Access method, communication device, chip, storage medium, and program product

Abstract

Provided in the present application are an access method, a terminal, a network device, a chip, a storage medium, and a product. The access method comprises: acquiring first information, wherein the first information is input information of an AI service; and on the basis of the first information, determining a response result of the AI service associated with an AI model, wherein the response result is a related result for a random access procedure. By means of the present application, the decision-making accuracy of cell reselection can be improved, thereby improving the communication performance.

Description

Access methods, communication equipment, chips, storage media and software products Technical Field

[0001] This application relates to the field of communication technology, and in particular to an access method, communication device, chip, storage medium and program product. Background Technology

[0002] In existing technologies, for cell reselection mechanisms, when multiple candidate cells meet the requirements, the cell with the best signal quality is usually selected as the target cell for reselection. Considering that terminals in NR systems access cells via beams, in order to increase the probability of successful access through a good beam, the determination of the target cell needs to consider both cell signal quality and an appropriate number of beams.

[0003] However, considering only the cell signal quality and the appropriate number of beams is too simplistic, which affects communication performance.

[0004] Summary of the Invention

[0005] This application provides an access method, communication device, chip, storage medium, and program product that can improve the accuracy of cell reselection decisions, thereby improving communication performance.

[0006] In a first aspect, embodiments of this application provide an access method applied to a first entity, the method comprising:

[0007] Obtain the first information; the first information is the input information for the AI ​​service;

[0008] Based on the first information, determine the response result of the AI ​​service associated with the AI ​​model; the response result is the relevant result for the random access process.

[0009] Secondly, embodiments of this application provide an access device, the device comprising:

[0010] The first communication unit is configured to acquire first information; the first information is input information for an AI service.

[0011] Based on the first information, determine the response result of the AI ​​service associated with the AI ​​model; the response result is the relevant result for the random access process.

[0012] Thirdly, embodiments of this application provide a communication device, the communication device comprising:

[0013] Memory is used to store executable instructions for a computer;

[0014] A processor, connected to the memory, is configured to implement the method described in the first aspect by executing the computer-executable instructions.

[0015] Fourthly, embodiments of this application provide a chip, the chip comprising:

[0016] A processor for retrieving and running a computer program from memory, causing a device on which the chip is mounted to perform the method described in the first aspect.

[0017] Fifthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by at least one processor, implements the method described in the first aspect.

[0018] In a sixth aspect, embodiments of this application provide a computer program product, including computer program instructions that, when executed by a processor, implement the method as described in the first aspect.

[0019] This application provides an access method, device, chip, storage medium, and program product. The method involves acquiring first information, which is input information for an AI service. Based on the first information, the method determines the response result of the AI ​​service associated with the AI ​​model. The response result is a relevant result for the random access process. By analyzing the first information using the AI ​​model, the random access process is optimized, enabling terminal devices to access the network faster and more stably. Based on the optimized results, the accuracy of cell reselection decisions is improved. Accurate cell reselection decisions ensure that devices select the optimal cell to camp on, guaranteeing signal quality and communication stability. Furthermore, it optimizes network resource allocation and improves overall network communication performance. Attached Figure Description

[0020] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the specification, serve to explain the technical solutions of this application. Obviously, the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0021] The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be broken down, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0022] Figure 1 is a schematic diagram of an optional application scenario provided by an embodiment of this application;

[0023] Figure 2 is a schematic diagram of an optional AI / ML model lifecycle management provided in an embodiment of this application;

[0024] Figure 3 is a flowchart illustrating an optional access method provided in an embodiment of this application;

[0025] Figure 4 is a schematic diagram of the structural composition of an optional communication device provided in an embodiment of this application;

[0026] Figure 5 is a schematic diagram of the structural composition of an optional communication device provided in an embodiment of this application;

[0027] Figure 6 is a schematic diagram of the structural composition of an optional chip provided in an embodiment of this application;

[0028] Figure 7 is a schematic diagram of the structural composition of an optional communication system provided in an embodiment of this application. Detailed Implementation

[0029] In order to gain a more detailed understanding of the features and technical content of the embodiments of this application, the implementation of the embodiments of this application will be described in detail below with reference to the accompanying drawings. The accompanying drawings are for reference and illustration only and are not intended to limit the embodiments of this application.

[0030] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0031] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0032] It should also be noted that the terms "first, second, and third" used in the embodiments of this application are only used to distinguish similar objects and do not represent a specific order of objects. It is understood that "first, second, and third" can be interchanged in a specific order or sequence where permitted, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0033] Figure 1 is a schematic diagram of an optional application scenario provided by an embodiment of this application.

[0034] As shown in Figure 1, the communication system 100 may include a terminal device 110 (also called a terminal) and a network device (NW) 120. The network device 120 can communicate with the terminal device 110 via an air interface. Multi-service transmission is supported between the terminal device 110 and the network device 120.

[0035] It should be understood that the embodiments of this application are only illustrated by way of example with communication system 100, but the embodiments of this application are not limited thereto. That is to say, the technical solutions of the embodiments of this application can be applied to various communication systems, such as: Long Term Evolution (LTE) system, LTE Time Division Duplex (TDD), Universal Mobile Telecommunication System (UMTS), Internet of Things (IoT) system, Narrow Band Internet of Things (NB-IoT) system, enhanced Machine-Type Communications (eMTC) system, 5G communication system (also known as New Radio (NR) communication system), or future communication systems, etc.

[0036] In the communication system 100 shown in Figure 1, the network device 120 can be an access network device that communicates with the terminal device 110. The access network device can provide communication coverage for a specific geographical area and can communicate with the terminal device 110 located within that coverage area.

[0037] In some embodiments, the network device may be an evolved Node B (eNB or eNodeB) in a Long Term Evolution (LTE) system, a Next Generation Radio Access Network (NG RAN) device, a base station (gNB) in an NR system, or a radio controller in a Cloud Radio Access Network (CRAN). Alternatively, the network device may be a macro base station, a micro base station (also known as a small station), a satellite, a Radio Network Controller (RNC), a Node B (NB), a Base Station Controller (BSC), a Base Transceiver Station (BTS), a Home Evolved Node B (or Home Node B, HNB), a Baseband Unit (BBU), an Access Point (AP), a Wireless Relay Node, a Wireless Backhaul Node, a Transmission Point (TP), or a Transmission and Reception Point (TRP) in a Wireless Fidelity (WiFi) system. This network equipment can also be used as a relay station, access point, vehicle-mounted equipment, wearable devices, hub, switch, bridge, router, or network equipment in the future evolved Public Land Mobile Network (PLMN).

[0038] In some embodiments, terminal device 110 may be any terminal device, including but not limited to terminal devices that are connected to network device 120 or other terminal devices via wired or wireless connection.

[0039] In some embodiments, terminal device 110 may refer to an access terminal, user equipment (UE), user unit, user station, mobile station, mobile station, remote station, remote terminal, mobile device, user terminal, terminal, wireless communication device, user agent, or user device. The access terminal may be a cellular phone, cordless phone, Session Initiation Protocol (SIP) phone, IoT device, satellite handheld terminal, Wireless Local Loop (WLL) station, Personal Digital Assistant (PDA), handheld device with wireless communication capabilities, computing device or other processing device connected to a wireless modem, in-vehicle device, wearable device, terminal device in a 5G network, or terminal device in a future evolved network, etc.

[0040] In some embodiments, the terminal device 110 can be used for device-to-device (D2D) communication.

[0041] Figure 1 illustrates an exemplary network device and two terminal devices. Optionally, the communication system 100 may include multiple network devices, and each network device may include other numbers of terminal devices within its coverage area. This application embodiment does not limit this.

[0042] It should be noted that Figure 1 is merely an example illustrating the system to which this application applies. Of course, the method shown in the embodiments of this application can also be applied to other systems. Furthermore, the terms "system" and "network" are often used interchangeably in this application.

[0043] It should be noted that Figure 1 is merely an example illustrating the system to which this application applies. Of course, the method shown in the embodiments of this application can also be applied to other systems. Furthermore, the terms "system" and "network" are often used interchangeably in this application. The term "and / or" in this application is simply a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this application generally indicates that the preceding and following related objects have an "or" relationship. It should also be understood that the "instruction" mentioned in the embodiments of this application can be a direct instruction, an indirect instruction, or an indication of an association relationship. For example, A instructing B can mean that A directly instructs B, for example, B can be obtained through A; it can also mean that A indirectly instructs B, for example, A instructs C, B can be obtained through C; or it can mean that there is an association relationship between A and B. It should also be understood that the term "correspondence" mentioned in the embodiments of this application can indicate a direct or indirect correspondence between two things, or an association between them, or a relationship of instruction and being instructed, configuration and being configured, etc. It should also be understood that the "predefined" or "predefined rules" mentioned in the embodiments of this application can be implemented by pre-storing corresponding codes, tables, or other methods that can be used to indicate relevant information in the device (e.g., including terminal devices and network devices), and this application does not limit the specific implementation method. For example, predefined can refer to what is defined in a protocol. It should also be understood that in the embodiments of this application, the term "protocol" can refer to standard protocols in the field of communication, such as the LTE protocol, the NR protocol, and related protocols applied to future communication systems, and this application does not limit this.

[0044] It should be understood that the term "and / or" in this application is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Furthermore, the character " / " in this application generally indicates that the preceding and following related objects have an "or" relationship.

[0045] It should also be understood that the term "instruction" mentioned in the embodiments of this application can be a direct instruction, an indirect instruction, or an indication of a relationship. For example, A instructing B can mean that A directly instructs B, such as B being able to obtain information through A; it can also mean that A indirectly instructs B, such as A instructing C, so B can obtain information through C; or it can mean that there is a relationship between A and B.

[0046] It should also be understood that the term "correspondence" mentioned in the embodiments of this application may indicate a direct or indirect correspondence between the two, or an association between the two, or a relationship of instruction and being instructed, configuration and being configured, etc.

[0047] It should also be understood that the "predefined" or "predefined rules" mentioned in the embodiments of this application can be implemented by pre-storing corresponding codes, tables, or other means that can be used to indicate relevant information in the device (e.g., including terminal devices and network devices), and this application does not limit the specific implementation method. For example, predefined can refer to those defined in a protocol. It should also be understood that in the embodiments of this application, the "protocol" can refer to standard protocols in the field of communication, such as LTE protocol, NR protocol, and related protocols applied to future communication systems, and this application does not limit it.

[0048] To facilitate understanding of the technical solutions of the embodiments of this application, the relevant technologies of the embodiments of this application are described below. The following relevant technologies are optional solutions and can be combined with the technical solutions of the embodiments of this application in any way, and they all fall within the protection scope of the embodiments of this application.

[0049] Related Technology 1:

[0050] Community Selection and Reselection

[0051] For User Equipment (UE) in RRC_IDLE (idle state) or RRC_INACTIVE (deactivated state), the prerequisite for camping on a cell is that the cell's signal quality (including the Reference Signal Received Power (RSRP) and Reference Signal Received Quality (RSRQ) measurements) meets the cell selection S criterion, which is the same as in the LTE system. After the UE selects a suitable cell, it will continuously evaluate cell reselection. The measurements to be performed for cell reselection are divided and performed according to the reselection priority of each frequency point. Specifically, this includes the following three points:

[0052] 1) For high-priority frequency points, neighbor cell measurement is always performed;

[0053] 2) For co-frequency points, when the RSRP and RSRQ values ​​of the serving cell are both higher than the co-frequency measurement threshold configured by the network, the terminal equipment (user equipment) can stop co-frequency neighbor cell measurement; otherwise, measurement must be performed.

[0054] 3) For frequency points of the same and lower priorities, when the RSRP and RSRQ values ​​of the serving cell are both higher than the inter-frequency measurement threshold configured by the network, the UE can stop the neighbor cell measurement of the same and lower priorities frequency points; otherwise, the measurement must be performed.

[0055] Furthermore, after obtaining multiple candidate cells through measurement, the process of determining the target cell for cell reselection is basically the same as in the LTE system, adopting the principle of prioritizing cells on high-priority frequency points for reselection. Specifically, this includes the following three points:

[0056] 1) For cell reselection on high-priority frequency points, the signal quality must be higher than a certain threshold and last for a specified duration, and the UE must remain in the source cell for no less than 1 second.

[0057] 2) For cell reselection on the same frequency and priority frequency points, the R criterion (ordered according to RSRP) must be met. The signal quality of the new cell is better than that of the current cell and lasts for a specified duration. The UE camps on the source cell for no less than 1 second.

[0058] 3) For cell reselection on low-priority frequency points, there must be no high-priority or same-priority frequency point cells that meet the requirements, the signal quality of the source cell must be below a certain threshold, the signal quality of the cell on the low-priority frequency point must be above a certain threshold and last for a specified duration, and the UE must stay in the source cell for no less than 1 second.

[0059] During cell reselection on the same frequency and priority frequency points, when multiple candidate cells meet the requirements, the LTE system selects the best cell as the target cell using the RSRP ranking method. Considering that UEs in NR systems access cells via beams, to increase the probability of successful access through a good beam, the target cell selection needs to consider both cell signal quality and the number of good beams. Ultimately, a scheme was adopted that does not change the ranking value (i.e., still uses cell measurement results for ranking): before selecting the target cell, the best few cells with similar signal quality are first selected, and then the cell with the most good beams is selected as the target cell.

[0060] Related Technology 2:

[0061] Artificial Intelligence (AI)

[0062] Artificial intelligence can be achieved through machine learning methods. Basic machine learning methods can include supervised learning, unsupervised learning, or reinforcement learning.

[0063] Supervised learning, also known as supervised learning or supervised instruction, involves learning or creating a pattern (also called a function or learning model) from training data and using this pattern to predict new instances. Training data consists of input objects (usually vectors) and expected outputs. The output can be a continuous value (called regression analysis) or a predicted classification label (called classification). A supervised learner's task is to predict the function's output for any possible input after observing some pre-labeled training examples (inputs and expected outputs). To achieve this, the learner must generalize from the existing data to unobserved situations in a "reasonable" (inductive bias) manner.

[0064] Reinforcement learning emphasizes how to act based on the environment to maximize expected benefits. Unlike supervised learning, reinforcement learning does not require labeled input-output pairs, nor does it require precise correction of suboptimal solutions. Its focus is on finding a balance between exploration (to the unknown) and exploitation (to existing knowledge). This "exploration-exploitation" exchange in reinforcement learning is most studied in multi-armed slot machine problems and finite Markov decision processes (MDPs). For MDPs, in machine learning problems, the environment is often abstracted as an MDP, and many reinforcement learning algorithms can only use dynamic programming methods under this assumption. The main difference between traditional dynamic programming methods and reinforcement learning algorithms is that the latter does not require knowledge about the MDP and is suitable for large-scale MDPs where an exact solution cannot be found.

[0065] Related technology 3:

[0066] Application of AI in the 3GPP System

[0067] AI for NW applies artificial intelligence technology to the network domain to improve network performance, efficiency, scalability, and intelligence. 3GPP began researching the AI / ML functional framework in 5G Rel-17 and identified high-level use cases based on AI / ML for NG-RAN. Rel-18 studied the standardization of AI / ML physical layer use cases (AI / ML-based localization, AI / ML-based beam management, and AI / ML-based channel state information) and high-level use cases (network power saving, load balancing, mobility optimization), and TR38.843 provides a general AI / ML framework and AI / ML model lifecycle management (LCM). As shown in Figure 2, LCM functions include the following aspects:

[0068] The data collection function provides input data for model training, management, and inference functions;

[0069] Training data: The data required for training AI / ML models;

[0070] Monitoring data: The input data required to manage AI / ML models or AI / ML functions;

[0071] Inference data: The data required for AI / ML inference functions.

[0072] In some embodiments, model training is a function that performs AI / ML model training, validation, and testing, and can generate model performance metrics that can be used as part of the model testing process. If necessary, the model training function is also responsible for data preparation (e.g., data preprocessing and cleaning, formatting, and transformation) based on the training data provided by the data collection function.

[0073] Training / Update Model: If a model storage function is available, it is used to transfer trained, validated, and tested AI / ML models to the model storage function, or to transfer updated versions of models to the model storage function.

[0074] Management is the function that oversees the operation (e.g., selection / (deactivation) activation / toggle / rollback) and monitoring (e.g., performance) of AI / ML models or functions. This function is also responsible for making decisions based on data received from data collection and inference functions to ensure correct inference operations.

[0075] Management instructions: The input information required to manage the inference function. This information may include the selection / (deactivation) activation / switching of AI / ML models or AI / ML-based functions, reverting to non-AI / ML operations (i.e., operations independent of the inference process), etc.

[0076] Model transfer / delivery request: Used to request a model from the model storage function.

[0077] Performance Feedback / Retraining Request: Information required for model training functions to input, such as for model (re)training or updating purposes.

[0078] Inference is a function that uses data provided by a data collection function (i.e., inference data) as input to provide the output of a process that applies an AI / ML model or AI / ML function. If necessary, the inference function is also responsible for data preparation based on the inference data provided by the data collection function (e.g., data preprocessing and cleaning, formatting and transformation).

[0079] Inference output: Management functions are used to monitor the performance of AI / ML models or AI / ML functions.

[0080] Model storage is a function responsible for storing trained / updated models that can be used to perform inference functions.

[0081] Model transfer / delivery: Used to deliver AI / ML models to inference functions.

[0082] The existing cell selection and reselection mechanism does not take into account the uplink transmission congestion level and network load balance, resulting in a suboptimal cell selection strategy and increasing the risk of initial access failure, rejection, or network load (network congestion) imbalance.

[0083] Based on this, this application provides an access method. The main idea of ​​this method is as follows: acquiring first information; the first information is the input information of an AI service; determining the response result of the AI ​​service associated with the AI ​​model based on the first information; the response result is a relevant result for the random access process. By using the first information indicating the uplink congestion situation, congestion can be monitored and handled in real time, reducing network load and avoiding overload. The response result of the AI ​​service associated with the AI ​​model is determined based on the first information, and the response result is a relevant result for the random access process. By analyzing the first information using the AI ​​model, the random access process is optimized, enabling terminal devices to access the network faster and more stably. Based on the optimized access result, the accuracy of cell reselection decisions is improved. Accurate cell reselection decisions ensure that devices select the optimal cell to camp on, guaranteeing signal quality and communication stability. It also optimizes the allocation of network resources and improves overall network communication performance.

[0084] To facilitate understanding of the technical solutions of the embodiments of this application, the technical solutions of this application are described in detail below through specific embodiments. The above-mentioned related technologies are optional solutions and can be arbitrarily combined with the technical solutions of the embodiments of this application, all of which fall within the protection scope of the embodiments of this application. The embodiments of this application include at least some of the following contents.

[0085] Figure 3 is a flowchart illustrating an optional access method provided in an embodiment of this application, applied to a first entity. As shown in Figure 3, the method may include S101 and S102:

[0086] S101. Obtain first information; the first information is the input information for the AI ​​service.

[0087] S102. Based on the first information, determine the response result of the AI ​​service associated with the AI ​​model; the response result is the relevant result for the random access process.

[0088] In some embodiments of this application, the AI ​​service includes at least one of the following: training of the AI ​​model, inference of the AI ​​model, and management of the AI ​​model.

[0089] In this embodiment of the application, when the AI ​​service is the training of an AI model, the AI ​​model is an initial AI model, and the response result is a first model determined by the training of the initial AI model.

[0090] When the AI ​​service is the inference of the AI ​​model, the AI ​​model is the first model, and the response result is the output data of the first model;

[0091] When the AI ​​service is used to manage the AI ​​model, the response result is the management result of the AI ​​model.

[0092] In this embodiment, training an AI model refers to using training data to learn and optimize the model to improve its performance on a specific task. In some embodiments, AI model training includes: data preparation: collecting, cleaning, preprocessing, and formatting training data; model training: training the AI ​​model using the training data and adjusting model parameters; validation and testing: evaluating model performance using validation and test data to prevent overfitting or underfitting; and model optimization: optimizing the model based on the evaluation results to ensure its effectiveness in practical applications. The goal of AI model training is to generate a high-performance AI model for use in inference.

[0093] In this application embodiment, inference refers to using a trained AI model to predict or make decisions based on newly input data. In some embodiments, the inference of the AI ​​model includes: data preparation: collecting, cleaning, and formatting inference data; model application: inputting the inference data into the AI ​​model to generate prediction results or decision outputs; and output results: feeding back the prediction results or decisions to the user or system. The purpose of AI model inference is to generate useful prediction or decision results based on real-time or batch data.

[0094] In this embodiment, AI model management includes operations such as model selection, activation, deactivation, switching, and performance monitoring. In some embodiments, AI model management includes: model selection and activation: selecting and activating appropriate AI models according to actual needs; model deactivation and switching: deactivating or switching models when needed to ensure stability and reliability; performance monitoring: continuously monitoring the performance of AI models, collecting and analyzing performance data; and model updates and rollback: updating models or rolling back to previous versions based on performance feedback and retraining requests. The purpose of AI model management is to ensure the efficient operation of AI models, respond promptly to changes, and improve overall system performance.

[0095] In this embodiment, the first entity can be deployed on a terminal device or a network device; this embodiment does not impose any restrictions.

[0096] In this embodiment of the application, the first information may include information related to indicating uplink congestion.

[0097] Understandably, by using first information that includes uplink congestion, congestion can be monitored and handled in real time. When performing random access based on this first information, the uplink congestion is taken into account, which allows for balancing network load and avoiding uplink conflicts.

[0098] In this embodiment of the application, the information on the uplink congestion situation can be second information; the second information is obtained by the network device.

[0099] Among them, network devices collect relevant uplink metrics, such as access-related information and uplink measurement information.

[0100] An AI model is deployed on the first entity to provide AI services based on the first information. The AI ​​model (e.g., the first model) can take the first information as input and perform inference analysis, considering the network conditions and impacts that occur during random access, to predict the relevant results during the random access process. The inference result of the AI ​​model is the response result generated by the first entity.

[0101] In this embodiment, the response result may include the first model output during AI model training, relevant information used by the terminal device during AI model inference (the output information differs when the network device is the first entity and when the terminal device is the first entity), and the model management result output during AI model management. Using the generated response result in real-time for the terminal device's random access process allows for timely optimization of the random access strategy, effectively reducing the access failure rate caused by network congestion or instability.

[0102] Understandably, the deployment of the AI ​​model enables automated and intelligent management of the access process, and the first entity can effectively cope with uplink congestion, providing stable and efficient support for the random access process.

[0103] In some embodiments of this application, where the first entity is deployed on a network device and the AI ​​service of the first entity includes AI model inference, the training and management of the AI ​​model are both performed on the first entity.

[0104] In some embodiments of this application, when the first entity is deployed on the first terminal device, at least one of the AI ​​model inference, AI model training, and AI model management in the AI ​​service is performed on the first entity, while all other AI services not performed on the first entity are performed on the second entity, which is deployed on the network device.

[0105] In this embodiment of the application, Table 1 shows the deployment of AI model training, AI model inference, and AI model management:

[0106] Table 1

[0107] As shown in Table 1, the deployment of AI model training, AI model inference, and AI model management includes the following eight scenarios:

[0108] Scenario 1: The training, inference, and management of the AI ​​model are all deployed on the first terminal device.

[0109] In this embodiment of the application, all AI-related functions are executed on the terminal device, including model training, real-time inference, and model management.

[0110] In some embodiments, training data is collected and prepared on the terminal device, and the model training and optimization process is performed; the trained model is used to perform real-time inference on the terminal device to generate prediction results or decisions; functions such as managing model updates, version control, and performance monitoring are all implemented on the terminal device.

[0111] Scenario 2: The training and inference of the AI ​​model are both deployed on the first terminal device, while the management of the AI ​​model is deployed on the network device.

[0112] In the embodiments of this application, model training and inference are performed on the terminal device, while model management is centralized on the network device.

[0113] In some embodiments, data acquisition and processing are performed on the terminal device, and training data is used for model training and optimization on the terminal device; the trained model is used for real-time inference and data processing; version control, updates and performance monitoring of the model are managed on the network device, which can be achieved through a remote management interface.

[0114] Scenario 3: The training and inference of the AI ​​model are both deployed on the first terminal device, while the inference of the AI ​​model is deployed on the network device.

[0115] In the embodiments of this application, the training and inference of the model are performed on the terminal device, but some inference functions are performed on network devices, such as edge computing nodes.

[0116] In some embodiments, data processing and training are performed on the terminal device, and data is transmitted to the edge node via the network for model optimization; the trained model is deployed on the terminal device and the edge node, and data transmission and coordination of inference tasks are achieved through the network device; the deployment, updates and performance monitoring of the model are centrally managed on the network device to ensure the effective execution and optimization of inference tasks.

[0117] Scenario 4: The training of the AI ​​model is deployed on network devices, while the inference and management of the AI ​​model are both deployed on the first terminal device.

[0118] In the embodiments of this application, the training of the model is performed on the network device, while the inference and management functions are executed on the terminal device.

[0119] In some embodiments, large-scale data processing and model training are performed on network devices to optimize model parameters and generate training results; the trained model is deployed to terminal devices to perform real-time inference and data processing tasks; model version control, performance monitoring and application management are performed on terminal devices to ensure stability and reliability.

[0120] Scenario 5: The training and management of the AI ​​model are both deployed on network devices, while the inference of the AI ​​model is deployed on the first terminal device.

[0121] In this embodiment of the application, the training and management of the model are performed on the network device, while the inference process is distributed across multiple terminal devices.

[0122] In some embodiments, large-scale data processing and model training are performed centrally on network devices to optimize model performance and data analysis; the trained first model is deployed to a first terminal device, and data transmission and inference tasks are coordinated through the network device; model version control, updates and performance monitoring are implemented on the network device, and inference tasks on multiple terminal devices are coordinated and managed.

[0123] Scenario 6: The training and inference of the AI ​​model are both deployed on network devices, while the management of the AI ​​model is deployed on the first terminal device.

[0124] In the embodiments of this application, model training and inference are performed on network devices, while model management is performed on terminal devices.

[0125] In some embodiments, data processing and model training are performed on network devices to optimize model performance and data analysis; the trained model is deployed to network devices and terminal devices to realize the distribution and coordination of inference tasks; model management, version control and performance monitoring are performed on terminal devices to ensure the efficient execution and management of inference tasks.

[0126] Scenario 7: The training of the AI ​​model is deployed on the first terminal device, while the management and inference of the AI ​​model are both deployed on network devices.

[0127] In the embodiments of this application, model training is performed on the terminal device, while model management and inference functions are centrally implemented on the network device.

[0128] In some embodiments, data is collected and processed on the terminal device, and the data is transmitted to a centralized training and optimization environment through the network device; the trained model is deployed on the network device to realize centralized management and coordination of inference tasks; version control, updates and performance monitoring of the model are implemented on the network device to ensure efficient execution of inference tasks and resource utilization.

[0129] Scenario 8: The training, inference, and management of the AI ​​model are all deployed on network devices.

[0130] In this embodiment, all AI-related functions are executed centrally on the network device, including model training, real-time inference, and model management.

[0131] It should be noted that the implementations of scenarios 2, 6, and 7 are not reasonable, but the possibility of implementing them in the above manner cannot be ruled out.

[0132] In some embodiments, large-scale data processing and model training are performed on network devices to optimize model performance and data analysis; the trained model is deployed on network devices to achieve real-time inference and data processing tasks; and model version control, updates, and performance monitoring are performed centrally on network devices to ensure stability and efficient operation.

[0133] Understandably, deploying AI model training, inference, and management functions across different devices allows for flexible adjustments based on needs and resource allocation. For example, centralizing training tasks on network devices leverages their powerful computing and storage resources, while deploying inference tasks on terminal devices enables faster response times and reduced latency. Distributing AI model training and inference across different devices according to task requirements helps optimize overall performance. For instance, deploying the inference process on terminal devices closer to the data source or data consumer reduces data transmission latency and bandwidth consumption. Effective resource utilization and task allocation can reduce deployment and maintenance costs. Centralizing some management functions on network devices enables more efficient monitoring and scheduling, saving manpower and material costs. In application scenarios requiring rapid response and real-time decision-making, deploying inference functions on terminal devices enables instant data processing and feedback, improving real-time performance and responsiveness. Centralizing the management and monitoring of model versions and performance on network devices strengthens security controls and data protection measures, ensuring sensitive information is not accessed or leaked without authorization. On the one hand, different deployment schemes can meet different application needs, such as the low latency requirements for edge computing and the high-efficiency requirements for large-scale data processing, thus adapting to diverse business scenarios and user needs. In general, rationally selecting and configuring the deployment methods for AI model training, inference, and management can improve overall performance while maximizing resource utilization, reducing costs, and enhancing security and responsiveness, thereby bringing users a better user experience and business benefits.

[0134] In some embodiments of this application, the first information includes: second information from the network device and / or third information from the first terminal device; the second information is related to the uplink congestion situation.

[0135] In this embodiment of the application, the second information is determined or obtained by the network device, and the third information is determined or obtained by the first terminal device.

[0136] It should be noted that when the first entity is deployed on the network device, the third information is transmitted from the second entity deployed on the first terminal device to the first entity. Similarly, when the first entity is deployed on the first terminal device, the second information is transmitted from the second entity deployed on the network device to the first entity.

[0137] In some embodiments of this application, the second information includes at least one of the following:

[0138] Initial access information; wherein, the initial access information is determined based on access frequency information and / or access time information;

[0139] Uplink measurement information used to indicate uplink congestion.

[0140] In this embodiment of the application, the initial access information is also referred to as the access information indicating successful initial access.

[0141] In the embodiments of this application, the initial access information refers to the time-related information and / or repetition number-related information of the initial successful access on each cell and / or synchronization signal block (SSB).

[0142] In this embodiment, the synchronization signal block corresponds to a beam. The terminal device can receive synchronization signal blocks transmitted on multiple beams.

[0143] For example, time-related information includes access timestamps and / or access durations. The access timestamp refers to the time point of each successful access recorded on each Cell or SSB, allowing for analysis and monitoring of the distribution of access times. The access duration refers to the duration of each recorded successful access, helping to assess the duration and potential variations of the access process. Repetition-related information includes the number of access attempts and / or the access success rate. The number of access attempts refers to the number of access attempts recorded on each Cell or SSB, including both successful and failed attempts, used to analyze the repetition rate and success rate during the access process. The access success rate is calculated based on the number of successful attempts and the total number of attempts, to assess the access efficiency and stability of the Cell or SSB.

[0144] In this embodiment, the access count information includes: Access Attempt Count. The Access Attempt Count refers to the number of times the device attempts to connect to the base station or network from its first attempt to successfully connect. This attempt process includes any one or more initial access procedures, such as sending the initial access preamble, sending the initial access MSG3, etc. The time-related information includes: Access Timestamp and / or Access Duration. The Access Timestamp refers to the recorded time of each successful access, and the Access Duration represents the duration used for each access, i.e., the time period from the start of access to the completion of connection establishment.

[0145] Understandably, initial access information helps operators or network administrators understand device access behavior and performance. By analyzing access frequency and timing information, network load, inter-device interference, and potential congestion can be assessed. Initial access information is crucial for optimizing network resource allocation, improving access strategies, and enhancing user experience.

[0146] In some embodiments of this application, the uplink measurement information includes the uplink measurement object and / or the uplink measurement result, which is determined based on the measurement quantity.

[0147] In some embodiments of this application, the uplink measurement information includes at least one of the following: uplink measurement information of each cell; uplink measurement information of each beam.

[0148] In this embodiment, the uplink measurement information of each cell records the uplink communication quality and signal strength between the terminal device and a specific cell. The uplink measurement information of each beam records the uplink communication quality and signal strength between the terminal device and a specific beam.

[0149] In this embodiment, the uplink measurement objects are various indicators and parameters used to evaluate uplink performance. The uplink measurement results are specific values ​​or states calculated or evaluated based on the uplink measurement objects, reflecting the actual performance of the current uplink.

[0150] In some embodiments of this application, the uplink measurement object includes at least one of the following:

[0151] Reference signal received power;

[0152] Signal-to-interference-plus-noise ratio;

[0153] Reference signal reception quality;

[0154] Received signal strength indication.

[0155] In this embodiment, the uplink measurement information includes measurement information for each cell and / or each beam. Exemplarily, the uplink measurement information includes: Reference Signal Received Power (RSRP), Reference Signal Received Quality (RSRQ), and Signal-to-Interference-plus-Noise Ratio (SINR). RSRP indicates the signal strength received by the device from the base station; changes in RSRP can reflect fluctuations in signal strength in the uplink, potentially due to congestion. RSRQ represents the ratio of signal strength to signal quality; changes in RSRQ can indicate the level of interference and resource utilization in the uplink. SINR indicates the strength of the received signal quality relative to environmental interference and noise; a low SINR may indicate high interference and noise in the uplink, possibly due to congestion.

[0156] In this embodiment, the received signal strength indicator can be a numerical value describing the level of uplink measurement results, or an indication of whether it is above a certain threshold. In some embodiments, the received signal strength indicator is a numerical quantification index used to represent the received signal strength; the higher the value, the stronger the signal. The signal strength indicator provides a quantitative assessment of signal strength, helping network devices understand the signal environment of the current link. Additionally, the signal strength indicator can be used as a comparison index for threshold values, indicating whether the received signal is above or below a preset threshold. By comparing thresholds, it is possible to quickly assess whether the signal quality meets expected requirements.

[0157] In the embodiments of this application, the preset threshold can be configured by the network, pre-configured, or defined by the protocol. That is, the preset threshold value is dynamically configured by the network device according to the current network status and needs; or, the preset threshold value is preset when the device is manufactured or deployed; or, the preset threshold value is specified in the communication protocol standard to ensure uniformity and compatibility between devices.

[0158] Understandably, received signal strength indicators can provide accurate signal strength assessment and monitoring through numerical quantization and threshold comparison. The application of signal strength indicators not only improves signal monitoring accuracy and network resource utilization efficiency, but also enhances network stability and robustness, supports intelligent network management, and ultimately improves the user's communication experience.

[0159] In some embodiments of this application, the uplink measurement result includes at least one of the following:

[0160] The grade value characterizing the upstream measurement result;

[0161] The result indication is determined by comparing the uplink measurement object with the preset threshold; the preset threshold is configured by the network device, or pre-configured, or preset by the protocol.

[0162] In this embodiment, the grade value characterizing the uplink measurement result is a numerical value that quantitatively describes the uplink measurement result, representing the relative level of signal quality or strength. The grade value provides a clear understanding of the uplink signal quality. The result indication, determined by comparing the uplink measurement object with a preset threshold, is a conclusive indication obtained by comparing the uplink measurement object (such as reference signal received power, signal-to-interference-plus-noise ratio, etc.) with a preset threshold. This indication is used to quickly assess whether the signal quality meets communication requirements and to make relevant decisions, such as power adjustment, resource allocation, and handover management.

[0163] Similarly, the aforementioned preset threshold values ​​are thresholds configured by the network device, or preset thresholds, or protocol-preset thresholds.

[0164] It should be understood that uplink measurement results provide a precise means of signal quality assessment through grade values ​​and result indicators. By utilizing dynamic thresholds configured in network devices, pre-configured initial thresholds, and protocol-preset standard thresholds, uplink measurement results can flexibly adapt to different network environments and requirements, ensuring communication stability and reliability while improving the network's intelligent management level and the user's communication experience.

[0165] Understandably, the aforementioned uplink measurement information is collected through network devices or base stations. By considering these parameters during the random access process, the access selection of terminal devices can be optimized, reducing connection problems and latency caused by congestion.

[0166] In some embodiments of this application, the initial access information includes at least one of the following:

[0167] One or more first access information; one or more first access information includes access count information and / or access time information reported by multiple second terminal devices that have successfully achieved initial access on each cell and / or each synchronization signal block;

[0168] At least one of the maximum, minimum, and average values ​​among multiple first access information;

[0169] One or more second access information, including access count information and / or access time information reported by the third terminal device that successfully initially accessed each cell and / or each synchronization signal block within the first time period;

[0170] At least one of the maximum, minimum, and average values ​​among multiple second access information;

[0171] Access factor information; access factor information is determined based on one or more first access information and at least one of a plurality of second access information.

[0172] In this embodiment of the application, the access information includes information related to the number of accesses and / or information related to the access time.

[0173] In the embodiments of this application, one or more first access information refers to information on the successful network access of multiple second terminal devices on multiple cells and / or multiple synchronization signal blocks.

[0174] This information helps understand the access status of different areas or signal blocks, thereby enabling network resource optimization and management. Multiple first access reports encompass data reports of multiple second terminal devices successfully accessing the network for the first time in different cells or different SSBs. This data is crucial for evaluating network coverage, signal quality, and the efficiency of device access.

[0175] In this embodiment, the maximum, minimum, or average value among multiple first access information refers to the maximum, minimum, or average value calculated for a specific indicator (such as access time, access count, etc.) among the information of multiple second terminal devices successfully accessing the network for the first time. For example, if access time information of multiple second terminal devices on multiple cells or multiple synchronization signal blocks is collected, the maximum value may be the access time of the device with the longest access time among all records; the minimum value may be the access time of the device with the shortest access time among all records; and the average value is the average of the access times of all devices, reflecting the overall average access efficiency. These statistical values ​​can help analyze and optimize the access situation on different cells or signal blocks to understand the uplink congestion or load situation.

[0176] In this embodiment, the maximum, minimum, or average value among multiple second access information refers to the statistical value of a specific indicator calculated based on the initial access success information of multiple third terminal devices in various cells or synchronization signal blocks within a first time period. For example, if access count information of multiple third terminal devices in various cells or synchronization signal blocks is collected within the first time period, then: the maximum value may be the access count of the device with the most access counts among all records; the minimum value may be the access count of the device with the fewest access counts among all records; and the average value is the average of the access counts of all devices, reflecting the overall average access frequency. These statistical values ​​can help understand the access situation of different devices in different areas within a specific time period, in order to understand the uplink congestion or load situation within the first time period.

[0177] In some embodiments of this application, the first time period includes at least one of the following:

[0178] Preset time period;

[0179] The first cycle related to the system information update cycle;

[0180] The second cycle related to the transmission cycle of the synchronization signal block;

[0181] The third cycle is related to the duration of network configuration.

[0182] In this application embodiment, the following time periods are defined: A preset time period: This is a fixed time length set in advance for planning and executing specific operations or events. For example, it can be set to execute a task once per hour or send a report once per day. A first period related to the system information update cycle: This refers to the time period related to the update frequency of system parameters or configuration information. A second period related to the transmission cycle of synchronization signal blocks: Synchronization signal blocks (SSBs) are key signals used for device positioning and synchronization in 5G networks. The second period refers to the time period related to the SSB transmission frequency. Typically, SSB transmission can be performed at specific time intervals, such as every few milliseconds or seconds. A third period related to the time length of network device configuration: This refers to the time length associated with network device configuration parameters or policies. Different time lengths may be set during communication based on different network requirements and service types, such as the time period for QoS (Quality of Service) configuration.

[0183] Understandably, for the first period, it allows for precise scheduling of tasks and events, such as periodically executing critical operations or generating reports. This effectively utilizes resources and avoids wasting them in unnecessary waiting or idle states. For the second period, it ensures that system parameters and configuration information are updated promptly to reflect the latest network status and needs. Timely updates help quickly identify and fix potential problems, improving stability and reliability. For the third period, it ensures that all devices receive synchronization signal blocks at the same time, ensuring coordination and synchronization among devices in the network, improving the accuracy and precision of location services, which is particularly important for applications and services requiring device location information. For the fourth period, configuration parameters can be adjusted according to network needs and traffic patterns to improve network flexibility and adaptability. Through periodic configuration adjustments, it ensures that the Quality of Service (QoS) meets the requirements of users and applications. In summary, setting and optimizing these time periods helps improve network efficiency, stability, and performance, ensuring effective response to dynamically changing needs and environments.

[0184] In this embodiment, the access factor information is a specific identifier or configuration determined based on one or more collected first access information or one or more second access information. These identifiers can be used to reflect numerical configurations related to the time of successful access and / or the number of repetitions, such as a number between 0 and 10, where a larger value indicates greater access difficulty. For example, if multiple first access information or multiple second access information are collected, a numerical configuration can be derived by analyzing and statistically analyzing this information. This configuration can be used to describe the ease or difficulty of successful access. For instance, fewer access records on a cell or synchronization block, and / or shorter access success times for users, and / or fewer access attempts may result in a lower value, indicating easier access; conversely, fewer successful access records in many cells or synchronization blocks within a specific time period, and / or longer access times, and / or more access attempts may result in a higher value, indicating more difficult access. This access factor information can help understand and evaluate the network's access situation.

[0185] In this embodiment, the access factor information further includes access indication information, wherein the access indication information is used to indicate whether the number of access attempts and / or access time exceeds a preset threshold. The preset threshold may be configured by the network, pre-configured, or specified by the protocol; this embodiment does not impose any restrictions.

[0186] In some embodiments of this application, the third information includes sensing information, which includes at least one of the following:

[0187] The moving speed of the first terminal device;

[0188] Location-related information of the first terminal device;

[0189] Predicted path information for the first terminal device.

[0190] In this embodiment, the moving speed of the first terminal device refers to the speed at which the terminal device moves. Rapid movement of the first terminal device may require more frequent cell handovers, therefore this factor needs to be considered during the random access process of the first terminal device. The location-related information of the first terminal device includes its current geographical coordinates, latitude and longitude, etc. By obtaining the location-related information of the first terminal device, better cell selection can be achieved during the random access process. The predicted path-related information of the first terminal device involves the predicted path of the device's future movement. Predicting the movement range in advance with the predicted path information can provide better cell selection results for the random access process, ensuring seamless handover of the first terminal device during movement.

[0191] In some embodiments of this application, the sensing information is determined by the communication parameters of the terminal device; the communication parameters include at least one of the following:

[0192] The value of the reference signal received power;

[0193] The change in the received power of the reference signal;

[0194] Its own sensor information;

[0195] Its own location information.

[0196] In this embodiment, the reference signal received power value refers to the power level of the reference signal transmitted by the base station received by the first terminal device. A higher reference signal received power value generally indicates that the device is closer to the base station or is in a good signal coverage area. This value can be used to estimate the distance from the device to the base station or to measure the quality of the received signal within the coverage area.

[0197] In this embodiment, the change in reference signal received power reflects the change in reference signal received power over time. Power changes can be caused by factors such as device movement and changes in the surrounding environment (e.g., building obstruction or multipath effects). Monitoring power changes can help identify the device's motion state and environmental changes, thereby adjusting signal processing and resource allocation.

[0198] In this embodiment, the first terminal device may be equipped with various sensors, such as accelerometers, gyroscopes, and magnetometers, utilizing its own sensor information. These sensors can provide information on the device's physical motion state, such as acceleration, angular velocity, and direction. This data can be used to estimate the device's trajectory and speed, thereby affecting mobility management and location services.

[0199] In this embodiment, the first terminal device can obtain location information via GPS (Global Positioning System), BeiDou Navigation Satellite System, or network-based positioning services using its own positioning information. The positioning information includes the device's current geographic coordinates, typically expressed as longitude and latitude. This information is crucial for providing accurate location services, navigation, geolocation, and location-based services.

[0200] In some embodiments of this application, the third information further includes at least one of the following:

[0201] Downlink measurement information from the first terminal device;

[0202] Slice information;

[0203] Scene information;

[0204] Business information.

[0205] In this embodiment, the downlink measurement information of the first terminal device includes downlink signal quality indicators received by the device, such as signal strength index (RSSI), signal-to-noise ratio (SNR), and bit error rate (BER), used to evaluate the device's receiving performance and network quality. Slicing information refers to the specific frequency band or range used by the device in communication, used to distinguish and manage different communication channels or service types. Scene information involves information about the environment and operating scenario of the device, such as indoor / outdoor, urban / rural, etc., which can affect the signal propagation characteristics and service requirements of communication. Service information involves information about the specific service or application currently being executed by the device, such as video traffic, voice calls, data transmission, etc., which can affect network resource allocation and service priority.

[0206] In some embodiments of this application, the first information further includes: fourth information from the network device, the fourth information representing the configuration information for cell reselection.

[0207] In this embodiment of the application, the configuration information for cell reselection includes: S / R criterion configuration and / or slice configuration.

[0208] In this embodiment, the cell reselection configuration information includes key parameters and criteria in the mobile communication system, used for handover and reselection between different cells. For example, the cell reselection configuration information may include: priority configuration of the serving cell and neighboring cells, which define the conditions and priorities for handover between the serving cell and surrounding neighboring cells. For instance, the device may decide when to hand over to another cell based on signal strength, quality, and interference levels. Reselection criteria configuration, where the R criteria specify that the device should search for and select a new cell when the quality of the current serving cell deteriorates or other conditions are not met. Criteria may include signal strength thresholds, SINR thresholds, etc. Slice configuration, in network slicing technology, defines the service quality requirements and priorities for different user groups or service types, including cell selection and resource allocation.

[0209] Understandably, this configuration information is crucial for ensuring that mobile devices maintain stable connections and efficient data transmission in different network environments.

[0210] The following section explains the input and output information of the AI ​​model based on the AI ​​service and its deployment.

[0211] The first type of AI service includes the training of AI models.

[0212] In some embodiments of this application, when the AI ​​service includes training of an AI model, the first information comes from a training dataset; the training dataset includes at least one of the following: third information of a first terminal device as a sample, second and fourth information obtained by a network device corresponding to the first terminal device, excitation parameters, and sample access information of a first terminal device that has been connected.

[0213] In the embodiments of this application, the activation parameters may be included in the training dataset or determined during the training process; the embodiments of this application do not impose any restrictions.

[0214] In some embodiments of this application, when the AI ​​service is used for training an AI model, the first information serves as the input data for training the initial AI model to obtain the first model. The output data of the AI ​​model is used to combine the sample access information of the first terminal device in the training dataset to adjust the initial AI model and obtain the first model. The first model is the response result.

[0215] In this embodiment of the application, when the AI ​​service is used for training an AI model, the initial input information of the AI ​​model includes:

[0216] 1) The third information of the first terminal device that has been connected as a sample includes at least one of the following: the moving speed of the first terminal device; the location-related information of the first terminal device; the predicted path-related information of the first terminal device; the downlink measurement information of the first terminal device; the slice information; the scene information; and the service information.

[0217] 2) The second information acquired by the network device includes at least one of the following: access information indicating successful initial access; wherein the access information includes: access count-related information and / or time-related information; uplink measurement information indicating uplink congestion. The access information includes at least one of the following: multiple first access information; the multiple first access information is reported by multiple first terminal devices that have successfully achieved initial access in each cell and / or each synchronization signal block; a first maximum value, a first minimum value, or a first average value among the multiple first access information; a second maximum value, a second minimum value, or a second average value among the multiple second access information; wherein the multiple second access information is reported by multiple second terminal devices that have successfully achieved initial access in each cell and / or each synchronization signal block within a first time period; access factor information; the access factor information is determined based on one or more first access information or multiple second access information.

[0218] 3) The fourth information obtained by the network device includes: the cell reselection configuration information, including: S / R criterion configuration and / or slice configuration.

[0219] In some embodiments of this application, when the first entity is deployed on the first terminal device, the second information and the fourth information come from the second entity, which is deployed on the network device.

[0220] In this embodiment, when the first entity is deployed on the first terminal device, the second entity deployed on the network device sends second information and fourth information to the first terminal device. Correspondingly, the first terminal device receives the second information and fourth information sent by the second network entity.

[0221] In this embodiment of the application, the output data of the initial AI model includes:

[0222] The selection results of the camping cell to be occupied;

[0223] The result of the beam selection for the desired dwell time;

[0224] Intermediate access result parameters; Intermediate access result parameters are used by the first terminal device to determine the cell or beam to be camped during the random access process.

[0225] In some embodiments, when a first terminal device performs random access, the initial AI model predicts one or more suitable cells based on received third information (e.g., device status, location, etc.) and second information (e.g., network load, cell quality, etc.). These recommendations are typically based on algorithmic analysis of network conditions and device requirements to ensure that the device can stably connect to the network and obtain good quality of service.

[0226] In some embodiments, when using beamforming technology, the initial AI model may recommend a set of beams as options for the devices to be connected. Beamforming optimizes the direction and power of signal transmission to improve data transmission rates and network coverage. By selecting appropriate beams, optimization can be achieved based on device location, network capacity, and beam quality, thereby enhancing connection stability and efficiency.

[0227] In some embodiments, intermediate access outcome parameters are used to guide the behavior of the first terminal device during the access process. These parameters may include access threshold settings (such as access signal strength thresholds), access timing optimization (such as optimal access time), frequency selection (such as preferred frequency bands), connection priority, etc. Optimization of these parameters can help the device complete access more quickly and reliably in complex wireless environments.

[0228] In some embodiments of this application, the intermediate access result parameters include at least one of the following:

[0229] Priority information of any one of the cell, frequency, or beam to be accessed; the priority information, when the first condition is met, is used to determine the cell or beam to which the first terminal device should camp during the random access process.

[0230] Information on adjusting measurement parameters.

[0231] In this embodiment, the cell, frequency, or beam to be accessed is specified as the target cell, frequency, or beam that the device should prioritize during the access process. Provided other conditions are met (e.g., network load, quality of service requirements), the priority information determines the order in which the device selects the cell or beam to camp on. Priority information helps the device make the optimal choice among multiple options to improve access success rate and quality of service.

[0232] In some embodiments of this application, the first condition includes at least one of the following:

[0233] The range of priority adjustments for network device configurations;

[0234] The first cell or first frequency configured for network devices has the highest priority.

[0235] The conditions under which network device configurations allow priority adjustments.

[0236] In this embodiment, the network device specifies an adjustable priority range for each cell, frequency, or beam. This range can range from the lowest to the highest priority, allowing the network device to make flexible adjustments based on actual conditions. The network device sets a certain priority range to adjust the priority of different cells or frequencies. This range is typically defined based on network topology, service policies, and user needs. For example, some cells may be set to high priority to provide services to specific users, while other cells may be set to low priority for general data transmission.

[0237] In the embodiments of this application, network configuration may specify certain cells or frequencies that can be prioritized. This configuration is typically based on network topology, service policies, or predicted demand to ensure that critical service areas or frequency bands receive priority access to resources. In some embodiments, network configuration is typically based on network load, service demand, or quality of service objectives. For example, in high-traffic areas or important service areas, the priority of specific cells or frequencies can be adjusted to ensure that users can obtain stable and high-quality service.

[0238] In the embodiments of this application, network devices may have certain specific conditions that allow for priority adjustments. These conditions may encompass factors such as network load, user demand, mobility speed, and service type. For example, during periods of high load or under specific service demands, the priority of certain cells or frequencies may be increased. In some embodiments, this includes, but is not limited to, network load levels, user device mobility speed, and the demand for specific service types. For example, during periods of low network load or for specific user groups, the priority of certain cells or frequencies may be decreased to achieve efficient resource utilization.

[0239] Understandably, the UE needs to further determine the selection of the camping cell and / or SSB based on the priority adjustment results, and this priority adjustment must be performed under network control. The purpose of priority adjustment is to optimize resource allocation in a dynamic network environment, improve network capacity utilization, and enhance user experience. Based on the priority information sent by the network, the UE selects a suitable camping cell or SSB to ensure timely connection to the most suitable cell under different conditions, thereby achieving more reliable and efficient communication services.

[0240] In this embodiment, the adjustment information for measurement parameters includes: the frequency or beam configuration to be measured, threshold configuration, and priority information. The frequency or beam configuration to be measured specifies the frequency or beam range that the device should consider when performing signal measurements. This can be dynamically adjusted based on current network load, device location, and communication requirements to ensure that the device selects the optimal frequency or beam during access. The threshold configuration specifies the threshold values ​​in the measurement parameters, such as the minimum acceptable signal strength threshold or the minimum requirements for quality indicators (such as RSRQ, SINR). These threshold values ​​help the device assess the quality of surrounding signals, thereby deciding whether to perform access or handover operations. The priority information specifies the priority order that the device should consider when making measurement and access decisions. Priority information can be configured according to different service quality requirements or network policies to ensure that critical services or high-priority users have priority access to network resources.

[0241] Understandably, the adjustment information for measurement parameters is output through the AI ​​model (the first model) and provided to the first entity, enabling it to intelligently manage the measurement process and make reasonable resource allocation and access decisions based on real-time network conditions and user needs. This intelligent adjustment helps improve the accuracy of access.

[0242] In some embodiments of this application, the access method further includes: performing random access of the first terminal device based on the output data.

[0243] In this embodiment of the application, when the first terminal device performs random access based on the output data, it usually determines the specific cell or beam to be accessed based on the output data of the previous model.

[0244] For example, the first terminal device (UE) needs to perform an initial access procedure upon startup. Without pre-allocated resources, the UE sends a Random Access Request (RACH request) to the base station (NodeB or gNB). The output data from the first model may include cell selection results or beam selection results, which will serve as the basis for the UE to select a suitable cell or beam during the random access procedure. If the response contains priority or adjustment information, the UE may use this information to decide whether to prioritize certain cells or beams during random access.

[0245] Understandably, by using the output data of the initial AI model to perform random access, the access result of the first terminal device during the random access process can be predicted. In order to compare the predicted access result with the actual sample access information of the first terminal device, the initial AI model can be adjusted, and thus the trained first model can be determined.

[0246] In some embodiments of this application, when the AI ​​service is used for training an AI model, incentive parameters (reward information) are determined during the random access process of the first terminal device; wherein,

[0247] The activation parameters are used to train the initial AI model by combining the sample access information of the first terminal device in the training dataset, and to determine the first model.

[0248] In this embodiment, the first model is determined by training an initial AI model based on excitation parameters and the sample access information of the first terminal device in the training dataset; wherein,

[0249] The activation parameters are obtained from the training dataset, or, in the case of training an AI model for an AI service, are determined during random access to the first terminal device.

[0250] In this embodiment, when the AI ​​service is used to train an AI model, determining the incentive parameters (reward information) is a crucial step. The incentive parameters are used to evaluate the performance of the first terminal device during random access and, in conjunction with the training dataset, to train an initial AI model to determine the first model.

[0251] In some embodiments, during the random access process of the first terminal device, a series of operations and feedback occur. For example, the terminal device attempts to access the network, the network device responds to the access request, and the device ultimately succeeds or fails to access the network. Incentive parameters are used to evaluate the performance of these operations and feedback. These parameters can reflect metrics such as access success rate, access latency, and number of access attempts. For example, information such as the time to successful access, the number of attempts, signal strength (e.g., RSRP, RSRQ, SINR), and resources used during the access process can all be used as incentive parameters.

[0252] It should be noted that the incentive parameters can be numerical, representing specific performance during the access process. For example, the reward for successful access can be set to a positive value, and the penalty for failed access can be set to a negative value. The specific values ​​of these parameters can be set according to specific network requirements and objectives. For example, a reward of +10 points for successful access, a penalty of -5 points for failed access, and a penalty of -1 point for every second increase in access latency. The initial AI model is trained using a training dataset. During training, the AI ​​model adjusts its strategy based on the incentive parameters to maximize the reward (minimize the penalty). The training process can employ reinforcement learning, where the model gradually optimizes its access strategy through interaction with the environment. Through continuous training and optimization, the initial AI model gradually evolves into the first model. The first model can make more accurate and efficient access decisions based on sample access information and incentive parameters in the training dataset. The trained first model can be deployed in a real-world environment to guide the first terminal device in selecting the optimal cell or beam during random access, improving access success rate and network efficiency.

[0253] For example, suppose there is an initial AI model that needs to learn how to optimize the access strategy during a random access process of a first terminal device. Sample access information and incentive parameters are collected: Sample access information: Device A attempts to access cell X with a signal strength of -85dBm, attempts 3 times, and finally successfully accesses after 5 seconds. Incentive parameters: Successful access reward +10 points, access attempt penalty -3 points (1 point per attempt), access delay penalty -5 points (1 point per second of delay). Calculate the total incentive: +10 - 3 - 5 = +2 points. Therefore, we can obtain: Sample 1: Access information is {cell X, -85dBm, 3 attempts, 5 seconds}, incentive parameter is +2 points. Further, a reinforcement learning algorithm is used to train the model, allowing the model to adjust the access strategy according to the incentive parameters to improve the total incentive score. For example, the model might learn to reduce the number of attempts when the signal strength is below -90dBm, or prioritize accessing cells with better signal strength. After multiple rounds of training and optimization, the initial AI model becomes more intelligent and can better predict and select the optimal access strategy. The first model is deployed to the first entity, providing real-time guidance to the first terminal device to make optimal decisions during random access, thereby improving access efficiency and success rate. Through this process, the AI ​​model can continuously optimize and improve access strategies, enabling the first terminal device to achieve a better experience and performance during random access.

[0254] In some embodiments of this application, the second information and / or the fourth information are sent by the second entity via a broadcast message or an RRC message.

[0255] In this embodiment of the application, a second entity deployed on a network device sends second information and / or fourth information to a first entity deployed on a first terminal device.

[0256] In some embodiments of this application, when the first entity is deployed on the first terminal device, the second information and the fourth information come from the second entity, which is deployed on the network device.

[0257] In this embodiment of the application, the training of the AI ​​model is deployed on the first terminal device.

[0258] In some embodiments of this application, when the training of the AI ​​model is deployed on the first terminal device, the second information and / or the fourth information are sent by the second entity through broadcast messages or Radio Resource Control (RRC) messages.

[0259] It should be noted that when the AI ​​model is trained and deployed on the first terminal device, the second information in the AI ​​model's input information is obtained by the network device and sent to the first terminal device. The third information is obtained by the first terminal device itself.

[0260] In some embodiments of this application, when the first entity is deployed on a network device, the third information comes from a second entity, which is deployed on the first terminal device.

[0261] In this embodiment of the application, the training of the AI ​​model is deployed on a network device.

[0262] In some embodiments of this application, the output data includes at least one of the following: priority information of any one of the cell, frequency, or beam to be accessed; and adjustment information of measurement parameters.

[0263] In this embodiment of the application, the output data is used to guide the terminal device in selecting and optimizing strategies when accessing the network.

[0264] In some embodiments, priority information for any one of the cell, frequency, or beam to be accessed is used to indicate the preferred cell, frequency, or beam for the terminal device during random access, thereby optimizing access success rate and communication quality. Specifically, cell priority information indicates the access priority of each cell in the network, helping the terminal device select the best cell for access; frequency priority information indicates the priority of available frequency bands, helping the terminal device select the best frequency band for access; and beam priority information indicates the priority of different beams, helping the terminal device select the best beam under beamforming technology. Measurement parameter adjustment information is used to adjust the measurement parameters of the terminal device to make more accurate access decisions.

[0265] It should be understood that by prioritizing high-priority cells, frequencies, or beams, access success rates and communication stability can be improved, and by reasonably allocating priorities, network load can be balanced to avoid overloading of certain cells or frequency bands.

[0266] Understandably, the output data includes priority information for the cell, frequency, or beam to be accessed, as well as adjustment information for measurement parameters. This information works together to help the terminal device make optimized decisions during random access, improving access efficiency and communication performance, and ultimately enhancing the user's communication experience.

[0267] In some embodiments of this application, the network device sends output data to the first terminal device, and the output data is used by the first terminal device when performing random access.

[0268] In some embodiments of this application, the access method further includes:

[0269] When the AI ​​service is used for training the AI ​​model, it receives the incentive parameters reported by the first terminal device; the incentive parameters are determined during the random access process of the first terminal device.

[0270] Based on the activation parameters and combined with the sample access information of the first terminal device in the training dataset, the initial AI model is trained to determine the first model;

[0271] Send the first model as a response result to the first terminal device.

[0272] In this embodiment of the application, when the AI ​​service is for training an AI model, the incentive parameters are reported by the first terminal device or obtained from the training dataset during the random access process of the first terminal device.

[0273] The activation parameters are used to train the initial AI model by combining the sample access information of the first terminal device in the training dataset, and to determine the first model.

[0274] In this embodiment, when the AI ​​service is used for training an AI model, the AI ​​model is trained and optimized by receiving, processing, and utilizing data reported by the terminal device. Finally, an optimized first model is generated and sent back to the terminal device to improve its access efficiency and communication performance. In some embodiments, incentive parameters reported by the first terminal device are received. These incentive parameters are determined during the random access process of the first terminal device. The received incentive parameters are combined with the sample access information of the first terminal device in the training dataset to train the initial AI model. Through training and optimization, a new AI model, namely the first model, is generated. This model can more accurately predict the access success rate of the terminal device and provide optimized access strategies. The trained and optimized first model is sent back to the first terminal device as a response result. After receiving the optimized model, the first terminal device can improve its access efficiency and communication performance based on the strategies and suggestions provided by the model.

[0275] Understandably, when the AI ​​service is used to train the AI ​​model, the initial AI model is trained and optimized by receiving the stimulus parameters reported by the first terminal device and combining them with the sample access information in the training dataset, generating an optimized first model. This first model can provide a more accurate access strategy and is sent back to the terminal device via the network, helping it improve access efficiency and communication performance, thereby enhancing the user experience.

[0276] In some embodiments of this application, the excitation parameters include at least one of the following:

[0277] The results of sample access;

[0278] The access time or number of access repetitions of the sample during the network access process;

[0279] The energy consumed in connecting or measuring a sample.

[0280] In this embodiment, the sample access result describes whether the first terminal device successfully accesses the network. That is, it records whether the device successfully connects to the network during a random access process. This can be represented as a binary variable (success or failure) or as a scoring system reflecting access quality. For example, a successful access result serves as a positive incentive, helping the AI ​​model identify and strengthen effective access strategies; a failed access result serves as a negative incentive, helping the model identify and avoid ineffective access strategies.

[0281] In this embodiment, the access time during network access describes the time required for the first terminal device to successfully access the network from the start of its access attempt. For example, a shorter access time indicates a more efficient access process, and the model will tend to optimize strategies that reduce access time, thereby improving access efficiency. The number of access attempts during network access describes how many times the first terminal device needs to attempt network access successfully. For example, fewer access attempts indicate a smoother access process, and the model will tend to optimize strategies that reduce the number of attempts, thereby improving access success rate and user experience.

[0282] In this embodiment, the energy consumed by the sample during access or measurement describes the energy consumed by the first terminal device during network access. For example, lower energy consumption indicates a more energy-efficient access process. The model tends to optimize strategies that reduce energy consumption; this parameter is particularly important for battery-powered devices as it can extend battery life.

[0283] It's worth noting that by comprehensively considering access results, access time, number of access repetitions, and energy consumption, the AI ​​model can more comprehensively optimize access strategies and improve overall network access efficiency. The model can dynamically adjust access strategies based on incentive parameters under different network environments to adapt to changing network conditions, thereby improving access success rate and user satisfaction. Furthermore, the focus on energy consumption ensures that the model not only optimizes access success rate but also considers device battery life, providing longer-lasting service.

[0284] Understandably, when training an AI model using AI services, incentive parameters (reward information) are determined and used to train the initial AI model to identify the first model. By training the AI ​​model using these incentive parameters, the random access strategy of the first terminal device can be optimized, significantly improving the access success rate. The model can identify and select the optimal access conditions, thereby reducing access failures. Incentive parameters can include penalty information for access latency; by optimizing the access strategy, the model can reduce access latency and improve user experience. Terminal devices can access the network faster, ensuring timely data transmission and communication. Through intelligent selection of access points and optimization of access strategies, the AI ​​model can effectively utilize network resources, reducing spectrum waste and signal interference, which helps improve the overall network resource utilization and performance. In summary, by introducing incentive parameters into AI model training, the network can achieve intelligent and adaptive access strategy optimization. This not only improves access efficiency and user experience but also significantly enhances network resource utilization, reduces operating costs, and supports a wide range of application scenarios.

[0285] The second type of AI service includes inference from AI models.

[0286] In some embodiments of this application, when the AI ​​service includes inference of an AI model, the first information comes from an inference dataset; the inference dataset includes: third information of the first terminal device to be accessed, and at least one of second and fourth information obtained by the network device corresponding to the first terminal.

[0287] In some embodiments of this application, when the AI ​​service is the inference of the AI ​​model, the first information is used as the input data of the first model, and the response result is the output data of the first model.

[0288] In some embodiments of this application, when the AI ​​service includes inference of an AI model, the first information comes from an inference dataset; the inference dataset includes: third information of the first terminal device to be accessed, and second and fourth information obtained by the network device corresponding to the first terminal device.

[0289] In this embodiment of the application, the input information of the first model includes:

[0290] 1) The third information of the first terminal device to be accessed includes at least one of the following: the moving speed of the first terminal device; the location-related information of the first terminal device; the predicted path-related information of the first terminal device; the downlink measurement information of the first terminal device; the slice information; the scene information; and the service information.

[0291] 2) The second information acquired by the network device corresponding to the first terminal device includes at least one of the following: access information indicating successful initial access; wherein the access information includes: access count-related information and / or time-related information; uplink measurement information indicating uplink congestion. The access information includes at least one of the following: multiple first access information; the multiple first access information is reported by multiple first terminal devices that have successfully achieved initial access in each cell and / or each synchronization signal block; a first maximum value, a first minimum value, or a first average value among the multiple first access information; a second maximum value, a second minimum value, or a second average value among the multiple second access information; wherein the multiple second access information is reported by multiple second terminal devices that have successfully achieved initial access in each cell and / or each synchronization signal block within a first time period; access factor information; the access factor information is determined based on one or more first access information or multiple second access information.

[0292] 3) The fourth information obtained by the network device includes: the cell reselection configuration information, including: S / R criterion configuration and / or slice configuration.

[0293] In this embodiment, when the first entity is deployed on the first terminal device, the second entity deployed on the network device sends second information and fourth information to the first terminal device. Correspondingly, the first terminal device receives the second information and fourth information sent by the second network entity.

[0294] In this embodiment of the application, the output data (i.e., the response result) of the first model includes:

[0295] The selection results of the camping cell to be occupied;

[0296] The result of the beam selection for the desired dwell time;

[0297] Intermediate access result parameters; Intermediate access result parameters are used by the first terminal device to determine the cell or beam to be camped during the random access process.

[0298] In some embodiments of this application, the intermediate access result parameters include at least one of the following:

[0299] Priority information of any one of the cell, frequency, or beam to be accessed; the priority information, when the first condition is met, is used to determine the cell or beam to which the first terminal device should camp during the random access process.

[0300] Information on adjusting measurement parameters.

[0301] In this embodiment, the cell, frequency, or beam to be accessed is specified as the target cell, frequency, or beam that the device should prioritize during the access process. Provided other conditions are met (e.g., network load, quality of service requirements), the priority information determines the order in which the device selects the cell or beam to camp on. Priority information helps the device make the optimal choice among multiple options to improve access success rate and quality of service.

[0302] In some embodiments of this application, the first condition includes at least one of the following:

[0303] The range of priority adjustments for network device configurations;

[0304] The first cell or first frequency configured for network devices has the highest priority.

[0305] The conditions under which network device configurations allow priority adjustments.

[0306] In some embodiments of this application, random access to the first terminal device is performed based on the response result.

[0307] In this embodiment of the application, when the first terminal device performs random access based on the response result, it typically determines the specific cell or beam to access based on the results output by the previous model. Random access refers to the process by which a device temporarily accesses the network through network signaling without a predetermined resource allocation.

[0308] In some embodiments of this application, when the first entity is deployed on the first terminal device, the second information and the fourth information come from the second entity, which is deployed on the network device.

[0309] In this embodiment, the inference of the AI ​​model is deployed on a first terminal device.

[0310] In some embodiments of this application, when the inference of the AI ​​model is deployed on the first terminal device, the second information and / or the fourth information are sent by the second entity via broadcast messages or Radio Resource Control (RRC) messages.

[0311] It should be noted that when the AI ​​model's inference is deployed on the first terminal device, the second and fourth pieces of information in the input information of the first model are obtained by the network device and sent to the first terminal device. The third piece of information is obtained by the first terminal device itself.

[0312] In some embodiments of this application, when the first entity is deployed on a network device, the third information comes from a second entity, which is deployed on the first terminal device.

[0313] In this embodiment of the application, the inference of the AI ​​model is deployed on a network device.

[0314] In this embodiment of the application, when the inference of the AI ​​model is deployed on a network device, the network device sends a response result to the first terminal device.

[0315] In some embodiments of this application, the response result is carried in a broadcast message or an RRC message.

[0316] In some embodiments of this application, the output data includes at least one of the following: priority information of any one of the cell, frequency, or beam to be accessed; and adjustment information of measurement parameters.

[0317] In this embodiment of the application, the output data is used to guide the terminal device in selecting and optimizing strategies when accessing the network.

[0318] It should be noted that the principles of AI model training and AI model inference are the same in terms of input information and output data, which will not be elaborated here.

[0319] The third type of AI service includes the management of AI models.

[0320] In some embodiments of this application, the management of the AI ​​model includes: management of the training service and / or management of the inference service, wherein the training service is training based on the AI ​​model and the inference service is inference based on the AI ​​model.

[0321] In the embodiments of this application, the management of the AI ​​model involves the supervision and optimization of the training and inference services to ensure the performance and applicability of the model.

[0322] In some embodiments, the training service involves managing the training process of the AI ​​model to ensure that the model can learn and optimize its performance. The inference service involves managing the inference process of the AI ​​model in practical applications to ensure that the model can perform its prediction tasks accurately and efficiently.

[0323] For example, the training service includes one or more of the following: data collection and preprocessing, model training and optimization, training process monitoring and version control and deployment. The inference service includes one or more of the following: inference request processing, inference process monitoring, feedback collection and processing, and model updates and maintenance.

[0324] In some embodiments of this application, the first information further includes: a second condition; the second condition serves as a judgment condition for model management when managing the AI ​​model.

[0325] In this embodiment, the second condition in the first information plays a crucial role in AI model management. The second condition is used to determine whether the AI ​​model needs management and adjustment. These conditions help the device monitor and optimize the performance of the AI ​​model in real-time operation.

[0326] In some embodiments of this application, the second condition includes at least one of the following:

[0327] The speed is greater than or equal to the speed threshold;

[0328] Location-related information is greater than or equal to the location change threshold;

[0329] The value of the reference signal received power is greater than or equal to the first power threshold;

[0330] The change in the received power of the reference signal is greater than or equal to the second power threshold;

[0331] Predicted path-related information characterizes the line-of-sight path scenario;

[0332] The number of access attempts exceeds the access attempts threshold.

[0333] Access time-related information exceeds the access time threshold;

[0334] Downlink measurement information does not exceed the downlink measurement threshold.

[0335] Understandably, adjusting the model at appropriate times to adapt to complex scenarios such as high-speed movement and significant location changes significantly improves the device's access success rate. Under high load or poor signal conditions, adjusting the model optimizes network resource allocation, ensuring efficient device access. This reduces access failures and latency, improving user experience and satisfaction, especially in mobile scenarios. Dynamically adjusting the model based on real-time data and environmental changes enhances its adaptability and flexibility, ensuring optimal performance across different scenarios. Continuous monitoring and adjustments maintain efficient network operation and stable performance, improving overall network service quality.

[0336] Understandably, on the one hand, by setting and monitoring specific second conditions, it's possible to accurately identify when AI model management and adjustments are needed, thereby optimizing network access decisions and improving access success rate and speed. On the other hand, adjusting the model based on the actual situation of terminal devices allows for better adaptation to changing network environments, enhancing the stability and reliability of network access. Furthermore, real-time monitoring and model adjustment enable dynamic adjustments to access strategies based on different network and device conditions, optimizing resource utilization and improving overall network performance. Finally, the application of second conditions allows for intelligent decision-making based on specific environments and conditions, enhancing adaptability and user experience.

[0337] In some embodiments of this application, when the first entity is deployed on a network device, the third information comes from a second entity, which is deployed on the first terminal device; when the AI ​​service is the management of an AI model, the method further includes:

[0338] Send the second condition to the first terminal device;

[0339] The device receives at least one of the third access information and third information reported by the first terminal device; the third access information and third information are determined by the first terminal device after establishing an access connection based on the second condition.

[0340] In some embodiments of this application, the third access information and the third information are reported to the network device by the first terminal device when the second condition is met based on at least one of the third information, the second information from the network device, and the fourth information, and the access is successful.

[0341] In other words, after the network device sends the second condition, the second information, and the fourth information to the first terminal device, the first terminal device can report the third access information and the third information to the network device based on whether at least one of the third information, the second information, and the fourth information meets the second condition. If at least one of the third information meets the second condition and the first terminal successfully accesses the network, the first terminal device can report the third access information and the third information to the network device.

[0342] In some embodiments of this application, the access method further includes:

[0343] Model management is performed based on at least one of the third access information and third information to determine the response result; the response result characterizes the model management result.

[0344] In some embodiments of this application, where at least one of the training and inference of the AI ​​model is performed by a second entity, the method further includes sending a response result to the second entity.

[0345] It should be noted that, in this embodiment of the application, after the first terminal device sends the third access information and the third information to the network device, the network device can, based on the third access information and the third information, determine again whether the second condition is met, and according to the determination result, perform model management to determine whether to use the AI ​​model or switch the AI ​​model, and send the management result to the first terminal device.

[0346] In some embodiments of this application, the third access information includes: the time used for access and the number of access repetitions.

[0347] In some embodiments of this application, the access method further includes: carrying the response result in a broadcast message or an RRC message.

[0348] In some embodiments of this application, AI model training and AI model inference are performed in a first entity, and the model's usage identifier indicates that random access is allowed using the AI ​​model.

[0349] In some embodiments of this application, when the first entity is deployed on the first terminal device, the second information, the fourth information, and the second condition come from the second entity, which is deployed on the network device; when the AI ​​service is the management of the AI ​​model, the response result is determined based on the second information, the third information, the fourth information, and the second condition, and is performed for model management.

[0350] In this embodiment of the application, random access is performed based on the response result to determine the access result; if the access result indicates successful access, the response result is sent to the second entity (i.e., the network device).

[0351] It should be noted that the first terminal device needs to determine whether the second condition is met based on the second, third, and fourth information. If the second condition is met, it determines whether to use the AI ​​model or switch the AI ​​model. If the first terminal device successfully accesses the network based on the response result, it sends the model management result to the network device.

[0352] In this embodiment, if the moving speed of the first terminal device exceeds a preset speed threshold, it may affect the stability and performance of network access. Therefore, it is necessary to trigger AI model management to adjust the model's parameters or strategies. For example, if the first terminal device is moving at high speed, such as on a vehicle or high-speed train, and its speed exceeds the threshold, model management is triggered.

[0353] In this embodiment, if the location change of a terminal device (e.g., the first terminal device) exceeds a preset location change threshold, it may affect the network access performance. Therefore, AI model management is needed to adjust the access strategy to ensure connection stability. For example, when a terminal device moves from one location to another, such as from indoors to outdoors, the location change exceeds the threshold, triggering model management.

[0354] In this embodiment of the application, if the power of the reference signal received by the terminal device exceeds a preset first power threshold, it may indicate that the signal is strong, and the access strategy can be optimized.

[0355] For example, if a terminal device is in an area with strong signal and the received power exceeds a threshold, model management is triggered.

[0356] In this embodiment, if the power change of the reference signal received by the terminal device exceeds a preset second power threshold, it may indicate significant signal fluctuations, requiring real-time adjustments to maintain connection stability. For example, in areas with unstable signal environments, if the power change exceeds the threshold, model management is triggered.

[0357] In this embodiment, if the predicted path of the terminal device indicates that it will enter a line-of-sight scene, network access may be affected, and the AI ​​model management needs to be adjusted to adapt to the new network conditions. For example, the predicted path triggers model management if the terminal device anticipates entering an open area or an area with obstacles.

[0358] In this embodiment, if a terminal device attempts to access the network more than a preset threshold within a certain period, it may indicate that the access process needs optimization or the network environment is unstable. For example, if one or more terminal devices attempt to access the network multiple times within a short period but fail, exceeding the threshold, model management is triggered.

[0359] In this embodiment, if the time taken for a terminal device to access the network exceeds a preset time threshold, it may indicate low access efficiency or a problem, requiring adjustment. For example, if a terminal device fails to access the network for an extended period, exceeding the time threshold, model management is triggered.

[0360] In this embodiment, if the downlink signal strength received by the terminal device is lower than a preset downlink measurement threshold, it may indicate poor network signal quality, requiring AI model management to improve network access and connection quality. For example, if the downlink signal strength received by the terminal device remains below the threshold, model management is triggered.

[0361] In this embodiment, the second condition is a specific condition used to determine whether model management operation is required, based on the current state of the first terminal device and the network environment. When any of the following conditions are met, the device will trigger model management to ensure the adaptability and effectiveness of the AI ​​model.

[0362] Understandably, adjusting the model at appropriate times to adapt to complex scenarios such as high-speed movement and significant location changes significantly improves the device's access success rate. Under high load or poor signal conditions, adjusting the model optimizes network resource allocation, ensuring efficient device access. This reduces access failures and latency, improving user experience and satisfaction, especially in mobile scenarios. Dynamically adjusting the model based on real-time data and environmental changes enhances its adaptability and flexibility, ensuring optimal performance across different scenarios. Continuous monitoring and adjustments maintain efficient network operation and stable performance, improving overall network service quality.

[0363] In some embodiments of this application, the response result includes at least one of the following:

[0364] AI models;

[0365] Model usage identifier; the usage identifier is used to indicate whether an AI model is used in the random access process.

[0366] In this embodiment, the model usage identifier is used to identify whether a model (e.g., the first model) is used for random access, that is, to indicate whether the first terminal device should use the model for the current or future random access procedure. The usage identifier is an indicator used to indicate whether a model should be used for random access. This identifier can be a simple Boolean value (yes / no) or a more complex policy instruction.

[0367] Understandably, by using the first model, terminal devices can perform random access more efficiently, significantly improving the access success rate. Using identifiers allows for dynamic decision-making based on real-time network conditions and terminal device status, enhancing network adaptability and flexibility. Flexible use of the first model enables priority adjustments during periods of network resource strain, ensuring the smooth execution of critical tasks. In this way, AI services not only provide optimized access strategies but also flexibly manage model usage, ensuring efficient utilization of network resources and continuous improvement in user experience.

[0368] In some embodiments of this application, the access method further includes: performing model management when the AI ​​service manages the AI ​​model, and the second information and / or the third information meet the second condition.

[0369] In this embodiment, model management is performed when the AI ​​service manages the AI ​​model and the second and / or third information meets the second condition. This includes determining whether to adjust, update, or replace the currently used AI model based on real-time information about network conditions and terminal device status to optimize service performance and user experience.

[0370] In this embodiment, the second condition is a specific condition used to trigger model management operations. When the second and / or third information meets these conditions, the device performs corresponding model management operations. When the network environment or device state changes and meets the specific conditions, the device automatically adjusts the model to adapt to the new situation. By triggering model management operations by meeting the second condition, high-quality service is ensured.

[0371] For example, in a smart city network, an AI service manages the access process of multiple terminal devices. Devices continuously collect second information (such as network load and signal strength) and third information (such as device location and speed). When a device detects an increase in network load reaching a preset second condition, it automatically triggers a model management operation, selecting a new optimized model or adjusting the parameters of the current model to handle high load conditions. During the access process, terminal devices randomly connect according to the new model strategy, ensuring access success rate and efficiency. The device continuously monitors network conditions and device status, adjusting the model in real time to maintain optimal service quality. In this way, the AI ​​service can not only dynamically manage and optimize models but also ensure efficient utilization of network resources and continuous improvement of user experience.

[0372] Understandably, dynamically adjusting the model allows it to better adapt to real-time network conditions and device demands, improving service adaptability and flexibility. Adjusting the model based on real-time data optimizes access strategies for terminal devices, increasing access success rates and efficiency. Optimizing model selection and adjustment allows for the rational allocation of network resources, reducing network load and improving overall network performance. In summary, this approach enables AI services not only to dynamically manage and optimize models but also to ensure efficient utilization of network resources and continuous improvement in user experience.

[0373] This application provides an access method, which includes: acquiring first information; the first information indicating uplink congestion; determining the response result of an AI service associated with an AI model based on the first information; and the response result being a relevant result for a random access process. By using the first information containing uplink congestion information, congestion can be monitored and processed in real time. When performing random access based on the first information, the uplink congestion is taken into account, allowing for random access while considering network load balancing, thereby selecting the optimal cell, improving access success rate, and avoiding network load imbalance. Determining the response result of the AI ​​service associated with the AI ​​model based on the first information, and the response result being a relevant result for the random access process, allows the AI ​​model to analyze the first information and optimize the random access process, enabling terminal devices to access the network faster and more stably. Based on the optimized access result, the accuracy of cell reselection decisions is improved. Accurate cell reselection decisions ensure that devices select the optimal cell to camp on, guaranteeing signal quality and communication stability. It also optimizes network resource allocation and improves overall network communication performance.

[0374] The communication method provided in this application will be explained in some specific embodiments below.

[0375] Example 1: Model Training (Training an AI Model)

[0376] In the first scenario, model training is implemented on the UE side:

[0377] In this embodiment of the application, the input information for model training is as follows:

[0378] 1) Information related to the time of initial successful access and / or the number of repetitions on each cell and / or SSB (to reflect the current congestion status of the UL) (equivalent to the access information of initial successful access mentioned above). This information includes:

[0379] Information reported by other UEs that have successfully achieved initial access (equivalent to one or more first access information mentioned above);

[0380] The maximum, minimum, or average value among all reported information (equivalent to the maximum, minimum, or average value among one or more first access information mentioned above); or

[0381] The first time length is the maximum, minimum, or average value of all reported information within the first time length (equivalent to the maximum, minimum, or average value of one or more second access information mentioned above). The first time length is a period value, i.e., the situation of reported information within each T time length. T can be related to the system information update cycle (equivalent to the preset time cycle mentioned above), or the paging (paging information) cycle (equivalent to the first cycle related to the system information update cycle mentioned above), or the SSB transmission cycle (equivalent to the second cycle related to the transmission cycle of the synchronization signal block mentioned above), or the network configured cycle length (equivalent to the third cycle related to the network configured time length mentioned above).

[0382] This information can be a numerical configuration (factor) that reflects the time-related information and / or the number of repetitions of successful access (equivalent to the access factor information mentioned above), such as a number from 0 to 10, with a larger value indicating that access is more difficult.

[0383] 2) UL measurement information (used to reflect the current UL congestion situation) (equivalent to the uplink measurement information mentioned above), which is the RSRP / RSRQ / SINR information for each cell and / or each beam.

[0384] 3) Current cell selection and reselection configuration information (equivalent to the cell reselection configuration information mentioned above), namely S / R criterion configuration, slice configuration, etc.

[0385] In this embodiment of the application, the network transmits the above information via broadcast, such as in the System Information Block (SIB) and / or the Master Information Block (MIB).

[0386] The UE can obtain the following information (equivalent to the third piece of information mentioned above):

[0387] 1) DL measurement information (equivalent to the downlink measurement information of the first terminal device mentioned above);

[0388] 2) Speed ​​(equivalent to the moving speed of the first terminal device mentioned earlier), location-related information (equivalent to the location-related information of the first terminal device mentioned earlier), and / or predicted path-related information (equivalent to the predicted path-related information of the first terminal device mentioned earlier), this information:

[0389] The change in RSRP (equivalent to the change in the reference signal received power mentioned earlier) / RSRP value (equivalent to the value of the reference signal received power mentioned earlier) can be used to reflect the situation.

[0390] It can be detected through the UE device's sensing information (equivalent to its own sensor information mentioned earlier) and positioning information (equivalent to its own positioning information mentioned earlier).

[0391] 3) Slice configuration (equivalent to the switching information mentioned above), whether it is a high-speed scenario (equivalent to the scenario information mentioned above), business information, and other related information.

[0392] In this embodiment of the application, for model training, the output information is as follows (at least one of the following):

[0393] 1) The selection results of the camping cell (equivalent to the selection results of the cells to be camped in the previous text);

[0394] 2) The selection result of the camping synchronization signal block (or beam) (equivalent to the selection result of the beam to be camped in the previous text);

[0395] 3) Priority adjustment of each cell / frequent / beam (equivalent to the intermediate access result parameters mentioned above);

[0396] 4) Adjustment of measurement parameters, such as the frequency to be measured, beam, threshold configuration, priority information, etc.

[0397] It should be noted that the UE needs to further determine the selection result of the camping cell and / or SSB (beam) based on the priority adjustment result. This priority adjustment must be carried out under network control, such as:

[0398] The scope of network configuration priority adjustment (equivalent to the scope of network device configuration priority adjustment mentioned above);

[0399] The priority of specific cells or frequencies configured in the network can be adjusted (equivalent to the first cell or first frequency configured in the network device above having the highest priority);

[0400] Specific conditions under which network configuration priorities can be adjusted (equivalent to the conditions under which network device configuration priorities can be adjusted as mentioned above).

[0401] In this embodiment of the application, for model training, the reward information (equivalent to the activation parameters mentioned above) is as follows:

[0402] 1) Access result, such as success or failure (equivalent to the access result of the sample in the previous text);

[0403] 2) Time and number of repetitions required to access the network (equivalent to the access time or number of access repetitions of the sample in the previous text during the network access process);

[0404] 3) Measure the power consumption (energy consumption) of the connection (equivalent to the energy consumed by the sample in the previous text when it was connected).

[0405] In the second scenario, model training is implemented on the network side:

[0406] In this embodiment of the application, the network needs to collect the following information from the UE side for training model input:

[0407] 1) SSB information and / or cell information of the access network;

[0408] 2) Access time information and / or access repetition count information:

[0409] 3) DL measurement information (which may include results from this cell and / or neighboring cells);

[0410] 4) Speed, location-related information, and / or predicted path-related information, this information:

[0411] This can be reflected by changes in RSRP / RSRP value;

[0412] It can be detected through the sensing information and location information of the UE device.

[0413] 5) Measurement information (used to reflect the current congestion status of UL), which includes RSRP / RSRQ / SINR information for each cell and / or each beam;

[0414] 6) Current cell selection and reselection configuration information, namely S / R criterion configuration, slice configuration, etc.

[0415] In this embodiment of the application, for model training, the reward information is reported by the UE as follows:

[0416] 1) Connection result, such as success or failure;

[0417] 2) The time and number of repetitions required to access the network;

[0418] 3) Measure the power consumption (energy consumption) of the connection;

[0419] In this embodiment of the application, the network side needs to send the trained model to the UE via broadcast (SIB) and / or unicast (RRC signaling such as reconfiguration message or RRC release message).

[0420] Example 2: Model Reasoning (Reasoning of AI Models)

[0421] In the first scenario, model inference is implemented on the UE side:

[0422] The UE needs to obtain the following information from the network side for inference input:

[0423] 1) Information related to the time of initial successful access and / or the number of repetitions on each cell and / or SSB (to reflect the current congestion status of the UL), this information:

[0424] Information reported by other UEs that have successfully achieved initial access;

[0425] The maximum, minimum, or average value among all reported information; or

[0426] The maximum, minimum, or average value among all reported information within the first time length is a period value, i.e., the situation of reported information within each T time length. T can be related to the system information update cycle, the paging cycle, the SSB sending cycle, or the period length configured by the network.

[0427] This information can be a numerical configuration (factor) that reflects the time-related information and / or the number of repetitions of a successful access, such as a number from 0 to 10, with a larger value indicating a more difficult access.

[0428] 2) UL measurement information (used to reflect the current UL congestion situation), which includes RSRP / RSRQ / SINR information for each cell and / or each beam;

[0429] 3) Current cell selection and reselection configuration information, namely S / R criterion configuration, slice configuration, etc.;

[0430] In this embodiment of the application, the network transmits the above information via broadcast, such as in the SIB and / or MIB.

[0431] In this embodiment of the application, the UE can obtain the following information:

[0432] 1) DL measurement information;

[0433] 2) Speed, location-related information, and / or predicted path-related information, this information:

[0434] This can be reflected by changes in RSRP / RSRP value;

[0435] It can be detected through the sensing information and location information of the UE device.

[0436] 3) Slice configuration, whether it is a high-speed scenario, business information, and other related information.

[0437] In this embodiment of the application, for model training, the output information is as follows (at least one of the following):

[0438] 1) Selection results of camping cells;

[0439] 2) The selection results of the camping SSB;

[0440] 3) Priority adjustment for each cell / frequent / beam.

[0441] It should be noted that the UE needs to further determine the selection result of the camping cell and / or SSB based on the priority adjustment result. This priority adjustment must be carried out under network control, such as:

[0442] The scope of network configuration priority adjustment;

[0443] Network configurations can prioritize specific cells or frequencies;

[0444] Specific conditions under which network configuration can be prioritized.

[0445] In the second scenario, model inference is implemented on the network side:

[0446] In this embodiment of the application, the network needs to obtain the following information from the UE side for inference input:

[0447] 1) Information related to the time of initial successful access and / or the number of repetitions on each cell and / or SSB (to reflect the current congestion status of the UL).

[0448] 2) DL measurement information, which is the RSRP / RSRQ / SINR information for each cell and / or each beam (including this cell and / or neighboring cells);

[0449] 3) Speed, location-related information, and / or predicted path-related information, this information:

[0450] This can be reflected by changes in RSRP / RSRP value;

[0451] It can be detected through the sensing information and location information of the UE device.

[0452] In this embodiment of the application, for model inference, the output information is as follows (at least one of the following):

[0453] 1) Configure S / R criteria parameters, such as measurement thresholds (per-cell / beam / frequency), etc.

[0454] 2) Priority adjustment for each cell / frequent / beam.

[0455] Example 3: Model Training (AI Model Management)

[0456] In the first scenario, model inference is implemented on the UE side:

[0457] The network can configure the conditions for UE to perform model management (such as switching models, falling back to non-AI mode, and using AI mode):

[0458] Speed / position change threshold, which can be expressed as the absolute value of speed or position, or as the change in RSRP / RSRP value;

[0459] DL (RSRP / RSRQ / SINR) measurement threshold

[0460] In LOS / NLOS scenarios, AI methods can be used, such as in LOS scenarios.

[0461] In addition, the network can also provide the UE with the following information for model management:

[0462] 1) Information related to the time of initial successful access and / or the number of repetitions on each cell and / or SSB (to reflect the current congestion status of the UL), this information:

[0463] Information reported by other UEs that have successfully achieved initial access;

[0464] The maximum, minimum, or average value among all reported information; or

[0465] The maximum, minimum, or average value among all reported information within the first time length is a period value, i.e., the situation of reported information within each T time length. T can be related to the system information update cycle, the paging cycle, the SSB sending cycle, or the period length configured by the network.

[0466] This information can be a numerical configuration (factor) that reflects the time-related information and / or the number of repetitions of a successful access, such as a number from 0 to 10, with a larger value indicating a more difficult access.

[0467] 2) UL measurement information (used to reflect the current UL congestion situation), which includes RSRP / RSRQ / SINR information for each cell and / or each beam;

[0468] In some embodiments, the UE can report the management result to the network after successfully accessing the network, i.e., the model used during access / whether the model was used.

[0469] In the second scenario, model inference is implemented on the network side:

[0470] In this embodiment of the application, the network can be configured to allow the UE to report model management-related information, and the conditions for reporting model management (such as switching models, falling back to non-AI mode, using AI mode, etc.):

[0471] Speed / position change threshold, which can be expressed as the absolute value of speed (equivalent to the moving speed of the first terminal device being greater than or equal to the speed threshold) and position (equivalent to the position-related information of the first terminal device being greater than or equal to the position change threshold), or as RSRP change / RSRP value.

[0472] DL (RSRP / RSRQ / SINR) measurement threshold (equivalent to the downlink measurement information not exceeding the downlink measurement threshold mentioned above).

[0473] LOS / NLOS scenario (equivalent to the line-of-sight path scenario represented by the predicted path information of the first terminal device mentioned above);

[0474] Thresholds for the time required for successful access and the number of repetitions (equivalent to the access information exceeding the access threshold in the previous text regarding initial successful access).

[0475] When the above conditions are met (equivalent to the second condition mentioned above), the UE reports to the network after establishing a connection:

[0476] Access time and number of repetitions

[0477] DL measurement thresholds (which may include results from this cell and / or neighboring cells),

[0478] Speed, location information

[0479] The current scenario is either LOS or NLOS.

[0480] In this embodiment of the application, the network sends the model management results to the UE in the cell via broadcast (SIB) or unicast (dedicated RRC message, such as reconfiguration message or RRC release message).

[0481] In the embodiments of this application, this application uses AI assistance to consider uplink congestion, network load balancing, and traditional cell selection and reselection conditions, so that the UE can make a better decision for the individual UE and / or the entire system when performing cell selection and reselection by taking into account multiple factors.

[0482] In this application embodiment, an AI approach is used to solve the cell selection and reselection problem in existing 3GPP systems. By considering multiple factors, a better cell selection and reselection decision and configuration is made, improving the probability of successful user access and simultaneously achieving load balancing in the network. This invention considers the entire lifecycle processing from model training and inference to model management, and considers mechanisms for handling the above functions on the UE side and the network side respectively.

[0483] The preferred embodiments of this application have been described in detail above with reference to the accompanying drawings. However, this application is not limited to the specific details of the above embodiments. Within the scope of the technical concept of this application, various simple modifications can be made to the technical solutions of this application, and these simple modifications all fall within the protection scope of this application. For example, the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, this application will not describe the various possible combinations separately. Furthermore, various different embodiments of this application can also be arbitrarily combined, as long as they do not violate the spirit of this application, they should also be considered as the content disclosed in this application. Moreover, without conflict, the various embodiments and / or the technical features in the various embodiments described in this application can be arbitrarily combined with related technologies, and the resulting technical solutions should also fall within the protection scope of this application.

[0484] It should also be understood that in the various method embodiments of this application, the sequence number of each process does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application. Furthermore, in the embodiments of this application, the terms "downlink," "uplink," and "sidelink" are used to indicate the transmission direction of signals or data. "Downlink" indicates that the transmission direction of signals or data is a first direction from the site to the user equipment in the cell; "uplink" indicates that the transmission direction of signals or data is a second direction from the user equipment in the cell to the site; and "sidelink" indicates that the transmission direction of signals or data is a third direction from user equipment 1 to user equipment 2. For example, "downlink signal" indicates that the transmission direction of the signal is the first direction. Additionally, in the embodiments of this application, the term "and / or" is merely a description of the association relationship between related objects, indicating that three relationships can exist. In some embodiments, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Furthermore, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0485] Figure 4 is a schematic diagram of the structural composition of an optional communication device provided in an embodiment of this application, applied to a terminal device. As shown in Figure 4, the communication device 10 includes a first communication unit 11; wherein,

[0486] The first communication unit 11 is configured to acquire first information; the first information is input information of an AI service; and based on the first information, determine the response result of the AI ​​service associated with the AI ​​model; the response result is a result related to the random access process.

[0487] In some embodiments, the AI ​​service includes at least one of the following: training of an AI model, inference of an AI model, and management of an AI model; wherein, when the AI ​​service is training an AI model, the AI ​​model is an initial AI model, and the response result is a first model determined by training the initial AI model;

[0488] In the case where the AI ​​service is inference using an AI model, the AI ​​model is a first model, and the response result is the output data of the first model;

[0489] When the AI ​​service is for the management of an AI model, the response result is the management result of the AI ​​model.

[0490] In some embodiments, the first information includes: second information from a network device and / or third information from a first terminal device; the second information is related to the uplink congestion situation.

[0491] In some embodiments, the second information includes at least one of the following: initial access information; wherein the initial access information is determined based on access count information and / or access time information; uplink measurement information; the uplink measurement information includes uplink measurement objects and / or uplink measurement results, wherein the uplink measurement results are determined based on the measurement quantities.

[0492] In some embodiments, the access information includes at least one of the following: one or more first access information; the one or more first access information includes access count-related information and / or access time-related information reported by multiple second terminal devices that have successfully achieved initial access on each cell and / or each synchronization signal block; at least one of the maximum, minimum, and average values ​​among the multiple first access information; one or more second access information, the one or more second access information including access count-related information and / or access time-related information reported by third terminal devices that have successfully achieved initial access on each cell and / or each synchronization signal block within a first time period; at least one of the maximum, minimum, and average values ​​among the multiple second access information; access factor information; the access factor information is determined based on the one or more first access information and at least one of the one or more second access information.

[0493] In some embodiments, the first time period includes at least one of the following: a preset time period; a first period related to the update period of system information; a second period related to the transmission period of synchronization signal blocks; and a third period related to the time length of network configuration.

[0494] In some embodiments, the uplink measurement information includes at least one of the following: uplink measurement information for each cell; uplink measurement information for each beam.

[0495] In some embodiments, the uplink measurement object includes at least one of the following: reference signal received power; signal-to-interference-plus-noise ratio; reference signal received quality; and received signal strength indication.

[0496] In some embodiments, the uplink measurement result includes at least one of the following: a grade value characterizing the uplink measurement result; a result indication determined based on a comparison between the uplink measurement object and a preset threshold; the preset threshold is configured by the network device, or pre-configured, or preset by the protocol.

[0497] In some embodiments, the third information includes perception information, which includes at least one of the following: the moving speed of the first terminal device; location-related information of the first terminal device; and predicted path-related information of the first terminal device.

[0498] In some embodiments, the sensing information is determined by the communication parameters of the first terminal device; the communication parameters include at least one of the following: the value of the reference signal received power; the change value of the reference signal received power; its own sensor information; and its own positioning information.

[0499] In some embodiments, the third information further includes at least one of the following: downlink measurement information of the first terminal device; slice information; scene information; service information.

[0500] In some embodiments, the first information further includes: fourth information from the network device; the fourth information represents cell reselection configuration information.

[0501] In some embodiments, where the AI ​​service includes training an AI model, the first information comes from a training dataset; the training dataset includes at least one of the following: third information of a first terminal device as a sample, second and fourth information obtained by a network device corresponding to the first terminal device, excitation parameters, and sample access information of the first terminal device.

[0502] In some embodiments, where the AI ​​service includes inference of an AI model, the first information comes from an inference dataset; the inference dataset includes: third information of a first terminal device to be accessed, and at least one of second and fourth information obtained by a network device corresponding to the first terminal.

[0503] In some embodiments, when the AI ​​service is used for training an AI model, the first information serves as input data for training an initial AI model to obtain a first model, and the output data of the AI ​​model is used to adjust the initial AI model by combining the sample access information of the first terminal device in the training dataset to obtain the first model; the first model is the response result.

[0504] In some embodiments, when the AI ​​service is inference for an AI model, the first information serves as input data for the first model, and the response result serves as output data for the first model.

[0505] In some embodiments, where the first entity is deployed on a first terminal device, the second information and the fourth information come from a second entity, which is deployed on a network device.

[0506] In some embodiments, the output data includes at least one of the following: the selection result of the cell to be camped; the selection result of the beam to be camped; intermediate access result parameters; the intermediate access result parameters are used by the first terminal device to determine the cell to be camped or the beam to be camped during the random access process.

[0507] In some embodiments, the intermediate access result parameters include at least one of the following: priority information of any one of the cell, frequency, or beam to be accessed; the priority information, when a first condition is met, is used to determine the cell or beam to be camped on by the first terminal device during the random access process; and adjustment information of measurement parameters.

[0508] In some embodiments, the first condition includes at least one of the following: the adjustment range of the priority configured by the network device; the first cell or first frequency configured by the network device having the highest priority; and the conditions configured by the network device that allow priority adjustment.

[0509] In some embodiments, the first communication unit 11 is further configured to perform random access to the first terminal device based on the output data.

[0510] In some embodiments, the first model is determined by training an initial AI model based on excitation parameters and sample access information from the first terminal device in the training dataset; wherein...

[0511] The incentive parameters are obtained from the training dataset, or, in the case where the AI ​​service is for training an AI model, are determined during the random access process of the first terminal device.

[0512] In some embodiments, the second information and / or the fourth information are sent by the second entity via a broadcast message or an RRC message.

[0513] In some embodiments, where the first entity is deployed on a network device, the third information comes from a second entity, which is deployed on the first terminal device.

[0514] In some embodiments, the output data includes at least one of the following: priority information of any one of the cell, frequency, or beam to be accessed; and adjustment information of measurement parameters.

[0515] In some embodiments, the first communication unit 11 is further configured to send the output data to the first terminal device, the output data being used by the first terminal device when performing random access.

[0516] In some embodiments, when the AI ​​service is for training an AI model, the activation parameters are reported by the first terminal device or obtained from the training dataset during the random access process of the first terminal device.

[0517] The activation parameters are used to train the initial AI model by combining the sample access information of the first terminal device in the training dataset, and to determine the first model.

[0518] The first communication unit 11 is also configured to send the first model as a response result to the first terminal device.

[0519] In some embodiments, the excitation parameters include at least one of the following: the sample's access result; the sample's access time or number of access repetitions during the network access process; and the energy consumed by the sample during access or measurement.

[0520] In some embodiments, the management of the AI ​​model includes: management of a training service and / or management of an inference service, wherein the training service is training based on the AI ​​model and the inference service is inference based on the AI ​​model.

[0521] In some embodiments, the first information further includes: a second condition; the second condition serves as a judgment condition for model management when managing the AI ​​model.

[0522] In some embodiments, the second condition includes at least one of the following: speed is greater than or equal to a speed threshold; location-related information is greater than or equal to a location change threshold; the value of the reference signal received power is greater than or equal to a first power threshold; the change value of the reference signal received power is greater than or equal to a second power threshold; predicted path-related information characterizes the line-of-sight scenario; access count-related information exceeds an access count threshold; access time-related information exceeds an access time threshold; and downlink measurement information does not exceed a downlink measurement threshold.

[0523] In some embodiments, when the first entity is deployed on a network device, the third information comes from a second entity, which is deployed on the first terminal device; when the AI ​​service is the management of an AI model, the first communication unit 11 is further configured to send a second condition to the first terminal device; receive at least one of the third access information and the third information reported by the first terminal device; the third access information and the third information are determined by the first terminal device after establishing an access connection based on the second condition.

[0524] In some embodiments, the third access information and the third information are reported to the network device by the first terminal device when the second condition is met based on at least one of the third information, the second information from the network device, and the fourth information, and the access is successful.

[0525] In some embodiments, the first communication unit 11 is further configured to perform model management based on at least one of the third access information and the third information, and determine the response result; the response result characterizes the model management result.

[0526] In some embodiments, if at least one of the training and inference of the AI ​​model is performed by the second entity, the first communication unit 11 is further configured to send the response result to the second entity.

[0527] In some embodiments, the first communication unit 11 is further configured to carry the response result in a broadcast message or an RRC message.

[0528] In some embodiments, AI model training and AI model inference are performed in a first entity, and the model's usage identifier representation allows for random access using the AI ​​model.

[0529] In some embodiments, when the first entity is deployed on the first terminal device, the second information, the fourth information, and the second condition come from the second entity, which is deployed on the network device; when the AI ​​service is the management of an AI model, the response result is determined based on the second information, the third information, the fourth information, and the second condition, and is performed for model management.

[0530] In some embodiments, the first communication unit 11 is further configured to perform random access based on the response result and determine the access result;

[0531] If the access result indicates successful access, the response result is sent to the second entity.

[0532] In some embodiments, the response result includes at least one of the following: an AI model; a model usage identifier; the usage identifier being used to indicate whether the AI ​​model is used for random access procedures.

[0533] In some embodiments, where the first entity is deployed on a network device and the AI ​​service of the first entity includes AI model inference, both the training and management of the AI ​​model are performed on the first entity.

[0534] In some embodiments, when the first entity is deployed on the first terminal device, at least one of the AI ​​model inference, AI model training, and AI model management in the AI ​​service is performed on the first entity, and all AI services not performed on the first entity are performed on the second entity, which is deployed on a network device.

[0535] Understandably, by using first information including uplink congestion, congestion can be monitored and addressed in real time. When performing random access based on this first information, uplink congestion is taken into account, allowing for random access while considering network load balancing. This leads to the selection of the optimal cell, improving access success rate and avoiding network load imbalance. Based on the first information, the response results of the AI ​​service associated with the AI ​​model are determined. These response results relate to the random access process. The AI ​​model analyzes the first information to optimize the random access process, enabling terminal devices to access the network faster and more stably. Based on the optimized access results, the accuracy of cell reselection decisions is improved. Accurate cell reselection decisions ensure that devices select the optimal cell to camp on, guaranteeing signal quality and communication stability. Furthermore, it optimizes network resource allocation and improves overall network communication performance.

[0536] Those skilled in the art should understand that the description of the communication device in the embodiments of this application can be understood with reference to the description of the communication method in the embodiments of this application.

[0537] Figure 5 is a schematic diagram of the structural composition of an optional communication device provided in an embodiment of this application. The communication device 30 can be a terminal device or a first network device. The communication device 30 shown in Figure 5 includes a processor 31, which can call and run computer programs from memory to implement the methods in the embodiments of this application.

[0538] Optionally, as shown in FIG5, the communication device 30 may further include a memory 32. The processor 31 can retrieve and run computer programs from the memory 32 to implement the methods in the embodiments of this application.

[0539] The memory 32 can be a separate device independent of the processor 31, or it can be integrated into the processor 31.

[0540] Optionally, as shown in FIG5, the communication device 30 may further include a transceiver 33, which the processor 31 may control to communicate with other devices. In some embodiments, it may send information or data to other devices or receive information or data sent by other devices.

[0541] Transceiver 33, also known as communication interface, is used for receiving and sending signals during the process of sending and receiving information with other external network elements.

[0542] The transceiver 33 may include a transmitter and a receiver. The transceiver 33 may further include an antenna, and the number of antennas may be one or more.

[0543] Figure 6 is a schematic diagram of the structure of an optional chip provided in an embodiment of this application. The chip 40 shown in Figure 6 includes a processor 41, which can call and run computer programs from memory to implement the methods in the embodiments of this application.

[0544] Optionally, as shown in FIG6, chip 40 may further include memory 42. Processor 41 can retrieve and run computer programs from memory 42 to implement the methods in the embodiments of this application.

[0545] The memory 42 can be a separate device independent of the processor 41, or it can be integrated into the processor 41.

[0546] Optionally, the chip 40 may also include a transceiver (also known as a communication interface) for receiving and sending signals during the exchange of information with a device or chip.

[0547] Optionally, as shown in FIG6, the transceiver may include an input interface 43. The processor 41 can control this input interface to communicate with other devices or chips, and in some embodiments, it can receive information or data sent by other devices or chips.

[0548] Optionally, as shown in FIG6, the transceiver may include an output interface 44. The processor 41 can control this output interface to communicate with other devices or chips; in some embodiments, it can send information or data to other devices or chips.

[0549] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0550] This application also provides a computer storage medium storing one or more programs, which can be executed by one or more processors to implement the methods in this application.

[0551] Figure 7 is a schematic diagram of the structural composition of an optional communication system provided in an embodiment of this application. As shown in Figure 7, the communication system 50 includes a first terminal device 51 and a network device 52.

[0552] The first terminal device 51 or network device 52 can be used to implement the corresponding functions implemented by the first entity or the second entity in the above method. For the sake of brevity, it will not be described in detail here.

[0553] It should be understood that the processor in the embodiments of this application may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method embodiments can be completed by integrated logic circuits in the processor's hardware or by instructions in software form. The processor described above can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor, etc. The steps of the methods disclosed in the embodiments of this application can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules can be located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory, and the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method.

[0554] It is understood that the memory in the embodiments of this application can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DR RAM). It should be noted that the memory used in the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0555] It should be understood that the above-described memory is exemplary and not a limiting description. For example, the memory in the embodiments of this application may also be static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct memory bus RAM (DR RAM), etc. That is to say, the memory in the embodiments of this application is intended to include, but is not limited to, these and any other suitable types of memory.

[0556] This application also provides a computer-readable storage medium for storing computer programs.

[0557] Optionally, the computer-readable storage medium can be applied to the mobile terminal / terminal device or network device in the embodiments of this application, and the computer program causes the computer to execute the corresponding processes implemented by the mobile terminal / terminal device or network device in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0558] This application also provides a computer program product, including computer program instructions.

[0559] Optionally, the computer program product can be applied to the mobile terminal / terminal device in the embodiments of this application, and the computer program instructions cause the computer to execute the corresponding processes implemented by the mobile terminal / terminal device in the various methods of the embodiments of this application. For the sake of brevity, they will not be described in detail here.

[0560] Optionally, the computer program product can be applied to the network device in the embodiments of this application, and the computer program instructions cause the computer to execute the corresponding processes implemented by the network device in the various methods of the embodiments of this application. For the sake of brevity, they will not be described in detail here.

[0561] This application also provides a computer program.

[0562] Optionally, the computer program can be applied to the mobile terminal / terminal device in the embodiments of this application. When the computer program is run on a computer, it causes the computer to execute the corresponding processes implemented by the mobile terminal / terminal device in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0563] Optionally, the computer program can be applied to the network device or terminal device in the embodiments of this application. When the computer program is run on the computer, it causes the computer to execute the corresponding processes implemented by the network device or terminal device in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0564] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0565] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0566] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0567] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0568] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0569] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to related technologies, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0570] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of protection of the embodiments of this application.

Claims

1. An access method applied to a first entity, the method comprising: Obtain first information; The first piece of information is the input information for the AI ​​service; Based on the first information, determine the response result of the AI ​​service associated with the AI ​​model; the response result is the relevant result for the random access process.

2. The method according to claim 1, wherein, The AI ​​service includes at least one of the following: training of an AI model, inference of an AI model, and management of an AI model; wherein, when the AI ​​service is training an AI model, the AI ​​model is an initial AI model, and the response result is a first model determined by training the initial AI model; In the case where the AI ​​service is inference using an AI model, the AI ​​model is a first model, and the response result is the output data of the first model; When the AI ​​service is for the management of an AI model, the response result is the management result of the AI ​​model.

3. The method according to claim 1 or 2, wherein, The first information includes: second information from the network device and / or third information from the first terminal device; the second information is related to the uplink congestion situation.

4. The method according to claim 3, wherein, The second information includes at least one of the following: Initial access information; wherein the initial access information is determined based on access frequency information and / or access time information; Uplink measurement information; the uplink measurement information includes the uplink measurement object and / or the uplink measurement result.

5. The method according to claim 4, wherein, The initial access information includes at least one of the following: One or more first access information; the one or more first access information includes access count information and / or access time information reported by multiple second terminal devices that have successfully initially accessed each cell and / or each synchronization signal block; At least one of the maximum, minimum, and average values ​​among multiple first access information; One or more second access information, the one or more second access information including access count information and / or access time information reported by the third terminal device that initially successfully accessed each cell and / or each synchronization signal block within the first time period; At least one of the maximum, minimum, and average values ​​among multiple second access information; Access factor information; The access factor information is determined based on one or more first access information and at least one of one or more second access information.

6. The method according to claim 5, wherein, The first time period includes at least one of the following: Preset time period; The first cycle related to the system information update cycle; The second cycle related to the transmission cycle of the synchronization signal block; The third cycle is related to the duration of network configuration.

7. The method according to any one of claims 4 to 6, wherein, The uplink measurement information includes at least one of the following: Uplink measurement information for each cell; Uplink measurement information for each beam.

8. The method according to any one of claims 4 to 7, wherein, The uplink measurement object includes at least one of the following: Reference signal received power; Signal-to-interference-plus-noise ratio; Reference signal reception quality; Received signal strength indication.

9. The method according to any one of claims 4 to 8, wherein, The uplink measurement results include at least one of the following: The grade value characterizing the upstream measurement result; The result indication is determined based on the comparison between the uplink measurement object and the preset threshold; the preset threshold is configured by the network device, or pre-configured, or preset by the protocol.

10. The method according to any one of claims 3 to 8, wherein, The third information includes sensory information, which includes at least one of the following: The moving speed of the first terminal device; Location-related information of the first terminal device; Predicted path information for the first terminal device.

11. The method according to claim 10, wherein, The sensed information is determined through the communication parameters of the first terminal device; the communication parameters include at least one of the following: The value of the reference signal received power; The change in the received power of the reference signal; Its own sensor information; Its own location information.

12. The method according to claim 10 or 11, wherein, The third information also includes at least one of the following: Downlink measurement information from the first terminal device; Sub-band slice information; Scene information; Business information.

13. The method according to any one of claims 1 to 12, wherein, The first information also includes: fourth information from the network device; the fourth information represents the configuration information for cell reselection.

14. The method according to any one of claims 2 to 11, wherein, In the case where the AI ​​service includes training an AI model, the first information comes from a training dataset; the training dataset includes at least one of the following: third information of a first terminal device as a sample, second and fourth information obtained by a network device corresponding to the first terminal device, excitation parameters, and sample access information of the first terminal device.

15. The method according to any one of claims 2 to 11, wherein, In the case where the AI ​​service includes inference of an AI model, the first information comes from an inference dataset; the inference dataset includes: third information of the first terminal device to be accessed, and at least one of second and fourth information obtained by the network device corresponding to the first terminal.

16. The method according to any one of claims 1 to 15, wherein, In the case where the AI ​​service is used to train an AI model, the first information serves as the input data for training an initial AI model to obtain the first model. The output data of the AI ​​model is used to adjust the initial AI model by combining the sample access information of the first terminal device in the training dataset to obtain the first model. The first model is the response result.

17. The method according to any one of claims 1 to 15, wherein, In the case where the AI ​​service is the inference of the AI ​​model, the first information serves as the input data of the first model, and the response result serves as the output data of the first model.

18. The method according to claim 16 or 17, wherein, In the case where the first entity is deployed on the first terminal device, the second and fourth information come from the second entity, which is deployed on the network device.

19. The method according to claim 18, wherein, The output data includes at least one of the following: The results of the selection of the community where the applicant will stay; The result of the beam selection for the desired dwell time; Intermediate access result parameters; the intermediate access result parameters are used by the first terminal device to determine the cell or beam to be camped during the random access process.

20. The method according to claim 19, wherein, The intermediate access result parameters include at least one of the following: Priority information of any one of the cell, frequency, or beam to be accessed; the priority information, when a first condition is met, is used to determine the cell or beam to which the first terminal device should camp during the random access process. Information on adjusting measurement parameters.

21. The method according to claim 20, wherein, The first condition includes at least one of the following: The range of priority adjustments for network device configurations; The first cell or first frequency configured for network devices has the highest priority. The conditions under which network device configurations allow priority adjustments.

22. The method according to any one of claims 19 to 21, wherein, The method further includes: Based on the output data, random access to the first terminal device is performed.

23. The method according to any one of claims 18 to 22, wherein, The first model is determined by training an initial AI model based on activation parameters and sample access information from the first terminal device in the training dataset; wherein, The incentive parameters are obtained from the training dataset, or, in the case where the AI ​​service is for training an AI model, are determined during the random access process of the first terminal device.

24. The method according to claims 18 to 23, wherein, The second information and / or the fourth information are sent by the second entity via a broadcast message or an RRC message.

25. The method according to claim 16 or 17, wherein, In the case where the first entity is deployed on a network device, the third information comes from a second entity, which is deployed on the first terminal device.

26. The method according to claim 16 or 17, wherein, The output data includes at least one of the following: Priority information for any of the cells, frequencies, or beams to be accessed; Information on adjusting measurement parameters.

27. The method according to claim 26, wherein, The output data is sent to the first terminal device, and the output data is used by the first terminal device when performing random access.

28. The method according to any one of claims 25 to 27, wherein, When the AI ​​service is used for training an AI model, the activation parameters are reported by the first terminal device or obtained from the training dataset during the random access process of the first terminal device. The activation parameters are used to train the initial AI model by combining the sample access information of the first terminal device in the training dataset, and to determine the first model. The method further includes: The first model, as a response, is sent to the first terminal device.

29. The method according to claim 23 or 28, wherein, The excitation parameters include at least one of the following: The results of sample access; The access time or number of access repetitions of the sample during the network access process; The energy consumed in connecting or measuring a sample.

30. The method according to any one of claims 2 to 24, wherein, The management of the AI ​​model includes: management of the training service and / or management of the inference service, wherein the training service is training based on the AI ​​model and the inference service is inference based on the AI ​​model.

31. The method according to claim 30, wherein, The first information also includes: a second condition; the second condition is used as a judgment condition for model management when managing the AI ​​model.

32. The method according to claim 31, wherein, The second condition includes at least one of the following: The speed is greater than or equal to the speed threshold; Location-related information is greater than or equal to the location change threshold; The value of the reference signal received power is greater than or equal to the first power threshold; The change in the received power of the reference signal is greater than or equal to the second power threshold; Predicted path-related information characterizes the line-of-sight path scenario; The number of access attempts exceeds the access attempts threshold. Access time-related information exceeds the access time threshold; Downlink measurement information does not exceed the downlink measurement threshold.

33. The method according to any one of claims 30 to 32, wherein, In the case where the first entity is deployed on a network device, the third information comes from a second entity, which is deployed on the first terminal device; When the AI ​​service is used for the management of AI models, the method further includes: Send the second condition to the first terminal device; The device receives at least one of the third access information and the third information reported by the first terminal device; the third access information and the third information are determined by the first terminal device based on the second condition.

34. The method according to claim 33, wherein, The method further includes: Based on the third access information and at least one of the third information, model management is performed to determine the response result; the response result characterizes the model management result.

35. The method according to claim 34, wherein, In cases where at least one of the training and inference processes of the AI ​​model is performed by a second entity, the method further includes: The response result is sent to the second entity.

36. The method according to any one of claims 28, 33 to 35, wherein, The method further includes: The response result is carried in a broadcast message or an RRC message.

37. The method of claim 34, wherein, The training and inference of the AI ​​model are both performed on the first entity, and the model's usage identifier representation allows for random access using the AI ​​model.

38. The method according to any one of claims 30 to 32, wherein, When the first entity is deployed on the first terminal device, the second information, the fourth information, and the second condition come from the second entity, which is deployed on the network device. When the AI ​​service is for managing AI models, the response result is determined based on the second information, the third information, the fourth information, and the second condition, and is performed for model management.

39. The method according to claim 38, wherein, The method further includes: Based on the response result, random access is performed to determine the access result; If the access result indicates successful access, the response result is sent to the second entity.

40. The method according to any one of claims 34 to 39, wherein, The response result includes at least one of the following: AI models; The model usage identifier; the usage identifier is used to indicate whether the AI ​​model is used in the random access process.

41. The method according to claim 33 or 34, wherein, The third access information and the third information are reported to the network device by the first terminal device when at least one of the third information, the second information from the network device, and the fourth information satisfies the second condition and the access is successful.

42. The method according to any one of claims 1 to 11, wherein, When the first node is deployed on a network device and the AI ​​service of the first entity includes AI model inference, both the training and management of the AI ​​model are performed on the first entity.

43. The method according to any one of claims 1 to 11, wherein, When the first entity is deployed on the first terminal device, at least one of the AI ​​model inference, AI model training, and AI model management in the AI ​​service is performed on the first entity, and all AI services not performed on the first entity are performed on the second entity, which is deployed on the network device.

44. A communication device, the means comprising: The first communication unit is configured to acquire first information; The first piece of information is the input information for the AI ​​service; Based on the first information, determine the response result of the AI ​​service associated with the AI ​​model; the response result is the relevant result for the random access process.

45. A communication device, the communication device comprising: Memory, used to store computer programs; A processor, connected to the memory, is configured to retrieve and run the computer program from the memory to implement the method of any one of claims 1 to 43; A transceiver is used to receive and send information when exchanging information with other external devices.

46. ​​A chip, the chip comprising: Memory, used to store computer programs; A processor, connected to the memory, is configured to retrieve and run a computer program from the memory, causing a device on which the chip is mounted to perform the method as described in any one of claims 1 to 43; A transceiver is used to receive and send information during the exchange of information with a device or chip.

47. A computer-readable storage medium storing a computer program that, when executed by at least one processor, implements the method as described in any one of claims 1 to 43.

48. A computer program product comprising a computer program or instructions which, when executed by a processor, implement the steps of the method as claimed in any one of claims 1 to 43.

49. A computer program comprising computer program instructions that, when executed by a processor, implement the method as described in any one of claims 1 to 43.

Images