Ai processing unit (APU) management method and apparatus, terminal, device, and medium

By coordinating the transmission time units and resource usage of AI models, the problem of unreasonable management of the AI ​​processing unit (APU) was solved, thereby improving communication efficiency.

WO2026237941A1PCT designated stage Publication Date: 2026-11-19VIVO SOFTWARE TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
PCT/CN2026/096433
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2025-05-16
Filing Date
2026-05-12
Publication Date
2026-11-19

AI Technical Summary

Technical Problem

In existing technologies, the management of AI processing units (APUs) is inefficient, resulting in low communication efficiency.

Method used

By coordinating the transmission time units and resource usage of AI models between the terminal and network-side devices, and clarifying the APU management scheme, both communicating parties have the same understanding of the occupied APU and the remaining APU, thus achieving reasonable allocation.

Benefits of technology

It improved communication efficiency and ensured the effective management of AI model-related functions and the rational use of resources.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN2026096433_19112026_PF_FP_ABST
    Figure CN2026096433_19112026_PF_FP_ABST
Patent Text Reader

Abstract

The present application pertains to the technical field of communications, and discloses an AI processing unit (APU) management method and apparatus, a terminal, a device, and a medium. The APU management method in an embodiment of the present application comprises: when enabling or activating a downlink transmission employing a first artificial intelligence (AI) model, a terminal occupies a first number of APUs in a first time unit; or, when enabling or activating an uplink transmission employing a second AI model, the terminal occupies a second number of APUs in a second time unit; or, on the basis of first information, the terminal occupies APUs in a first time window or on a first resource, the first information being used to indicate related information of the first time window or indicate related information of the first resource.
Need to check novelty before this filing date? Find Prior Art

Description

AI Processing Unit (APU) Management Methods, Devices, Terminals, Equipment, and Media

[0001] Cross-references to related applications

[0002] This application claims priority to Chinese Patent Application No. 202510636361.3, filed on May 16, 2025, entitled "AI Processing Unit APU Management Method, Apparatus, Terminal, Device and Medium", the entire contents of which are incorporated herein by reference. Technical Field

[0003] This application belongs to the field of communication technology, specifically relating to an AI processing unit (APU) management method, device, terminal, equipment, and medium. Background Technology

[0004] With the rapid development of artificial intelligence (AI) technology, the application of AI technology in wireless communication networks has received increasing attention. Integrating AI technology into wireless communication networks can help improve technical indicators such as throughput, latency, mobility performance, and user capacity.

[0005] The relevant conferences proposed the AI ​​Processing Unit (APU), but there is no clear solution on how to effectively manage the APU, which can easily lead to problems such as unreasonable APU usage and low communication efficiency. Summary of the Invention

[0006] This application provides an AI processing unit (APU) management method, apparatus, terminal, device, and medium to effectively manage the APU and improve communication efficiency.

[0007] Firstly, an AI processing unit (APU) management method is provided, including:

[0008] When the terminal enables or activates downlink transmission based on the first artificial intelligence (AI) model, it occupies the first number of AI processing units (APUs) in the first time unit.

[0009] Alternatively, when the terminal enables or activates uplink transmission based on the second AI model, it occupies a second number of APUs in the second time unit.

[0010] Alternatively, the terminal may occupy the APU within a first time window or on a first resource based on first information, wherein the first information is used to indicate relevant information of the first time window or relevant information of the first resource.

[0011] Secondly, a method for managing an AI processing unit (APU) is provided, including:

[0012] The network-side device sends first information to the terminal, the first information being used to indicate relevant information of the first time window or relevant information of the first resource;

[0013] Alternatively, the network-side device receives second information from the terminal, which is used to indicate relevant information about the artificial intelligence processing unit (APU) occupied by the terminal.

[0014] Thirdly, an AI processing unit (APU) management device is provided, comprising:

[0015] The first processing module is used to occupy a first number of AI processing units (APUs) in a first time unit when downlink transmission based on a first artificial intelligence (AI) model is enabled or activated.

[0016] Alternatively, when uplink transmission based on the second AI model is enabled or activated, a second number of APUs may be occupied in the second time unit.

[0017] Alternatively, based on first information, the APU may be occupied within a first time window or on a first resource, whereby the first information is used to indicate relevant information of the first time window or relevant information of the first resource.

[0018] Fourthly, an AI processing unit (APU) management device is provided, comprising a fourth transmitting module or a first receiving module, wherein:

[0019] The fourth sending module is used to send first information to the terminal, the first information being used to indicate relevant information of the first time window or relevant information of the first resource;

[0020] The first receiving module is configured to receive second information from the terminal, the second information being used to indicate relevant information about the artificial intelligence processing unit (APU) occupied by the terminal.

[0021] Fifthly, an AI processing unit (APU) management device is provided, the device being configured to perform the steps of the method as described in the first aspect, or to implement the steps of the method as described in the second aspect.

[0022] In a sixth aspect, a terminal is provided, the terminal including a processor and a memory, the memory storing a program or instructions executable on the processor, the program or instructions, when executed by the processor, implementing the steps of the method as described in the first aspect.

[0023] In a seventh aspect, a terminal is provided, including a processor and a communication interface, wherein the processor is used to run programs or instructions to implement the steps of the method as described in the first aspect, and the communication interface is used to couple with the processor.

[0024] In an eighth aspect, a network-side device is provided, the network-side device including a processor and a memory, the memory storing a program or instructions executable on the processor, the program or instructions, when executed by the processor, implementing the steps of the method as described in the second aspect.

[0025] In a ninth aspect, a network-side device is provided, including a processor and a communication interface, wherein the processor is used to run programs or instructions to implement the steps of the method as described in the second aspect, and the communication interface is used to couple with the processor.

[0026] In a tenth aspect, a readable storage medium is provided, on which a program or instructions are stored, which, when executed by a processor, implement the steps of the method as described in the first aspect, or implement the steps of the method as described in the second aspect.

[0027] Eleventhly, a wireless communication system is provided, comprising: a terminal and a network-side device, wherein the terminal can be used to perform the steps of the method as described in the first aspect, and the network-side device can be used to perform the steps of the method as described in the second aspect.

[0028] In a twelfth aspect, a chip is provided, the chip including a processor and a communication interface coupled to the processor, the processor being configured to run a program or instructions to implement the steps of the method as described in the first aspect, or to implement the steps of the method as described in the second aspect.

[0029] In a thirteenth aspect, a computer program / program product is provided, which is stored in a storage medium and is executed by at least one processor to implement the steps of the method as described in the first aspect, or to implement the steps of the method as described in the second aspect.

[0030] In this embodiment, when the terminal enables or activates downlink transmission of the first AI model, it occupies a first number of APUs in the first time unit; or, when the terminal enables or activates uplink transmission based on the second AI model, it occupies a second number of APUs in the second time unit; or, based on the first information, the terminal occupies APUs within the first time window or on the first resource. This clarifies the APU management scheme, enabling both communicating parties to have the same understanding of the occupied APUs and the remaining APUs, thereby enabling the reasonable invocation of AI model-related functions, achieving effective APU management, and helping to improve communication efficiency. Attached Figure Description

[0031] Figure 1 is a block diagram of a wireless communication system applicable to an embodiment of this application;

[0032] Figure 2 is a schematic diagram of a neural network in related technologies;

[0033] Figure 3 is a schematic diagram of a neuron in a related technology;

[0034] Figure 4 is a flowchart of an implementation method for an AI processing unit (APU) management method in an embodiment of this application;

[0035] Figure 5 is a schematic diagram of APU occupancy in an embodiment of this application;

[0036] Figure 6 is a schematic diagram of an energy-saving window in an embodiment of this application;

[0037] Figure 7 is another schematic diagram of the energy-saving window in an embodiment of this application;

[0038] Figure 8 is a block diagram of the DPD principle in related technologies;

[0039] Figure 9 is a flowchart illustrating the implementation of another AI processing unit (APU) management method in this application embodiment;

[0040] Figure 10 is a schematic diagram of the AI ​​processing unit (APU) management device corresponding to Figure 4 in an embodiment of this application;

[0041] Figure 11 is a schematic diagram of the AI ​​processing unit (APU) management device corresponding to Figure 9 in an embodiment of this application;

[0042] Figure 12 is a schematic diagram of the structure of a communication device according to an embodiment of this application;

[0043] Figure 13 is a schematic diagram of the structure of a terminal in an embodiment of this application;

[0044] Figure 14 is a schematic diagram of the structure of a network-side device according to an embodiment of this application;

[0045] Figure 15 is a schematic diagram of the structure of another network-side device in an embodiment of this application. Specific Implementation

[0046] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0047] The terms "first," "second," etc., used in this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first" and "second" are generally of the same class, not limited in number; for example, the first object can be one or more. Furthermore, "or" in this application indicates at least one of the connected objects. For example, the scope of protection for "A or B" covers at least three scenarios: Scenario 1: including A but not B; Scenario 2: including B but not A; Scenario 3: including both A and B. In addition, the terms "A and / or B," "at least one of A and B," and "at least one of A or B" also cover at least the above three scenarios. The character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0048] The term "instruction" in this application can be either a direct instruction (or explicit instruction) or an indirect instruction (or implicit instruction). A direct instruction can be understood as the sender explicitly informing the receiver of specific information, the required operation, or the requested result in the instruction sent. An indirect instruction can be understood as the receiver determining the corresponding information based on the instruction sent by the sender, or making a judgment and determining the required operation or requested result based on the judgment result.

[0049] It is worth noting that the technologies described in this application are not limited to Long Term Evolution (LTE) / LTE-Advanced (LTE-A) systems, but can also be used in other wireless communication systems, such as Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), Orthogonal Frequency Division Multiple Access (OFDMA), Single-carrier Frequency-Division Multiple Access (SC-FDMA), or other systems. The terms "system" and "network" in this application are often used interchangeably, and the described technologies can be used with the systems and radio technologies mentioned above, as well as with other systems and radio technologies. The following description describes New Radio (NR) systems for illustrative purposes, and the term NR is used in most of the following description; however, these technologies can also be applied to systems other than NR systems, such as 6th generation (6G) radio systems. th Generation 6G communication system.

[0050] Figure 1 shows a block diagram of a wireless communication system applicable to an embodiment of this application. The wireless communication system includes a terminal 11 and a network-side device 12. The terminal 11 can also be referred to as User Equipment (UE), and can be a mobile phone, tablet computer, laptop computer, notebook computer, personal digital assistant (PDA), handheld computer, netbook, ultra-mobile personal computer (UMPC), mobile internet device (MID), augmented reality (AR), virtual reality (VR) device, robot, wearable device, flight vehicle, vehicle user equipment (VUE), shipboard equipment, pedestrian user equipment (PUE), smart home (home devices with wireless communication capabilities, such as refrigerators, televisions, washing machines, or furniture), game console, personal computer (PC), ATM, or self-service machine, etc. Wearable devices include: smartwatches, smart bracelets, smart headphones, smart glasses, smart jewelry (smart bracelets, smart chains, smart rings, smart necklaces, smart anklets, smart anklets, etc.), smart wristbands, smart clothing, etc. Among these, in-vehicle devices can also be referred to as in-vehicle terminals, in-vehicle controllers, in-vehicle modules, in-vehicle components, in-vehicle chips, or in-vehicle units, etc. It should be noted that the specific type of terminal 11 is not limited in this application embodiment. Network-side equipment 12 may include access network equipment or core network equipment, wherein access network equipment may also be referred to as Radio Access Network (RAN) equipment, radio access network function, or radio access network unit. Access network equipment may include base stations, Wireless Local Area Network (WLAN) access points (APs), or Wireless Fidelity (WiFi) nodes, etc.Among them, base stations can be referred to as Node B (NB), Evolved Node B (eNB), Next Generation Node B (gNB), New Radio Node B (NR Node B), Access Point, Relay Base Station (RBS), Serving Base Station (SBS), Base Transceiver Station (BTS), Radio Base Station, Radio Transceiver, Basic Service Set (BSS), Extended Service Set (ESS), Home Node B (HNB), Home Evolved Node B, Transmit / Receive Point (TRP), Non-Terrestrial Network (NTN) equipment (such as satellite or high altitude platform stations). The term "base station" can be any suitable term in the field, such as "station" or any other appropriate term in the relevant field, as long as the same technical effect is achieved. The term "base station" is not limited to specific technical terms. It should be noted that the embodiments of this application only use the base station in the NR system as an example for introduction, and do not limit the specific type of base station.

[0051] Core network equipment, also known as core network nodes, core network functions, or core network elements, includes, but is not limited to, at least one of the following: Mobility Management Entity (MME), Access and Mobility Management Function (AMF), Session Management Function (SMF), User Plane Function (UPF), Policy Control Function (PCF), Policy and Charging Rules Function (PCRF), Edge Application Server Discovery Function (EASDF), Unified Data Management (UDM), Unified Data Repository (UDR), Home Subscriber Server (HSS), Centralized network configuration (CNC), Network Repository Function (NRF), Network Exposure Function (NEF), Local NEF (L-NEF), and Binding Support. Functions include BSF, Application Function (AF), Location Management Function (LMF), Gateway Mobile Location Centre (GMLC), Network Data Analytics Function (NWDAF), and Non-Terrestrial Network (NTN) equipment (such as satellite or high altitude platform station).It should be noted that the embodiments of this application only use the core network equipment in the NR system as an example for introduction, and do not limit the specific type of core network equipment. If the name of the core network equipment mentioned in the embodiments of this application changes in subsequent protocol versions (e.g., 6G), it is also within the scope of protection of this application.

[0052] Optionally, the core network equipment can be implemented by one or more functional modules in a single device, or by multiple devices working together; this application does not specifically limit this. It is understood that the aforementioned functional modules can be network elements in hardware devices, software functional modules running on dedicated hardware, or virtualized functional modules instantiated on a platform (e.g., a cloud platform).

[0053] To facilitate understanding, the relevant technologies and concepts involved in the embodiments of this application will be introduced first.

[0054] I. Artificial Intelligence (AI)

[0055] Artificial intelligence (AI) has wide applications in various fields such as communications, healthcare, and education. Integrating AI into wireless communication networks to improve technical indicators such as throughput, latency, and user capacity is an important task for future wireless communication networks. AI modules can be implemented in various ways, such as neural networks, decision trees, support vector machines, and Bayesian classifiers. This application's embodiments mainly use a neural network as an example for illustration, but this does not constitute a limitation on the specific type of AI module.

[0056] Figure 2 shows a schematic diagram of a neural network, which includes an input layer (X1, X2, ..., X...). n The neural network consists of three layers: a hidden layer (Y), a hidden layer, and an output layer (Y). A schematic diagram of a neuron is shown in Figure 3, where: z = a1w1 + ... + a... k w k +…+a K w K +b;

[0057] Among them, a1, a2, ..., a k ... a K The input is w, where w is the weight (multiplicative coefficient), b is the bias (additive coefficient), and σ(.) is the activation function. Common activation functions include the sigmoid function, the hyperbolic tangent function, and the rectified linear unit (ReLU) (or linear rectified function).

[0058] The parameters of a neural network are optimized using optimization algorithms. An optimization algorithm is an algorithm that minimizes or maximizes an objective function (or loss function), which is often a mathematical combination of model parameters and data. For example, given data X and its corresponding label Y, a neural network model f(.) is constructed. After obtaining the neural network model, the predicted output f(x) can be obtained based on the input x, and the difference between the predicted value and the true value (f(x) - Y) can be calculated; this is the loss function. The goal is to find suitable values ​​W and b that minimize the value of the loss function. The smaller the loss value, the closer the prediction result of the neural network model is to the reality.

[0059] Most common optimization algorithms are based on the error back propagation (BP) algorithm. The basic idea of ​​the BP algorithm is that the learning process consists of two parts: forward propagation of the signal and backward propagation of the error. During forward propagation, the input sample is introduced from the input layer, processed layer by layer by the hidden layers, and then propagated to the output layer. If the actual output of the output layer does not match the expected output, the process transitions to the error back propagation stage. Error back propagation involves propagating the output error back to the input layer layer by layer through the hidden layers, distributing the error to all units in each layer, thus obtaining the error signal of each unit. This error signal serves as the basis for adjusting the weights of each unit. This process of adjusting the weights through forward and backward propagation is repeated continuously. This continuous adjustment of weights is the learning and training process of the network. This process continues until the error of the network output is reduced to an acceptable level, or until the predetermined number of learning iterations is reached.

[0060] Common optimization algorithms include gradient descent, stochastic gradient descent (SGD), mini-batch gradient descent, momentum method, Nesterov (named after the inventor, specifically stochastic gradient descent with momentum), adaptive gradient descent (Adagrad), adaptive learning rate adjustment (Adadelta), root mean square prop (RMSprop), and adaptive momentum estimation (Adam).

[0061] During error backpropagation, these optimization algorithms calculate the gradient based on the error / loss obtained from the loss function with respect to the current neuron, add the learning rate, previous gradients / derivatives / partial derivatives, etc., and then pass the gradient to the previous layer.

[0062] II. AI Units / AI Models

[0063] In this application embodiment, AI unit / AI model may also be referred to as AI unit, AI model, AI module, machine learning (ML) model, ML unit, ML module, AI structure, AI function, AI characteristic, neural network, neural network function, neural network functionality, etc. Alternatively, AI unit / AI model may also refer to a processing unit or processing module capable of implementing specific algorithms, formulas, processing flows, capabilities, etc. related to AI. Alternatively, AI unit / AI model may be a processing method, algorithm, function, characteristic, module, or unit for a specific dataset. Alternatively, AI unit / AI model may be a processing method, algorithm, function, characteristic, module, or unit running on AI / ML related hardware such as Graphics Processing Unit (GPU), Neural Processing Unit (NPU), Tensor Processing Unit (TPU), or Application-Specific Integrated Circuit (ASIC). This application embodiment does not specifically limit this. Optionally, the specific dataset includes the input and / or output of the AI ​​unit / AI model.

[0064] Optionally, the identifier of an AI unit / AI model can be understood as an AI unit identifier, AI model identifier, AI module identifier, AI structure identifier, AI algorithm identifier, or the identifier of a specific dataset associated with an AI unit / AI model, or the identifier of a specific scenario, environment, region, cell, channel characteristics, or device related to AI / ML, or the identifier of a function, characteristic, capability, or module related to AI / ML. This application embodiment does not specifically limit this.

[0065] III. Model Lifecycle

[0066] The lifecycle management (LCM) of AI / ML models (e.g., model training, model deployment, model inference, model monitoring, and model updates) and AI / ML functions are clearly defined.

[0067] The lifecycle of a model includes one or more of the following: model training, model deployment, model inference, model monitoring, and model updates.

[0068] IV. Calculation of Traditional Channel State Information (CSI) Processing Unit (CPU) (Legacy CPU)

[0069] The CPU is the basic unit for processing Channel State Information (CSI) in a terminal. There are usually two configuration modes. One configuration mode is that all CPUs process a CSI together, calculate the CSI and report it in the shortest possible time. The other configuration mode is that the processing latency can be larger, and the number of CPUs allocated for each CSI calculation is the number of Channel State Information-Reference Signal (CSI-RS) resources.

[0070] CPU allocation principles:

[0071] 1) Terminal reporting instructions support N CPU One CPU is used to process CSI reports;

[0072] 2) For N CSI reports, how to allocate the remaining N resources on the terminal? CPU -L unused CPUs:

[0073] a) Analyze the remaining N CPU Are L unused CPUs sufficient to process N CSI reports simultaneously?

[0074] b) If not, process the X highest-priority CSI reports based on priority, and the number of CPUs used by the X CSI reports should be less than or equal to N. CPU -L.

[0075] Where L represents the number of CPUs that are currently in use.

[0076] The relevant technologies and concepts involved in the embodiments of this application have been introduced above. The AI ​​processing unit (APU) management method provided in the embodiments of this application will be described in detail below with reference to the accompanying drawings and through some embodiments and application scenarios.

[0077] Referring to Figure 4, which is an implementation flowchart of an AI processing unit (APU) management method provided in this application embodiment, the method includes the following steps:

[0078] S410: When the terminal enables or activates downlink transmission based on the first artificial intelligence (AI) model, it occupies a first number of AI processing units (APUs) in the first time unit.

[0079] Alternatively, when the terminal enables or activates uplink transmission based on the second AI model, it occupies a second number of APUs in the second time unit.

[0080] Alternatively, the terminal may occupy the APU within a first time window or on a first resource based on the first information, wherein the first information is used to indicate relevant information of the first time window or relevant information of the first resource.

[0081] It's important to note that "AI Processing Unit (APU)" is just one possible name; it could also be "ML Processing Unit (MPU)." It can also be understood as a dedicated processing unit for measuring AI or ML-related business and functions, or simply a dedicated processing unit for computing or processing AI or ML-related business and functions. In other words, as long as the functions are the same, the naming should not be restricted.

[0082] By applying the method provided in the embodiments of this application, when the terminal enables or activates downlink transmission of the first AI model, it occupies a first number of APUs in the first time unit; or, when the terminal enables or activates uplink transmission based on the second AI model, it occupies a second number of APUs in the second time unit; or, based on the first information, the terminal occupies APUs within the first time window or on the first resource. This clarifies the management scheme for APUs, enabling both communicating parties to have the same understanding of the occupied APUs and the remaining APUs, thereby enabling the reasonable invocation of functions related to the AI ​​model, achieving effective management of APUs, and helping to improve communication efficiency.

[0083] In some embodiments of this application, the terminal may enable or activate downlink transmission based on a first AI model, such as channel estimation based on the first AI model or intelligent reception based on the first AI model. In this case, a first number of APUs may be occupied in a first time unit.

[0084] Downlink transmission can be understood as downlink reception (DL reception).

[0085] Optionally, the scheduling or configuration information of the downlink transmission may include indication information to enable or activate downlink transmission based on the first AI model.

[0086] Optionally, the first time unit may include one of the following:

[0087] Downlink time units used for downlink transmission;

[0088] Downlink time units for downlink transmission, and a third time unit, the third time unit including uplink time units or flexible time units associated with downlink transmission.

[0089] The time unit can include time slots, symbols, etc.

[0090] This can be understood as follows: if downlink transmission based on the first AI model is enabled or activated, a first number of APUs will be used for downlink transmission in the downlink time units. In other words, when downlink transmission based on the first AI model is enabled or activated, the first number of APUs will be used in all downlink time units indicated as downlink (D).

[0091] Considering that the downlink time unit accounts for a large proportion and is insufficient to deactivate and activate the AI ​​model used for downlink transmission in the uplink time unit, enabling or activating downlink transmission based on the first AI model allows the first number of APUs to be used in both the downlink time unit and the third time unit used for downlink transmission. The third time unit includes the uplink time unit or flexible time unit associated with downlink transmission. The uplink time unit or flexible time unit associated with downlink transmission can be determined according to the protocol agreement or the configuration of the network-side equipment.

[0092] In one embodiment, considering that the downlink time unit accounts for a large proportion and is insufficient to deactivate and activate the AI ​​model for downlink transmission in the uplink time unit, if downlink transmission based on the first AI model is enabled or activated, the first number of APUs can be occupied in all time units, such as all downlink time units, all uplink time units, and all flexible time units.

[0093] In some embodiments of this application, the terminal can enable or activate uplink transmission based on the second AI model, such as performing uplink power compensation, intelligent transmission, constellation mapping, or modulation based on the second AI model. In this case, a second number of APUs can be occupied in the second time unit.

[0094] Optionally, the second time unit may include one of the following:

[0095] Uplink time unit used for uplink transmission;

[0096] Uplink time unit for uplink transmission, and a fourth time unit, the fourth time unit including a downlink time unit or a flexible time unit for preparing for uplink transmission.

[0097] The uplink time unit used for uplink transmission, and the time unit between the downlink time unit that receives downlink control information and the time unit before uplink transmission;

[0098] The time unit between the downlink time unit from the receipt of downlink control information and the uplink time unit used for uplink transmission.

[0099] The time unit can include a time slot and a symbol.

[0100] Optionally, the uplink time unit used for uplink transmission can be a time unit indicated as uplink (U) in a time division duplex (TDD) configuration, or it can be a time unit related to the actual transmission of uplink signals or data.

[0101] This can be understood as follows: if uplink transmission based on the second AI model is enabled or activated, a second number of APUs will be used for uplink transmission in the uplink time units. In other words, when uplink transmission based on the second AI model is enabled or activated, a second number of APUs will be used in all uplink time units indicated as U.

[0102] Considering the relatively small proportion of uplink time units, some downlink time units are needed to prepare for uplink transmission before it begins. For example, the uplink time units before sending uplink signals or data may not be sufficient to load the second AI model into the AI ​​engine to complete the preparation work for the uplink signals or data. Therefore, some downlink time units are needed to complete the preparation work for the uplink signals or data. Similarly, downlink time units are needed to complete the preparation work for the corresponding uplink signals or data transmission. Based on this, if uplink transmission based on the second AI model is enabled or activated, a second number of APUs are required in both the uplink time units used for uplink transmission and the fourth time unit. The fourth time unit includes downlink time units or flexible time units used to prepare for uplink transmission. For example, if uplink signals or data are to be transmitted in the first uplink symbol, preparation work for the uplink signals or data transmission needs to be completed in some downlink symbols beforehand. Therefore, a second number of APUs are required in both the uplink symbols and these downlink symbols.

[0103] Considering that uplink transmission can be determined upon receiving downlink control information (DCI), if uplink transmission based on the second AI model is enabled or activated, the second number of APUs can be used in the uplink time unit used for uplink transmission and in the time unit between the downlink time unit receiving downlink control information and the time unit before uplink transmission, or the second number of APUs can be used in the time unit between the downlink time unit receiving downlink control information and the uplink time unit used for uplink transmission.

[0104] The time unit preceding uplink transmission can be either a downlink time unit or a flexible time unit. If there is no flexible time unit between the uplink and downlink time units, the time unit preceding uplink transmission is a downlink time unit; if there is a flexible time unit between the uplink and downlink time units, the time unit preceding uplink transmission is a flexible time unit. This can be understood as follows: if uplink transmission based on the second AI model is enabled or activated, the second number of APUs can be used in both the uplink time unit used for uplink transmission and the time unit between the downlink time unit receiving downlink control information and the downlink time unit preceding uplink transmission; or the second number of APUs can be used in both the uplink time unit used for uplink transmission and the time unit between the downlink time unit receiving downlink control information and the flexible time unit preceding uplink transmission.

[0105] The time unit between the downlink time unit that receives downlink control information and the uplink time unit used for uplink transmission can include downlink time units and uplink time units, or it can include downlink time units, uplink time units, and flexible time units. This can be understood as the APU being used only when there is uplink data scheduling or uplink signal transmission; therefore, the time the APU occupies is related to the reception time of the scheduling information.

[0106] In this embodiment, considering that uplink transmissions are relatively infrequent and the second AI model used for uplink transmission does not need to be constantly in an active or inference state, in one optional implementation, if the second AI model is in an active or inference state during a statically configured uplink time unit, a second number of APUs needs to be occupied during that statically configured uplink time unit. Alternatively, considering that the second AI model needs to be in an active or inference state before the statically configured uplink time unit, an additional second number of APUs needs to be occupied during the downlink time unit or flexible time unit before the statically configured uplink time unit for loading or preprocessing uplink signals or uplink data to be transmitted by the second AI model. Furthermore, considering that there is not actual uplink signal or uplink data transmission in every uplink time unit, uplink scheduling information needs to be considered, and a second number of APUs needs to be occupied during the time unit between the uplink scheduling information and the uplink transmission.

[0107] Optionally, the first quantity or the second quantity can be determined based on at least one of the following:

[0108] 1) Terminal capability information;

[0109] The terminal's capability information may include at least one of the following:

[0110] Maximum number of APUs running; maximum number of APUs running simultaneously; processing time required for one or more AI models using the maximum number of APUs; processing time required for one or more AI models using the third largest number of APUs; maximum number of APUs stored; information about the AI ​​models stored.

[0111] The stored AI model information includes at least one of the following:

[0112] The maximum number of AI models stored; the input information associated with the stored AI models; the output information associated with the stored AI models; the signal information associated with the stored AI models; the configuration information associated with the stored AI models; and the APU information associated with the stored AI models.

[0113] The terminal can determine, based on its own capability information, the first number of APUs it occupies in the first time unit, or the second number of APUs it occupies in the second time unit.

[0114] 2) The number of APUs reported by the terminal;

[0115] The number of APUs reported by the terminal can be understood as the number of APUs required to run the corresponding AI model, such as the number of APUs required to run the first AI model. The first number of APUs occupied by the terminal in the first time unit can be the same as the number of APUs reported by the terminal as required to run the first AI model. The network-side device can determine the first number of APUs occupied by the terminal in the first time unit, or the second number of APUs occupied by the terminal in the second time unit, based on the number of APUs reported by the terminal. Alternatively, the network-side device can determine the number of APUs occupied by the terminal within the first time window or on the first resource, based on the number of APUs reported by the terminal.

[0116] 3) Information related to the AI ​​model used;

[0117] Optionally, the relevant information about the AI ​​model used may include at least one of the following:

[0118] Input information associated with the AI ​​model used; output information associated with the AI ​​model used; signal information associated with the AI ​​model used; configuration information associated with the AI ​​model used; APU information associated with the AI ​​model used; processing requirements information of the AI ​​model used.

[0119] The terminal can determine the complexity and processing requirements of the AI ​​model based on relevant information about the AI ​​model used, and thus determine the number of APUs to be used. For example, based on relevant information about the first AI model used, it can determine the first number of APUs used in the first time unit. Similarly, based on relevant information about the second AI model used, it can determine the second number of APUs used in the second time unit.

[0120] This means that different AI models may use different numbers of APUs. For example, complex AI models use more APUs, while simple AI models use fewer. Furthermore, AI models with high latency requirements use more APUs, such as four, while AI models with low latency requirements use fewer, such as two.

[0121] Optionally, the number of APUs occupied by the corresponding AI model can be determined during the interaction of applicability information related to the AI ​​model. For example, before enabling or activating the AI ​​model, the network-side device communicates with the terminal to determine the number of APUs occupied by the AI ​​model. That is, as the terminal completes the update of the AI ​​model for the corresponding function, the number of APUs occupied by the corresponding AI model will be updated.

[0122] 4) As stipulated in the agreement;

[0123] The protocol can specify a first number of APUs to be used in the first time unit, or a second number of APUs to be used in the second time unit. Alternatively, the protocol can define rules to determine the first or second number of APUs to be used in the first or second time unit. For example, the rule could state that the first or second number is related to the bandwidth of the Bandwidth Part (BWP). Another example is that the rule could state that the first or second number is related to the bandwidth used by downlink data / signals or uplink data / signals.

[0124] 5) Indication information from network-side devices.

[0125] The network-side device can indicate the corresponding number of APUs to the terminal based on the capability information reported by the terminal or the number of APUs reported by the terminal. The terminal can then determine the first number of APUs to occupy in the first time unit or the second number of APUs to occupy in the second time unit based on the indication information from the network-side device.

[0126] The terminal can determine the first number of APUs occupied in the first time unit or the second number of APUs occupied in the second time unit based on one of the above-mentioned factors. Alternatively, it can combine the above-mentioned factors to determine the first number of APUs occupied in the first time unit or the second number of APUs occupied in the second time unit. This clarifies the determination scheme for the first or second number, enabling the terminal to accurately determine the first or second number.

[0127] For example, as shown in Figure 5, when downlink reception based on the first AI model is enabled or activated, a first number of APUs are occupied in the downlink time unit 510 used for downlink reception.

[0128] When uplink transmission based on the second AI model is enabled or activated, there are four scenarios in which the APU is occupied. The first scenario is that the second number of APUs is occupied in the uplink time unit 520 used for uplink transmission. The second scenario is that the second number of APUs is occupied in both the uplink time unit 520 used for uplink transmission and the downlink time unit 530 used to prepare for uplink transmission. The third scenario is that the second number of APUs is occupied in both the uplink time unit 520 used for uplink transmission and the time unit 540 between the downlink time unit that receives downlink control information and the downlink time unit before uplink transmission. The fourth scenario is that the second number of APUs is occupied in both the downlink time unit that receives downlink control information and the uplink time unit 550 used for uplink transmission.

[0129] In some embodiments of this application, when there is an overlap between the second time unit and the first time unit, the method further includes:

[0130] The terminal determines to perform at least one of downlink transmission and uplink transmission in the overlapping time unit based on at least one of the following:

[0131] The protocol specifies; the indication information of the network-side equipment; and the priority information corresponding to downlink and uplink transmissions.

[0132] In the embodiments of this application, when downlink transmission of the first AI model is enabled or activated, a first number of APUs are occupied in the first time unit; when uplink transmission of the second AI model is enabled or activated, a second number of APUs are occupied in the second time unit. The first time unit and the second time unit may overlap. For example, in Figure 5, when the first time unit includes 510 and the second time unit includes 520 and 530, or when the second time unit includes 540, the first time unit and the second time unit overlap.

[0133] Downlink transmission, uplink transmission, or both can be performed within overlapping time units.

[0134] According to the protocol, the terminal can determine whether to perform downlink transmission or uplink transmission in the overlapping time unit. For example, if the protocol stipulates that downlink transmission is performed in the overlapping time unit, then the first number of APUs will be used for downlink transmission in the overlapping time unit, and the second number of APUs will not be used for uplink transmission.

[0135] Alternatively, the terminal can determine, based on the instructions from the network-side device, whether to perform downlink or uplink transmission within the overlapping time unit. For example, if the network-side device indicates uplink transmission within the overlapping time unit, then the second number of APUs will be used for uplink transmission within the overlapping time unit, and the first number of APUs will no longer be used for downlink transmission within the overlapping time unit.

[0136] Alternatively, the terminal can determine whether to perform at least one of downlink and uplink transmissions in the overlapping time unit based on the priority information corresponding to downlink and uplink transmissions.

[0137] For example, if downlink transmission has a higher priority than uplink transmission, then the first number of APUs can be used for downlink transmission in the overlapping time unit. Conversely, if uplink transmission has a higher priority than downlink transmission, then the second number of APUs can be used for uplink transmission in the overlapping time unit.

[0138] Optionally, the priorities of the current downlink and uplink transmissions can be determined based on their respective priority information. Then, based on these priorities, the priorities of the current downlink and uplink transmissions are further determined, thus determining which APU is allocated in the overlapping time unit for at least one of the current downlink and uplink transmissions. For example, if the current downlink transmission has the highest priority among downlink transmissions and the current uplink transmission has the lowest priority among uplink transmissions, then the current downlink transmission can be considered to have a higher priority than the current uplink transmission, and the APU allocated in the overlapping time unit can be used for the current downlink transmission.

[0139] Optionally, priority can be determined based on a priority threshold.

[0140] When there is an overlap between the first time unit and the second time unit, the number of APUs occupied in the overlapping time unit and at least one of the following can be determined according to the protocol, the instruction information of the network-side device, and the priority information corresponding to the downlink and uplink transmissions: downlink transmission and uplink transmission.

[0141] In some embodiments of this application, the terminal determining that at least one of downlink transmission and uplink transmission must be performed in an overlapping time unit may include the following steps:

[0142] If, in an overlapping time unit, the number of APUs required for downlink and uplink transmissions exceeds the number of APUs remaining available in the terminal, the terminal determines to perform either downlink or uplink transmission in the overlapping time unit.

[0143] It is understandable that, in the case of overlapping time units between the second time unit and the first time unit, if the number of APUs required for downlink and uplink transmission in the overlapping time unit does not exceed the number of APUs remaining available in the terminal, then the first number of APUs can be used for downlink transmission in the overlapping time unit, while the second number of APUs can be used for uplink transmission.

[0144] If, within an overlapping time unit, the number of APUs required for downlink and uplink transmissions exceeds the number of remaining available APUs in the terminal, it is considered to exceed the terminal's capacity. The terminal can then determine, based on at least one of the following: protocol specifications, network-side device instructions, and priority information corresponding to downlink and uplink transmissions, whether to perform downlink or uplink transmissions within the overlapping time unit. This clarifies for both the network and terminal sides whether the terminal can apply the AI ​​model to execute downlink and uplink transmissions on the corresponding resources. Otherwise, if there are insufficient APUs, it is unclear whether the terminal will use APUs for downlink or uplink transmissions.

[0145] For example, if the terminal has 5 remaining available APUs, and in an overlapping time unit, downlink transmission requires 3 APUs and uplink transmission requires 2 APUs, then the total number of APUs required for downlink and uplink transmission does not exceed the terminal's remaining available APUs. Therefore, it can be determined that 3 APUs will be used for downlink transmission and 2 APUs for uplink transmission in the overlapping time unit. However, if in an overlapping time unit, downlink transmission requires 4 APUs and uplink transmission requires 2 APUs, then the total number of APUs required for downlink and uplink transmission exceeds the terminal's remaining available APUs. Based on at least one of the following: protocol agreement, network-side device indication information, and priority information corresponding to downlink and uplink transmission, it can be determined that 4 APUs will be used for downlink transmission or 2 APUs for uplink transmission in the overlapping time unit.

[0146] If, in the overlapping time unit, the number of APUs required for downlink and uplink transmissions is determined based on the number of APUs remaining available in the terminal, the APUs occupied in the overlapping time unit are used for at least one of downlink and uplink transmissions. This helps to ensure the smooth execution of downlink or uplink transmissions, enabling the network side and the terminal side to clearly determine whether the terminal can apply the AI ​​model to perform downlink and uplink transmissions on the corresponding resources, or whether it can only perform one of them.

[0147] In some embodiments of this application, the method may further include the following steps:

[0148] The terminal sends a second message to the network-side device, which indicates the relevant information of the APU occupied by the terminal.

[0149] In this embodiment, the terminal can send second information to the network-side device, which receives the second information from the terminal. The second information indicates relevant information about the APUs occupied by the terminal, enabling the network-side device to know the relevant information, such as a first time unit of APU occupation, a second time unit of APU occupation, a first number of APUs occupied in the first time unit, and a second number of APUs occupied in the second time unit. This facilitates accurate scheduling by the network-side device. Based on the second information, the network-side device can determine the APUs required to run an AI model, the time units occupied by running the AI ​​model, etc., or determine which AI model-related functions to schedule, or when to schedule AI model-related functions. This avoids continuous scheduling failures for one or more AI model-related functions due to insufficient remaining available APUs on the terminal.

[0150] In some embodiments of this application, the first time unit is a time unit within an energy-saving window, which is a window in which the terminal periodically listens to downlink signals in a power-saving state.

[0151] In the embodiments of this application, for downlink transmission in the idle state, since the introduction of power-saving state or discontinuous reception (DRX), it is possible to occupy the APU in the time unit within the power-saving window.

[0152] Alternatively, for downlink transmissions in an inactive state, due to the introduction of power-saving states or DRX, it is possible to utilize the APU within the time unit of the power-saving window. The inactive state can be understood as an inactive state, an idle state, or a power-saving state.

[0153] Alternatively, for connected downlink transmission, connected discontinuous reception (Connected mode DRX, CDRX) is also introduced, so it is possible to occupy the APU in the time unit within the energy-saving window.

[0154] The energy-saving window can be configured by the network-side device, such as a periodic time window. Optionally, the configuration information of the energy-saving window may include one or more of the following: period, start time, end time, and window length.

[0155] A power-saving window can be understood as a window during which a terminal periodically listens to downlink signals while in a power-saving state. For example, within the power-saving window, it listens to the Physical Downlink Control Channel (PDCCH) and measurement signals, and outside the power-saving window, it sleeps. For instance, within the power-saving window, the terminal can listen to the Positioning Reference Signal (PRS), Synchronization Signal Block (SSB) bursts, paging occasions, and Inter-Frequency (Inter-F) Radio Resource Management (RRM) SSB bursts.

[0156] As shown in Figure 6, during the measurement of the inactive positioning reference signal under low signal-to-interference plus noise ratio (SINR), there is an energy-saving window between deep sleep and micro sleep states. Within this window, the terminal is in light sleep or micro sleep states. This can be understood as follows: if scheduling or listening is not required for a period of time, but the time is insufficient to switch to deep sleep, the terminal can switch to light sleep or micro sleep states.

[0157] As shown in Figure 7, during the measurement of the inactive state positioning reference signal under high SINR, there is an energy-saving window between deep sleep states. Within the energy-saving window, the terminal is in a light sleep state.

[0158] Optionally, the energy-saving window includes an active energy-saving window. An active energy-saving window can be understood as a window enabled by the enable information for activating energy-saving window measurements. It is understood that even if a periodic energy-saving window is configured, it may not need to be activated every period. Activation signaling can be detected first; if the activation signaling indicates activation of the energy-saving window, then the APU will be occupied on the corresponding active energy-saving window.

[0159] When the terminal enables or activates downlink transmission based on the first AI model, it occupies a first number of APUs within the time units of the energy-saving window. Optionally, the first number of APUs can be occupied within the downlink time units used for downlink transmission within the energy-saving window, or the first number of APUs can be occupied within the downlink time units used for downlink transmission within the energy-saving window, as well as the uplink time units or flexible time units associated with downlink transmission. This helps the terminal save energy.

[0160] In combination with the above scheme, if downlink transmission based on the first AI model has occupied the first number of APUs before the actual downlink scheduling, then in the downlink time unit, the number of APUs remaining available in the terminal is the difference between the total number of APUs and the first number.

[0161] In combination with the above scheme, if the uplink transmission based on the second AI model has occupied the second number of APUs before the actual uplink and downlink scheduling, then in the uplink time unit, the number of APUs remaining available in the terminal is the difference between the total number of APUs and the second number.

[0162] Based on the above scheme, if downlink transmission based on the first AI model has occupied a first number of APUs and uplink transmission based on the second AI model has occupied a second number of APUs before the actual uplink scheduling, then in some time units before the uplink time unit, the number of APUs remaining available in the terminal is the difference between the total number of APUs and the sum of the first and second numbers.

[0163] Alternatively, it can be understood that, in combination with the above scheme, if downlink transmission based on the first AI model has already occupied the first number of APUs before actual downlink scheduling, then the number of APUs remaining available in the terminal is the difference between the total number of APUs and the first number.

[0164] In combination with the above scheme, if the uplink transmission based on the second AI model has occupied the second number of APUs before the actual uplink and downlink scheduling, the number of APUs remaining available in the terminal is the difference between the total number of APUs and the second number.

[0165] Based on the above scheme, before actual uplink and downlink scheduling, if downlink transmission based on the first AI model has occupied a first number of APUs and uplink transmission based on the second AI model has occupied a second number of APUs, then the number of APUs remaining available in the terminal is the difference between the total number of APUs and the sum of the first and second numbers.

[0166] In some embodiments of this application, the terminal may occupy the APU within a first time window or on a first resource based on first information, wherein the first information is used to indicate relevant information of the first time window or relevant information of the first resource. Optionally, the network-side device may send the first information to the terminal, and the terminal may receive the first information from the network-side device and occupy the APU within the first time window or on the first resource based on the first information.

[0167] This embodiment of the application takes into account that APU is required when acquiring the AI ​​model, but the AI ​​model does not occupy APU or occupies a small amount of APU during actual use. Therefore, the terminal occupies APU for acquiring the AI ​​model within the first time window or on the first resource. AI model acquisition includes AI model adjustment, AI model fine-tuning, and AI model updating.

[0168] Alternatively, considering that an AI model can be trained to create a table that can be used at a later time, such as a table for power amplifier (PA) compensation, the APU is used when training, validating, or updating the table, but not or only slightly when the table is actually used. Therefore, the terminal uses the APU in the first time window or on the first resource to obtain information for downlink or uplink transmission.

[0169] In some embodiments of this application, when the terminal occupies the APU within a first time window or on a first resource based on the first information, the method may further include the following:

[0170] 1) The terminal sends the number of APUs it occupies to the network-side equipment;

[0171] Based on the first information, the terminal can occupy an APU within a first time window or on a first resource, and can send the number of occupied APUs to the network-side device so that the network-side device can clearly know the number of APUs occupied by the terminal.

[0172] 2) The terminal sends the number of APUs it occupies in the first time window or on the first resource to the network-side device;

[0173] Based on the first information, the terminal occupies an APU within a first time window or on a first resource. It can send the number of APUs occupied within the first time window or on the first resource to the network-side device so that the network-side device can clearly understand the number of APUs occupied by the terminal within the first time window or on the first resource.

[0174] 3) When the terminal is performing uplink or downlink transmission, it does not occupy the APU or occupies one APU;

[0175] Based on the first information, the terminal occupies the APU within the first time window or on the first resource. The occupied APU is used to acquire the AI ​​model or to obtain information for uplink or downlink transmission. When the terminal is performing uplink or downlink transmission, it does not occupy the APU or occupies only one APU. This helps to save APU resources.

[0176] Based on the first information, the terminal allocates APU within the first time window or on the first resource. The allocated APU is used to acquire the AI ​​model or to obtain information for uplink or downlink transmission. When actually performing uplink or downlink transmission, the terminal does not allocate APU or only allocates one APU. This is because the actual uplink or downlink transmission is based on table lookup or simple inference. However, acquiring the AI ​​model within the first time window or on the first resource requires consuming more APU.

[0177] The uplink or downlink transmission of a terminal can be understood as the actual uplink or downlink transmission of the terminal. The uplink or downlink transmission of a terminal can be based on the AI ​​model obtained by occupying the APU within the first time window or on the first resource, or on the information obtained by occupying the APU within the first time window or on the first resource for the purpose of uplink or downlink transmission.

[0178] 4) The terminal does not occupy the APU or occupies one APU for uplink or downlink transmission in time units outside the first time window;

[0179] Based on the first information, the terminal occupies the APU within the first time window. The occupied APU is used to acquire the AI ​​model or to obtain information for uplink or downlink transmission. Outside the first time window, the terminal does not occupy the APU, or it occupies one APU for uplink or downlink transmission. This helps to save APU resources and avoid waste.

[0180] 5) The terminal does not perform uplink or downlink transmission within the first time window or on the first resource.

[0181] Based on the first information, the terminal occupies the APU within the first time window or on the first resource. The occupied APU is used to acquire the AI ​​model or to obtain information for uplink or downlink transmission. Within the first time window or on the first resource, the terminal does not perform uplink or downlink transmissions scheduled by the network-side equipment. This can also be understood as the terminal autonomously sending uplink data within the first time window or on the first resource to collect data and obtain the AI ​​model actually used for uplink or downlink transmission. Alternatively, if the APU is occupied within the first time window or on the first resource to obtain information for downlink transmission, then no uplink transmission is performed within the first time window or on the first resource. For example, receiving signals or data within the first time window to determine channel estimation or the AI ​​model of the intelligent receiver, no uplink transmission is performed within the first time window. Similarly, if the APU is occupied within the first time window or on the first resource to obtain information for uplink transmission, then no downlink transmission is performed within the first time window or on the first resource. For example, within the first time window, various configuration signals can be automatically transmitted and received to calibrate the AI-based Digital Pre-Distortion (DPD) model, while no downlink transmission occurs within the first time window. Alternatively, within the first time window, various configuration signals or data can be automatically transmitted and received to calibrate the AI-based Digital Pre-Distortion (DPD) model, while uplink signals or data are sent according to the terminal's needs within the first time window, omitting or not sending uplink signals or data scheduled by the network side.

[0182] Figure 8 shows the block diagram of the DPD principle. The basic principle of predistortion is: a predistortion processing module is set before the power amplifier. The combined effect of the predistortion processing module and the power amplifier module linearizes the overall input-output characteristics, and the output power is fully utilized.

[0183] Where x(n) represents the system input signal, z(n) represents the system output signal, and y(n) represents the output of the predistortion processing module.

[0184] AI-based DPD models refer to models that use AI to fit the inverse function of a power amplifier's response to obtain a more accurate predistortion model. As shown in Figure 8, the role of the AI-based DPD model is to fit the inverse function of the power amplifier's response; therefore, the input and labels for model training are the power amplifier's output signal and input signal, respectively.

[0185] In some embodiments of this application, the relevant information for the first time window includes at least one of the following:

[0186] The duration of the first time window; the type of the first time window; the use case information associated with the first time window; the period of the first time window; the start time of the first time window; the end time of the first time window; the priority of the first time window; the relevant information of the AI ​​model associated with the first time window; and the business information associated with the first time window.

[0187] The information related to the AI ​​model associated with the first time window may include configuration information related to the AI ​​model; the service information associated with the first time window is used to indicate the specific service used by the first time window, such as downlink transmission, uplink transmission, power adjustment, etc.

[0188] With at least one of the above pieces of information, the terminal can clearly identify the first time window for occupying the APU.

[0189] In some embodiments of this application, the relevant information of the first resource may include at least one of the following:

[0190] Used for acquiring uplink data of AI models; used for acquiring uplink channels of AI models; used for acquiring uplink signal resources of AI models; used for acquiring data resources of AI models; used for acquiring downlink data of AI models; used for acquiring downlink channels of AI models; used for acquiring downlink signal resources of AI models; used for data collection resources associated with uplink transmission; used for monitoring resources associated with uplink transmission; used for data collection resources associated with downlink transmission; used for monitoring resources associated with downlink transmission.

[0191] Among them, the uplink signal resources used to acquire AI models can be sounding reference signal (SRS) resources, such as SRS associated with use cases related to acquiring AI models, or SRS not used for reception by network-side devices;

[0192] Downlink or uplink data used to acquire AI models can include scheduling constraints, such as specified power and bandwidth limits, for acquiring AI models.

[0193] With at least one of the above pieces of information, the terminal can clearly occupy the primary resource of the APU.

[0194] In some embodiments of this application, the duration of APU occupation within the first time window is the same as the duration of the first time window, or the duration of APU occupation within the first time window is the duration from the start time of the first time window to the first reporting time, where the first reporting time is the time when the terminal sends response information to the network-side device.

[0195] In this embodiment of the application, the terminal occupies the APU within a first time window based on the first information in order to obtain an AI model or information for uplink or downlink transmission. The duration of APU occupation is the same as the duration of the first time window.

[0196] Alternatively, the terminal collects data within the first time window, and then needs some time outside the first time window to adjust the AI ​​model. After the adjustment is complete, it can send response information, such as information related to the AI ​​model, to the network-side device. The time of sending the response information is the first reporting time. The terminal occupies the APU within the first time window, and the APU occupancy time is the duration from the start time of the first time window to the first reporting time. This ensures the integrity of the AI ​​model.

[0197] In some embodiments of this application, the APU occupied within the first time window or on the first resource includes all remaining available APUs of the terminal.

[0198] It's understandable that the first time window is only used to acquire AI models or obtain information for downlink or uplink transmission. Therefore, the APUs used within the first time window can be all the remaining available APUs in the terminal. This allows for the maximum utilization of APU resources within the first time window.

[0199] Optionally, the number of APUs available to the terminal can be considered independently within and outside the first time window.

[0200] In some embodiments of this application, the APU occupied within a first time window or on a first resource is used to acquire a third AI model, wherein:

[0201] The third AI model correlates the number of APUs used for adjusting the third AI model with the number of APUs used for inference in the third AI model;

[0202] Alternatively, the third AI model can be associated with the first or second time unit, as well as the first time window or the first resource.

[0203] It should be noted that the third AI model may be the same as or different from the first or second AI model.

[0204] Optionally, the state of the third AI model may include at least one of the following:

[0205] Activated state; Inactive state; Adjusted state; Training state; Inference state.

[0206] In some embodiments of this application, the computing power of an APU is determined based on at least one of the number of reference parameters, the number of APUs corresponding to the reference AI model, and the number of APUs corresponding to the reference AI function.

[0207] In this embodiment of the application, in order to achieve unified management of APUs, it is necessary to unify the AI ​​metric unit and define a minimum APU. A unified AI metric unit can abstract common AI information that is independent of use cases, including AI computing power, AI parallel processing capability, AI application time, runtime, supported AI complexity, AI storage space, etc.

[0208] Optionally, the computing power of an APU can be determined based on the number of reference parameters. The reference parameters can be floating-point operations per second (FLOPs). For example, a certain number of FLOPs can be defined as an APU, or an order of magnitude of FLOPs can be defined as an APU.

[0209] Optionally, the computing power of an APU can be determined based on the number of APUs corresponding to a reference AI model. For example, the number of APUs corresponding to a reference AI model can be agreed upon through a protocol, thereby determining the computing power of an APU. For other AI models, the number of APUs corresponding to that other AI model can be obtained based on the computing power of an APU.

[0210] Optionally, the computing power of an APU can be determined based on the number of APUs corresponding to a reference AI function. For example, different AI modules or functions can be converted into several APUs, or one or more reference AI functions can be designated as several APUs, thereby calculating the number of APUs corresponding to different AI models.

[0211] In some embodiments of this application, the method may further include the following steps:

[0212] The terminal sends capability information to the network-side equipment;

[0213] The capability information includes at least one of the following:

[0214] Maximum number of APUs running; maximum number of APUs running simultaneously; processing time required for one or more AI models using the maximum number of APUs; processing time required for one or more AI models using the third largest number of APUs; maximum number of APUs stored; information about the AI ​​models stored.

[0215] Optionally, the stored information about the AI ​​model includes at least one of the following:

[0216] The maximum number of AI models stored; the input information associated with the stored AI models; the output information associated with the stored AI models; the signal information associated with the stored AI models; the configuration information associated with the stored AI models; and the APU information associated with the stored AI models.

[0217] In this embodiment, the terminal sends capability information to the network-side device, and the network-side device receives the capability information from the terminal, enabling the network-side device to perform accurate scheduling based on the terminal's capability information.

[0218] Capability information may include at least one of the above. The maximum number of running APUs can have multiple options, such as 1, 2, ..., L1, etc. The maximum number of APUs running simultaneously can be understood as the maximum number of APUs that can be scheduled or run in parallel, or the maximum number of APUs that can be scheduled within a certain period, and has multiple options, such as 1, 2, ..., L2, etc. The maximum number of running APUs simultaneously is less than or equal to the maximum number of running APUs. The processing time required for one or more AI models running the maximum number of APUs has multiple options, such as 1, 2, ..., K candidate time lengths. The maximum number of stored APUs can be the same as the maximum number of running APUs; it can be assumed that any APU that can be stored or loaded can be used for computation. Alternatively, the maximum number of running APUs is less than the maximum number of stored APUs; it can be assumed that there are more APUs that can be stored, but the actual number of APUs that can run simultaneously is limited.

[0219] Considering that the actual AI model (which can be understood as software) on the terminal side can be updated and changed, and different AI models may be used in different scenarios, use cases, network conditions, etc., it is necessary to report the relevant information of the stored AI model, such as the input information, output information, signal information, configuration information, APU information associated with the AI ​​model, to the network side device.

[0220] In some embodiments of this application, the method may further include the following steps:

[0221] The terminal controls the loading and running of AI models through the AI ​​management layer.

[0222] In this embodiment of the application, a unified AI management layer can be deployed. The AI ​​management layer can also be understood as an AI management unit. The AI ​​management layer can manage the AI ​​business or AI models that need to be performed, such as which AI models can perform inference and which AI business has a higher priority.

[0223] The terminal can control the loading and running of AI models through the AI ​​management layer. For example, the AI ​​management layer can control the start, stop, and operation of AI model inference and monitoring, or determine the priority of AI services based on the terminal's capability information.

[0224] In one embodiment, the number of AI models that can be loaded onto the chip running the AI ​​needs to be determined, and all models can be run after loading. The loading and unloading times need to be determined.

[0225] Optionally, the chip running AI can run a CPU (which can be used for non-AI or AI processing units) and an APU (a processing unit dedicated to AI).

[0226] Optionally, it may be necessary to further determine whether the loaded AI model can run based on CPU usage.

[0227] In another embodiment, it is checked whether the AI ​​model can complete its operation before the reference time.

[0228] Optionally, a reference time can be defined.

[0229] Optionally, if the AI ​​model cannot complete the run before the reference time, report "No available APU resources";

[0230] Alternatively, the reference time can be determined based on the maximum running time of the AI ​​model.

[0231] The above embodiments can be combined to achieve refined management and efficient utilization of APU resources, thereby improving communication efficiency.

[0232] Corresponding to the above method embodiments, this application embodiment also provides an AI processing unit (APU) management method, as shown in Figure 9, which includes the following steps:

[0233] S910: The network-side device sends first information to the terminal. The first information is used to indicate relevant information of the first time window or relevant information of the first resource.

[0234] Alternatively, the network-side device receives second information from the terminal, which is used to indicate relevant information about the artificial intelligence processing unit (APU) occupied by the terminal.

[0235] Applying the method provided in the embodiments of this application, the network-side device sends first information to the terminal. The first information indicates relevant information of a first time window or relevant information of a first resource. Based on the first information, the terminal occupies the APU within the first time window or on the first resource. Alternatively, the network-side device receives second information from the terminal. The second information is used to indicate relevant information of the APU occupied by the terminal, clarifying the management scheme for the APU. This enables both communicating parties to have the same understanding of the occupied APU and the remaining APU, thereby enabling reasonable invocation of AI model-related functions, achieving effective management of the APU, and helping to improve communication efficiency.

[0236] In some embodiments of this application, the second information includes at least one of the following:

[0237] The terminal occupies the first number of APUs in the first time unit;

[0238] The terminal occupies a second number of APUs in the second time unit;

[0239] The number of APUs used by the terminal;

[0240] The number of APUs occupied by the terminal within the first time window or on the first resource.

[0241] In some embodiments of this application, the first time unit includes one of the following:

[0242] Downlink time units used for downlink transmission;

[0243] Downlink time units for downlink transmission, and a third time unit, the third time unit including uplink time units or flexible time units associated with downlink transmission.

[0244] In some embodiments of this application, the second time unit includes one of the following:

[0245] Uplink time unit used for uplink transmission;

[0246] Uplink time unit for uplink transmission, and a fourth time unit, the fourth time unit including a downlink time unit or a flexible time unit for preparing for uplink transmission.

[0247] The uplink time unit used for uplink transmission, and the time unit between the downlink time unit that receives downlink control information and the time unit before uplink transmission;

[0248] The time unit between the downlink time unit from the receipt of downlink control information and the uplink time unit used for uplink transmission.

[0249] In some embodiments of this application, the first time unit is a time unit within an energy-saving window, which is a window in which the terminal periodically listens to downlink signals in a power-saving state.

[0250] In some embodiments of this application, the energy-saving window includes an active energy-saving window.

[0251] In some embodiments of this application, the relevant information for the first time window includes at least one of the following:

[0252] The duration of the first time window; the type of the first time window; the use case information associated with the first time window; the period of the first time window; the start time of the first time window; the end time of the first time window; the priority of the first time window; the information of the AI ​​model associated with the first time window; and the business information associated with the first time window.

[0253] In some embodiments of this application, the relevant information of the first resource includes at least one of the following:

[0254] Used for acquiring uplink data of AI models; used for acquiring uplink channels of AI models; used for acquiring uplink signal resources of AI models; used for acquiring data resources of AI models; used for acquiring downlink data of AI models; used for acquiring downlink channels of AI models; used for acquiring downlink signal resources of AI models; used for data collection resources associated with uplink transmission; used for monitoring resources associated with uplink transmission; used for data collection resources associated with downlink transmission; used for monitoring resources associated with downlink transmission.

[0255] In some embodiments of this application, the computing power of an APU is determined based on at least one of the following: the number of reference parameters, the number of APUs corresponding to the reference AI model, the number of APUs corresponding to the reference AI function, the protocol agreement, and the indication information of the network-side device.

[0256] In some embodiments of this application, the method further includes:

[0257] Network-side devices receive capability information from the terminal;

[0258] The capability information includes at least one of the following:

[0259] Maximum number of APUs running; maximum number of APUs running simultaneously; processing time required for one or more AI models using the maximum number of APUs; processing time required for one or more AI models using the third largest number of APUs; maximum number of APUs stored; information about the AI ​​models stored.

[0260] The AI ​​processing unit (APU) management method provided in this application embodiment can implement the various processes implemented in the method embodiment shown in Figure 4 and achieve the same technical effect. To avoid repetition, it will not be described again here.

[0261] The AI ​​processing unit (APU) management method provided in this application can be executed by an AI processing unit (APU) management device. This application uses the execution of the AI ​​processing unit (APU) management method by an AI processing unit (APU) management device as an example to illustrate the AI ​​processing unit (APU) management device provided in this application.

[0262] This application provides an AI processing unit (APU) management device. As an example, the AI ​​processing unit (APU) management device can be a communication device or a component within a communication device, such as a chip. The communication device can be a terminal, a network-side device, or a server, etc. Exemplarily, the terminal can be, but is not limited to, the type of terminal 11 listed above, and the network-side device can be, but is not limited to, the type of network-side device 12 listed above. This application does not impose specific limitations on these aspects.

[0263] The AI ​​processing unit (APU) management device includes a receiving module, a transmitting module, and a processing module. These modules can be implemented in software or hardware. When implemented in hardware, the processing module can be implemented by a processor. For example, the processor can include general-purpose processors, special-purpose processors, etc., such as central processing units (CPUs), microprocessors, digital signal processors (DSPs), artificial intelligence (AI) processors, graphics processing units (GPUs), application-specific integrated circuits (ASICs), network processors (NPs), field-programmable gate arrays (FPGAs), or other programmable logic devices, gate circuits, transistors, discrete hardware components, etc. The receiving and transmitting modules can be implemented by a communication interface, which can include one or more of the following: transceivers, pins, circuits, buses, radio frequency units, etc.

[0264] Specifically, referring to Figure 10, when the AI ​​processing unit (APU) management device is a terminal or a component within a terminal, the AI ​​processing unit (APU) management device 1000 includes:

[0265] The first processing module 1010 is used to occupy a first number of AI processing units (APUs) in a first time unit when enabling or activating downlink transmission based on a first artificial intelligence (AI) model.

[0266] Alternatively, when uplink transmission based on the second AI model is enabled or activated, a second number of APUs may be occupied in the second time unit.

[0267] Alternatively, based on first information, the APU may be occupied within a first time window or on a first resource, whereby the first information is used to indicate relevant information of the first time window or relevant information of the first resource.

[0268] The apparatus provided in this application occupies a first number of APUs in a first time unit when enabling or activating downlink transmission of the first AI model, or occupies a second number of APUs in a second time unit when enabling or activating uplink transmission based on the second AI model, or occupies APUs within a first time window or on a first resource based on first information. This clarifies the APU management scheme, enabling both communicating parties to have the same understanding of the occupied APUs and the remaining APUs, thereby enabling reasonable invocation of AI model-related functions, achieving effective APU management, and helping to improve communication efficiency.

[0269] Optionally, the first time unit includes one of the following:

[0270] Downlink time units used for downlink transmission;

[0271] Downlink time units for downlink transmission, and a third time unit, the third time unit including uplink time units or flexible time units associated with downlink transmission.

[0272] Optionally, the second time unit includes one of the following:

[0273] Uplink time unit used for uplink transmission;

[0274] Uplink time unit for uplink transmission, and a fourth time unit, the fourth time unit including a downlink time unit or a flexible time unit for preparing for uplink transmission.

[0275] The uplink time unit used for uplink transmission, and the time unit between the downlink time unit that receives downlink control information and the time unit before uplink transmission;

[0276] The time unit between the downlink time unit from the receipt of downlink control information and the uplink time unit used for uplink transmission.

[0277] Optionally, the first quantity or the second quantity is determined based on at least one of the following:

[0278] Terminal capability information; number of APUs reported by the terminal; relevant information about the AI ​​model used; protocol conventions; and instructions from network-side devices.

[0279] Optionally, if there is an overlap between the second time unit and the first time unit, the first processing module 1010 is further configured to:

[0280] Determine at least one of downlink and uplink transmissions in the overlapping time unit based on at least one of the following:

[0281] The protocol specifies; the indication information of the network-side equipment; and the priority information corresponding to downlink and uplink transmissions.

[0282] Optionally, the first processing module 1010 is specifically used for:

[0283] If, during an overlapping time unit, the number of APUs required for downlink and uplink transmissions exceeds the number of remaining available APUs in the terminal, then it is determined that one of the downlink and uplink transmissions should be performed during the overlapping time unit. Optionally, a first transmitting module is also included, configured to:

[0284] Send a second message to the network-side device. The second message is used to indicate the relevant information of the APU occupied by the terminal.

[0285] Optionally, the first time unit is a time unit within the energy-saving window, which is a window in which the terminal periodically listens to downlink signals in a power-saving state.

[0286] Optionally, the energy-saving window includes the activated energy-saving window.

[0287] Optionally, a second transmitting module is also included, for:

[0288] Based on the first information, and assuming APU usage within the first time window or on the first resource:

[0289] Send the number of APUs occupied to the network-side devices;

[0290] Alternatively, send the number of APUs occupied in the first time window or on the first resource to the network-side device.

[0291] Optionally, the first processing module 1010 is further configured to:

[0292] Based on the first information, and assuming APU usage within the first time window or on the first resource:

[0293] In the case of uplink or downlink transmission, it does not occupy the APU or occupies one APU;

[0294] Alternatively, in time units outside the first time window, the APU may not be used or one APU may be used for uplink or downlink transmission.

[0295] Alternatively, no uplink or downlink transmission may be performed within the first time window or on the first resource.

[0296] Optionally, the relevant information for the first time window includes at least one of the following:

[0297] The duration of the first time window; the type of the first time window; the use case information associated with the first time window; the period of the first time window; the start time of the first time window; the end time of the first time window; the priority of the first time window; the relevant information of the AI ​​model associated with the first time window; and the business information associated with the first time window.

[0298] Optionally, the relevant information of the first resource includes at least one of the following:

[0299] Used for acquiring uplink data of AI models; used for acquiring uplink channels of AI models; used for acquiring uplink signal resources of AI models; used for acquiring data resources of AI models; used for acquiring downlink data of AI models; used for acquiring downlink channels of AI models; used for acquiring downlink signal resources of AI models; used for data collection resources associated with uplink transmission; used for monitoring resources associated with uplink transmission; used for data collection resources associated with downlink transmission; used for monitoring resources associated with downlink transmission.

[0300] Optionally, the duration of APU usage within the first time window is the same as the duration of the first time window, or the duration of APU usage within the first time window is the duration from the start time of the first time window to the first reporting time, where the first reporting time is the time when the terminal sends response information to the network-side device.

[0301] Optionally, the APUs occupied within the first time window or on the first resource include all remaining available APUs in the terminal.

[0302] Optionally, the APUs used within the first time window or on the first resource are used to acquire a third AI model, wherein:

[0303] The third AI model correlates the number of APUs used for adjusting the third AI model with the number of APUs used for inference in the third AI model;

[0304] Alternatively, the third AI model can be associated with the first or second time unit, as well as the first time window or the first resource.

[0305] Optionally, the state of the third AI model includes at least one of the following:

[0306] Activated state; Inactive state; Adjusted state; Training state; Inference state.

[0307] Optionally, the computing power of an APU is determined based on at least one of the following: the number of reference parameters, the number of APUs corresponding to the reference AI model, the number of APUs corresponding to the reference AI function, the protocol agreement, and the indication information of the network-side device.

[0308] Optionally, a third transmitting module is also included, for:

[0309] Send capability information to network-side devices;

[0310] The capability information includes at least one of the following:

[0311] Maximum number of APUs running; maximum number of APUs running simultaneously; processing time required for one or more AI models using the maximum number of APUs; processing time required for one or more AI models using the third largest number of APUs; maximum number of APUs stored; information about the AI ​​models stored.

[0312] Optionally, the first processing module 1010 is further configured to:

[0313] The AI ​​management layer controls the loading and execution of AI models.

[0314] The AI ​​processing unit (APU) management device provided in this application embodiment can implement the various processes implemented in the method embodiment shown in Figure 4 and achieve the same technical effect. To avoid repetition, it will not be described again here.

[0315] Referring to Figure 11, when the AI ​​processing unit (APU) management device is a network-side device or a component within a network-side device, the AI ​​processing unit (APU) management device 1100 includes a fourth transmitting module 1110 or a first receiving module 1120, wherein:

[0316] The fourth sending module 1110 is used to send first information to the terminal. The first information is used to indicate relevant information of the first time window or relevant information of the first resource.

[0317] The first receiving module 1120 is used to receive second information from the terminal, the second information being used to indicate relevant information about the artificial intelligence processing unit (APU) occupied by the terminal.

[0318] Using the apparatus provided in this application embodiment, a first message is sent to the terminal. The first message indicates relevant information of a first time window or relevant information of a first resource. Based on the first message, the terminal occupies the APU within the first time window or on the first resource. Alternatively, the network-side device receives a second message from the terminal. The second message indicates relevant information of the APU occupied by the terminal, clarifying the APU management scheme. This enables both communicating parties to have the same understanding of the occupied APU and the remaining APU, thereby enabling reasonable invocation of AI model-related functions, achieving effective APU management, and helping to improve communication efficiency.

[0319] Optionally, the second information includes at least one of the following:

[0320] The terminal occupies a first number of APUs in a first time unit; the terminal occupies a second number of APUs in a second time unit; the number of APUs occupied by the terminal; the number of APUs occupied by the terminal within a first time window or on a first resource.

[0321] Optionally, the first time unit includes one of the following:

[0322] Downlink time units used for downlink transmission;

[0323] Downlink time units for downlink transmission, and a third time unit, the third time unit including uplink time units or flexible time units associated with downlink transmission.

[0324] Optionally, the second time unit includes one of the following:

[0325] Uplink time unit used for uplink transmission;

[0326] Uplink time unit for uplink transmission, and a fourth time unit, the fourth time unit including a downlink time unit or a flexible time unit for preparing for uplink transmission.

[0327] The uplink time unit used for uplink transmission, and the time unit between the downlink time unit that receives downlink control information and the time unit before uplink transmission;

[0328] The time unit between the downlink time unit from the receipt of downlink control information and the uplink time unit used for uplink transmission.

[0329] Optionally, the first time unit is a time unit within the energy-saving window, which is a window in which the terminal periodically listens to downlink signals in a power-saving state.

[0330] Optionally, the energy-saving window includes the activated energy-saving window.

[0331] Optionally, the relevant information for the first time window includes at least one of the following:

[0332] The duration of the first time window; the type of the first time window; the use case information associated with the first time window; the period of the first time window; the start time of the first time window; the end time of the first time window; the priority of the first time window; the information of the AI ​​model associated with the first time window; and the business information associated with the first time window.

[0333] Optionally, the relevant information of the first resource includes at least one of the following:

[0334] Used for acquiring uplink data of AI models; used for acquiring uplink channels of AI models; used for acquiring uplink signal resources of AI models; used for acquiring data resources of AI models; used for acquiring downlink data of AI models; used for acquiring downlink channels of AI models; used for acquiring downlink signal resources of AI models; used for data collection resources associated with uplink transmission; used for monitoring resources associated with uplink transmission; used for data collection resources associated with downlink transmission; used for monitoring resources associated with downlink transmission.

[0335] Optionally, the computing power of an APU is determined based on at least one of the following: the number of reference parameters, the number of APUs corresponding to the reference AI model, the number of APUs corresponding to the reference AI function, the protocol agreement, and the indication information of the network-side device.

[0336] Optionally, the first receiving module 1120 is also used for:

[0337] Receive capability information from the terminal; wherein the capability information includes at least one of the following:

[0338] Maximum number of APUs running; maximum number of APUs running simultaneously; processing time required for one or more AI models using the maximum number of APUs; processing time required for one or more AI models using the third largest number of APUs; maximum number of APUs stored; information about the AI ​​models stored.

[0339] The AI ​​processing unit (APU) management device provided in this application embodiment can implement the various processes implemented in the method embodiment shown in FIG9 and achieve the same technical effect. To avoid repetition, it will not be described again here.

[0340] As shown in Figure 12, this application embodiment also provides a communication device 1200, including a processor 1201 and a memory 1202. The memory 1202 stores a program or instructions that can run on the processor 1201. For example, when the communication device 1200 is a terminal, the program or instructions executed by the processor 1201 implement the various steps of the above-described terminal-side method embodiment and achieve the same technical effect. When the communication device 1200 is a network-side device, the program or instructions executed by the processor 1201 implement the various steps of the above-described network-side device-side method embodiment and achieve the same technical effect. To avoid repetition, this will not be described again here.

[0341] This application also provides a terminal, including a processor and a communication interface, with the communication interface coupled to the processor. The processor is used to run programs or instructions to implement the steps in the method embodiment shown in FIG4. This terminal embodiment corresponds to the above-described terminal-side method embodiment, and all implementation processes and methods of the above-described method embodiments can be applied to this terminal embodiment and achieve the same technical effect. This terminal may be the AI ​​processing unit (APU) management device shown in FIG10.

[0342] Specifically, Figure 13 is a schematic diagram of the structure of a terminal implementing an embodiment of this application.

[0343] The terminal 1300 includes, but is not limited to, at least some of the following components: radio frequency unit 1301, network module 1302, audio output unit 1303, input unit 1304, sensor 1305, display unit 1306, user input unit 1307, interface unit 1308, memory 1309, and processor 1310.

[0344] Those skilled in the art will understand that terminal 1300 may also include a power supply (such as a battery) for powering various components. The power supply can be logically connected to processor 1310 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system. The terminal structure shown in Figure 13 does not constitute a limitation on the terminal. The terminal may include more or fewer components than shown, or combine certain components, or have different component arrangements, which will not be elaborated here.

[0345] It should be understood that, in this embodiment, the input unit 1304 may include a graphics processor 13041 and a microphone 13042. The graphics processor 13041 processes image data of still images or videos obtained by an image capture device (such as a camera) in video capture mode or image capture mode. The display unit 1306 may include a display panel 13061, which may be configured in the form of a liquid crystal display, an organic light-emitting diode, or the like. The user input unit 1307 includes a touch panel 13071 and at least one of other input devices 13072. The touch panel 13071 is also called a touch screen. The touch panel 13071 may include a touch detection device and a touch controller. Other input devices 13072 may include, but are not limited to, physical keyboards, function keys (such as volume control buttons, power buttons, etc.), trackballs, mice, and joysticks, which will not be described in detail here.

[0346] In this embodiment, after receiving downlink data from the network-side device, the radio frequency unit 1301 can transmit it to the processor 1310 for processing; in addition, the radio frequency unit 1301 can send uplink data to the network-side device. Typically, the radio frequency unit 1301 includes, but is not limited to, antennas, amplifiers, transceivers, couplers, low-noise amplifiers, duplexers, etc.

[0347] The memory 1309 can be used to store software programs or instructions, as well as various data. The memory 1309 may primarily include a first storage area for storing programs or instructions and a second storage area for storing data. The first storage area may store the operating system, application programs or instructions required for at least one function (such as sound playback, image playback, etc.). Furthermore, the memory 1309 may include volatile memory or non-volatile memory. The non-volatile memory may 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. Volatile memory can be random access memory (RAM), static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct memory bus RAM (DRRAM). The memory 1309 in this embodiment includes, but is not limited to, these and any other suitable types of memory.

[0348] Processor 1310 may include one or more processing units; optionally, processor 1310 integrates an application processor and a modem processor, wherein the application processor mainly handles operations involving the operating system, user interface, and applications, and the modem processor mainly handles wireless communication signals, such as a baseband processor. It is understood that the aforementioned modem processor may also not be integrated into processor 1310.

[0349] The processor 1310 is used to occupy a first number of AI processing units (APUs) in a first time unit when enabling or activating downlink transmission based on a first artificial intelligence (AI) model.

[0350] Alternatively, when uplink transmission based on the second AI model is enabled or activated, a second number of APUs may be occupied in the second time unit.

[0351] Alternatively, based on first information, the APU may be occupied within a first time window or on a first resource, whereby the first information is used to indicate relevant information of the first time window or relevant information of the first resource.

[0352] In this embodiment, when enabling or activating downlink transmission of the first AI model, a first number of APUs are occupied in the first time unit; or, when enabling or activating uplink transmission based on the second AI model, a second number of APUs are occupied in the second time unit; or, based on the first information, APUs are occupied within the first time window or on the first resource. This clarifies the APU management scheme, enabling both communicating parties to have the same understanding of the occupied and remaining APUs, thereby enabling the reasonable invocation of AI model-related functions, achieving effective APU management, and helping to improve communication efficiency.

[0353] It is understood that the implementation process of each implementation method mentioned in this embodiment can refer to the relevant description of the method embodiment and achieve the same or corresponding technical effect. To avoid repetition, it will not be described again here.

[0354] This application also provides a network-side device, including a processor and a communication interface, wherein the communication interface and the processor are coupled, and the processor is used to run programs or instructions to implement the steps of the method embodiment shown in FIG9. This network-side device embodiment corresponds to the above-described network-side device method embodiment, and all implementation processes and methods of the above-described method embodiments can be applied to this network-side device embodiment and can achieve the same technical effect.

[0355] Specifically, this application embodiment also provides a network-side device, which may be the AI ​​processing unit (APU) management device shown in FIG11. As shown in FIG14, the network-side device 1400 includes: an antenna 1401, a radio frequency device 1402, a baseband device 1403, a processor 1404, and a memory 1405. The antenna 1401 is connected to the radio frequency device 1402. In the uplink direction, the radio frequency device 1402 receives information through the antenna 1401 and sends the received information to the baseband device 1403 for processing. In the downlink direction, the baseband device 1403 processes the information to be transmitted and sends it to the radio frequency device 1402, which processes the received information and then transmits it through the antenna 1401.

[0356] The method executed by the network-side device in the above embodiments can be implemented in the baseband device 1403, which includes a baseband processor.

[0357] The baseband device 1403 may include at least one baseband board, on which multiple chips are disposed, as shown in FIG14. One of the chips is, for example, a baseband processor, which is connected to the memory 1405 via a bus interface to call the program or instructions in the memory 1405 to execute the network-side device operation shown in the above method embodiment.

[0358] The network-side device may also include a network interface 1406, such as a Common Public Radio Interface (CPRI).

[0359] The radio frequency device 1402 is used to send first information to the terminal, the first information being used to indicate relevant information of the first time window or relevant information of the first resource.

[0360] Alternatively, a second piece of information may be received from the terminal, which indicates relevant information about the artificial intelligence processing unit (APU) occupied by the terminal.

[0361] In this embodiment, a first message is sent to the terminal, indicating relevant information of a first time window or relevant information of a first resource. Based on the first message, the terminal occupies the APU within the first time window or on the first resource. Alternatively, the terminal receives a second message, which indicates relevant information of the APU occupied by the terminal. This clarifies the APU management scheme, enabling both communicating parties to have the same understanding of the occupied and remaining APUs. This allows for the reasonable invocation of AI model-related functions, achieving effective APU management and improving communication efficiency.

[0362] In addition, the network-side device 1400 of this application embodiment also includes: a program or instructions stored in the memory 1405 and executable on the processor 1404. The processor 1404 calls the program or instructions in the memory 1405 to execute the methods executed by each module shown in FIG11 and achieve the same technical effect. To avoid repetition, it will not be described in detail here.

[0363] Specifically, this application embodiment also provides a network-side device. As shown in FIG15, the network-side device 1500 includes: a processor 1501, a network interface 1502, and a memory 1503. The network-side device may be the AI ​​processing unit (APU) management device shown in FIG11. The network interface 1502 is, for example, a common public radio interface (CPRI).

[0364] Among them, network interface 1502 is used to send first information to the terminal, the first information being used to indicate relevant information of the first time window or relevant information of the first resource;

[0365] Alternatively, a second piece of information may be received from the terminal, which indicates relevant information about the artificial intelligence processing unit (APU) occupied by the terminal.

[0366] In this embodiment, a first message is sent to the terminal, indicating relevant information of a first time window or relevant information of a first resource. Based on the first message, the terminal occupies the APU within the first time window or on the first resource. Alternatively, the terminal receives a second message, which indicates relevant information of the APU occupied by the terminal. This clarifies the APU management scheme, enabling both communicating parties to have the same understanding of the occupied and remaining APUs. This allows for the reasonable invocation of AI model-related functions, achieving effective APU management and improving communication efficiency.

[0367] In addition, the network-side device 1500 of this application embodiment also includes: a program or instructions stored in the memory 1503 and executable on the processor 1501. The processor 1501 calls the program or instructions in the memory 1503 to execute the methods executed by each module shown in FIG11 and achieve the same technical effect. To avoid repetition, it will not be described in detail here.

[0368] This application also provides a readable storage medium storing a program or instructions. When the program or instructions are executed by a processor, they implement the various processes of the above method embodiments and achieve the same technical effect. To avoid repetition, they will not be described again here.

[0369] The processor mentioned above is either the processor in the terminal described in the above embodiments or the processor in the network-side device. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk. In some examples, the readable storage medium may be a non-transient readable storage medium.

[0370] This application embodiment also provides a chip, which includes a processor and a communication interface. The communication interface is coupled to the processor. The processor is used to run programs or instructions to implement the various processes of the above method embodiments and achieve the same technical effect. To avoid repetition, it will not be described again here.

[0371] 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.

[0372] This application also provides a computer program / program product, which is stored in a storage medium and executed by at least one processor to implement the various processes of the above method embodiments and achieve the same technical effect. To avoid repetition, it will not be described again here.

[0373] This application also provides a wireless communication system, including: a terminal and a network-side device. The terminal can be used to execute the steps of the method embodiment shown in FIG4, and the network-side device can be used to execute the steps of the method embodiment shown in FIG9.

[0374] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0375] From the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of computer software products plus necessary general-purpose hardware platforms, and of course, they can also be implemented by hardware. The computer software product is stored in a storage medium (such as ROM, RAM, magnetic disk, optical disk, etc.), and the computer software product includes several instructions to cause the terminal or network-side device to execute the methods described in the various embodiments of this application.

[0376] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other implementations under the guidance of this application without departing from the spirit and scope of the claims. All of these implementations are within the protection scope of this application.

Claims

1. A method for managing an AI processing unit (APU), comprising: When the terminal enables or activates downlink transmission based on the first artificial intelligence (AI) model, it occupies the first number of AI processing units (APUs) in the first time unit. Alternatively, when the terminal enables or activates uplink transmission based on the second AI model, it occupies a second number of APUs in the second time unit. Alternatively, the terminal may occupy the APU within a first time window or on a first resource based on first information, wherein the first information is used to indicate relevant information of the first time window or relevant information of the first resource.

2. The method of claim 1, wherein, The first time unit includes one of the following: Downlink time units used for downlink transmission; Downlink time units for downlink transmission, and a third time unit, the third time unit including an uplink time unit or a flexible time unit associated with downlink transmission.

3. The method of claim 1 or 2, wherein, The second time unit includes one of the following: Uplink time unit used for uplink transmission; Uplink time unit for uplink transmission, and a fourth time unit, the fourth time unit including a downlink time unit or a flexible time unit for preparing for uplink transmission. The uplink time unit used for uplink transmission, and the time unit between the downlink time unit that receives downlink control information and the time unit before uplink transmission; The time unit between the downlink time unit from the receipt of downlink control information and the uplink time unit used for uplink transmission.

4. The method of any one of claims 1 to 3, wherein, The first quantity or the second quantity is determined based on at least one of the following: The terminal's capability information; The number of APUs reported by the terminal; Information related to the AI ​​model used; The agreement stipulates; Indication information from network-side devices.

5. The method of any one of claims 1 to 4, wherein, When the second time unit overlaps with the first time unit, the method further includes: The terminal determines, based on at least one of the following, whether to perform downlink transmission or uplink transmission in the overlapping time unit: The agreement stipulates; Indication information from network-side devices; Priority information for downlink and uplink transmissions.

6. The method of claim 5, wherein, The terminal determines to perform at least one of downlink transmission and uplink transmission in the overlapping time unit, including: If, during the overlapping time unit, the number of APUs required for downlink and uplink transmissions exceeds the number of APUs remaining available in the terminal, the terminal determines to perform either downlink or uplink transmission during the overlapping time unit.

7. The method of any one of claims 1 to 6, wherein, The method further includes: The terminal sends a second message to the network-side device, the second message being used to indicate information related to the APU occupied by the terminal.

8. The method of any one of claims 1 to 7, wherein, The first time unit is a time unit within the energy-saving window, which is a window in which the terminal periodically listens to downlink signals in power-saving mode.

9. The method of claim 8, wherein, The energy-saving window includes the active energy-saving window.

10. The method of any one of claims 1 to 9, wherein, When the terminal occupies the APU within a first time window or on a first resource based on the first information, the method further includes the following: The terminal sends the number of APUs it occupies to the network-side device; The terminal sends the number of APUs occupied in the first time window or on the first resource to the network-side device. When the terminal is performing uplink or downlink transmission, it does not occupy the APU or occupies one APU. The terminal does not occupy the APU or occupies one APU for uplink or downlink transmission in time units outside the first time window. The terminal does not perform uplink or downlink transmissions within the first time window or on the first resource.

11. The method of any one of claims 1 to 10, wherein, The relevant information for the first time window includes at least one of the following: The duration of the first time window; The type of the first time window; The use case information associated with the first time window; The period of the first time window; The start time of the first time window; The end time of the first time window; The priority of the first time window; Information related to the AI ​​model associated with the first time window; The business information associated with the first time window.

12. The method of any one of claims 1 to 11, wherein, The relevant information of the first resource includes at least one of the following: Used to acquire uplink data for AI models; Used to obtain the uplink channel of the AI ​​model; Used to acquire uplink signal resources for AI models; Used to acquire data resources for AI models; Used to acquire downlink data for AI models; Used to obtain the downlink channel of the AI ​​model; Used to acquire downlink signal resources for AI models; Data collection resources used in association with uplink transmissions; Monitoring resources associated with uplink transmissions; Data collection resources associated with downlink transmission; Monitoring resources used in association with downlink transmissions.

13. The method of any one of claims 1 to 12, wherein, The duration of APU usage within the first time window is the same as the duration of the first time window, or the duration of APU usage within the first time window is the duration from the start time of the first time window to the first reporting time, where the first reporting time is the time when the terminal sends response information to the network-side device.

14. The method of any one of claims 1 to 13, wherein, The APUs occupied within the first time window or on the first resource include all remaining available APUs of the terminal.

15. The method of any one of claims 1 to 14, wherein, The APU used within the first time window or on the first resource is used to acquire the third AI model, wherein: The third AI model is associated with the number of APUs used for adjusting the third AI model and the number of APUs used for inference in the third AI model; Alternatively, the third AI model may be associated with the first time unit or the second time unit, as well as the first time window or the first resource.

16. The method of claim 15, wherein, The state of the third AI model includes at least one of the following: Activated state; Inactive state; Adjusted state; Training state; Inference state.

17. The method of any one of claims 1 to 16, wherein, The computing power of an APU is determined based on at least one of the following: the number of reference parameters, the number of APUs corresponding to the reference AI model, the number of APUs corresponding to the reference AI function, the protocol agreement, and the indication information from the network-side device.

18. The method of any one of claims 1 to 17, wherein, The method further includes: The terminal sends capability information to the network-side device; The capability information includes at least one of the following: Maximum number of APUs running; The maximum number of APUs that can run simultaneously; The processing time required for one or more AI models that utilize the largest number of APUs; The processing time required for one or more AI models occupying the third number of APUs; Maximum number of APUs to store; Information related to the stored AI model.

19. The method of any one of claims 1 to 18, wherein, The method further includes: The terminal controls the loading and running of AI models through the AI ​​management layer.

20. A method for managing an AI processing unit (APU), comprising: The network-side device sends first information to the terminal, the first information being used to indicate relevant information of the first time window or relevant information of the first resource; Alternatively, the network-side device receives second information from the terminal, which is used to indicate relevant information about the artificial intelligence processing unit (APU) occupied by the terminal.

21. The method of claim 20, wherein, The second information includes at least one of the following: The terminal occupies a first number of APUs in the first time unit; The terminal occupies a second number of APUs in the second time unit; The number of APUs occupied by the terminal; The number of APUs occupied by the terminal within the first time window or on the first resource.

22. The method of claim 21, wherein, The first time unit includes one of the following: Downlink time units used for downlink transmission; Downlink time units for downlink transmission, and a third time unit, the third time unit including an uplink time unit or a flexible time unit associated with downlink transmission.

23. The method of claim 21 or 22, wherein, The second time unit includes one of the following: Uplink time unit used for uplink transmission; Uplink time unit for uplink transmission, and a fourth time unit, the fourth time unit including a downlink time unit or a flexible time unit for preparing for uplink transmission. The uplink time unit used for uplink transmission, and the time unit between the downlink time unit that receives downlink control information and the time unit before uplink transmission; The time unit between the downlink time unit from the receipt of downlink control information and the uplink time unit used for uplink transmission.

24. The method of any one of claims 21-23, wherein, The first time unit is a time unit within the energy-saving window, which is a window in which the terminal periodically listens to downlink signals in power-saving mode.

25. The method of claim 24, wherein, The energy-saving window includes the active energy-saving window.

26. The method of any one of claims 20-25, wherein, The relevant information for the first time window includes at least one of the following: The duration of the first time window; The type of the first time window; The use case information associated with the first time window; The period of the first time window; The start time of the first time window; The end time of the first time window; The priority of the first time window; Information about the AI ​​model associated with the first time window; The business information associated with the first time window.

27. The method of any one of claims 20 to 26, wherein, The relevant information of the first resource includes at least one of the following: Used to acquire uplink data for AI models; Used to obtain the uplink channel of the AI ​​model; Used to acquire uplink signal resources for AI models; Used to acquire data resources for AI models; Used to acquire downlink data for AI models; Used to obtain the downlink channel of the AI ​​model; Used to acquire downlink signal resources for AI models; Data collection resources used in association with uplink transmissions; Monitoring resources associated with uplink transmissions; Data collection resources associated with downlink transmission; Monitoring resources used in association with downlink transmissions.

28. The method of any one of claims 20-27, wherein, The computing power of an APU is determined based on at least one of the following: the number of reference parameters, the number of APUs corresponding to the reference AI model, the number of APUs corresponding to the reference AI function, the protocol agreement, and the indication information from the network-side device.

29. The method of any one of claims 20 to 28, wherein, The method further includes: The network-side device receives capability information from the terminal; The capability information includes at least one of the following: Maximum number of APUs running; The maximum number of APUs that can run simultaneously; The processing time required for one or more AI models that utilize the largest number of APUs; The processing time required for one or more AI models occupying the third number of APUs; Maximum number of APUs to store; Information related to the stored AI model.

30. An AI processing unit (APU) management device, comprising: The first processing module is used to occupy a first number of AI processing units (APUs) in a first time unit when downlink transmission based on a first artificial intelligence (AI) model is enabled or activated. Alternatively, when uplink transmission based on the second AI model is enabled or activated, a second number of APUs may be occupied in the second time unit. Alternatively, based on first information, the APU may be occupied within a first time window or on a first resource, whereby the first information is used to indicate relevant information of the first time window or relevant information of the first resource.

31. The apparatus of claim 30, wherein, When the second time unit overlaps with the first time unit, the first processing module is further configured to: Determine at least one of downlink and uplink transmissions during the overlap time based on at least one of the following: The agreement stipulates; Indication information from network-side devices; Priority information for downlink and uplink transmissions.

32. The apparatus of claim 30 or 31, wherein, It also includes a first sending module, used for: Send a second message to the network-side device, the second message being used to indicate relevant information about the occupied APU.

33. The apparatus of any one of claims 30-32, wherein, It also includes a second sending module, used for: Based on the first information, and assuming APU usage within the first time window or on the first resource: Send the number of APUs occupied to the network-side devices; Alternatively, send the number of APUs occupied within the first time window or on the first resource to the network-side device.

34. The apparatus of any one of claims 30-33, wherein, The first processing module is further configured to: Based on the first information, and assuming APU usage within the first time window or on the first resource: In the case of uplink or downlink transmission, it does not occupy the APU or occupies one APU; Alternatively, in time units outside the first time window, the APU may not be used or one APU may be used for uplink or downlink transmission. Alternatively, no uplink or downlink transmission may be performed within the first time window or on the first resource.

35. The apparatus of any one of claims 30-34, wherein, It also includes a third sending module, used for: Send capability information to network-side devices; The capability information includes at least one of the following: Maximum number of APUs running; The maximum number of APUs that can run simultaneously; The processing time required for one or more AI models that utilize the largest number of APUs; The processing time required for one or more AI models occupying the third number of APUs; Maximum number of APUs to store; Information related to the stored AI model.

36. An AI processing unit (APU) management device, comprising a fourth transmitting module or a first receiving module, wherein: The fourth sending module is used to send first information to the terminal, the first information being used to indicate relevant information of the first time window or relevant information of the first resource; The first receiving module is configured to receive second information from the terminal, the second information being used to indicate relevant information about the artificial intelligence processing unit (APU) occupied by the terminal.

37. The apparatus of claim 36, wherein, The first receiving module is further configured to: Receive capability information from the terminal; The capability information includes at least one of the following: Maximum number of APUs running; The maximum number of APUs that can run simultaneously; The processing time required for one or more AI models that utilize the largest number of APUs; The processing time required for one or more AI models occupying the third number of APUs; Maximum number of APUs to store; Information related to the stored AI model.

38. A terminal comprising a processor and a memory, the memory storing a program or instructions executable on the processor, the program or instructions, when executed by the processor, implementing the steps of the AI ​​processing unit (APU) management method as claimed in any one of claims 1 to 19.

39. A network-side device, comprising a processor and a memory, the memory storing a program or instructions executable on the processor, the program or instructions, when executed by the processor, implementing the steps of the AI ​​processing unit (APU) management method as claimed in any one of claims 20 to 29.

40. A readable storage medium storing a program or instructions that, when executed by a processor, implement the steps of the AI ​​processing unit (APU) management method as claimed in any one of claims 1 to 19, or implement the steps of the AI ​​processing unit (APU) management method as claimed in any one of claims 20 to 29.