qin et al. | AQ-learning approach for mobility robustness optimization in Lte-Son | |
| US9264960B1 (en) | Systems and methods for determinng access node candidates for handover of wireless devices | |
| CN116349308A (en) | Cell reselection related information associated with a network slice or closed access group of a wireless network | |
| CN114765789A (en) | Data processing method and device in wireless communication network | |
| US11576055B2 (en) | Method, apparatus and computer readable media for network optimization | |
| US11337131B1 (en) | Method and apparatus for recommending real-time handover to a target cell in open-radio access network (O-RAN) environment | |
| CN117545019A (en) | Method for load balancing and related equipment | |
| Sun et al. | Data driven smart load balancing in wireless networks | |
| CN106559800B (en) | Joint evaluation method and device for Radio Resource Management (RRM) measurement event | |
| WO2024109084A1 (en) | Systems and methods for predicting quality of experience of user equipment using an artificial intelligence model | |
| US20230379774A1 (en) | Systems, apparatuses and methods for cognitive service driven handover optimization | |
| US20240172016A1 (en) | Prediction of cell traffic in a network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |