WO2024000533A1 - 人工智能应用管理方法、装置及通信设备 - Google Patents

人工智能应用管理方法、装置及通信设备 Download PDF

Info

Publication number
WO2024000533A1
WO2024000533A1 PCT/CN2022/103169 CN2022103169W WO2024000533A1 WO 2024000533 A1 WO2024000533 A1 WO 2024000533A1 CN 2022103169 W CN2022103169 W CN 2022103169W WO 2024000533 A1 WO2024000533 A1 WO 2024000533A1
Authority
WO
WIPO (PCT)
Prior art keywords
processing
task
tasks
processing tasks
priority
Prior art date
Application number
PCT/CN2022/103169
Other languages
English (en)
French (fr)
Inventor
牟勤
Original Assignee
北京小米移动软件有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 北京小米移动软件有限公司 filed Critical 北京小米移动软件有限公司
Priority to PCT/CN2022/103169 priority Critical patent/WO2024000533A1/zh
Priority to CN202280002428.2A priority patent/CN117642720A/zh
Publication of WO2024000533A1 publication Critical patent/WO2024000533A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W72/00—Local resource management
    • H04W72/04—Wireless resource allocation

Definitions

  • the present disclosure relates to the field of mobile communication technology, and in particular to an artificial intelligence application management method, device and communication equipment.
  • AI Artificial Intelligence
  • the present disclosure proposes an artificial intelligence application management method, device and communication equipment, provides a solution for coordinating artificial intelligence processing tasks, optimizes the scheduling of each task in the communication process, and avoids problems caused by processing tasks exceeding the terminal processing capacity. Low efficiency or even stuck problems.
  • a first aspect embodiment of the present disclosure provides an artificial intelligence application management method, which is executed by a network device.
  • the method includes: receiving artificial intelligence AI processing capabilities sent by user equipment; and according to the AI processing capabilities and the AI to be configured
  • the relationship between processing tasks and processing capacity requirements determines the configuration rules for configuring AI processing tasks to user devices.
  • determining the configuration rules for configuring AI processing tasks to user equipment includes: when there are multiple AI processing tasks to be configured, and the sum of the processing capacity requirements of the multiple AI processing tasks to be configured is less than or When equal to the AI processing capability, configure multiple AI processing tasks to be configured to the user device.
  • determining the configuration rules for configuring AI processing tasks to user equipment includes: when there are multiple AI processing tasks to be configured, and the sum of the processing capacity requirements of the multiple AI processing tasks to be configured is greater than the AI When processing capabilities, determine at least one of the occurrence time, delay interval, and priority of the AI processing task; determine the configuration rules based on at least one of the occurrence time, delay interval, and priority of the AI processing task.
  • determining the configuration rule according to at least one of the occurrence time, delay interval and priority of the AI processing task includes: when two or more AIs in the plurality of AI processing tasks to be configured The processing task's demand for processing power meets the simultaneous processing condition, and the configuration rule is determined as follows: configure two or more AI processing tasks to occur at the same time.
  • determining the configuration rule according to at least one of the occurrence time, delay interval and priority of the AI processing task includes: when two or more AIs in the plurality of AI processing tasks to be configured If the processing task's demand for processing power does not meet the simultaneous processing conditions, the configuration rule is determined to be any of the following: Prohibiting the configuration of two or more AI processing tasks to the user device; Configuring the occurrence of two or more AI processing tasks There is a delay interval between times; two or more AI processing tasks are configured to have different priorities.
  • the condition for simultaneous processing is: the sum of the processing power requirements of the two AI processing tasks is less than or equal to the AI processing power.
  • determining the priority of AI processing tasks includes: determining the priority of each AI processing task to be configured according to a network protocol between the network device and the user device; or according to multiple AI processing tasks to be configured. The importance level or urgency level of the processing task, and configure the priority of each AI processing task.
  • the AI processing capabilities include storage capabilities and/or computing capabilities of the user device.
  • a second aspect embodiment of the present disclosure provides an artificial intelligence application management method, which is executed by a user device.
  • the method includes: sending artificial intelligence AI processing capabilities to a network device; and receiving and executing AI processing tasks configured by the network device. .
  • providing the AI processing capability to the network device includes: sending the storage capability and/or computing capability of the user device to the network device.
  • receiving and executing the AI processing task configured by the network device includes: receiving at least one of the occurrence time, delay interval and priority of the AI processing task; according to the occurrence time, delay interval and priority of the AI processing task. At least one of the priorities performs AI processing tasks.
  • executing the AI processing task includes: executing the corresponding AI processing task at the occurrence time; and/or during the AI processing task Execute AI processing tasks after the corresponding delay interval; and/or execute AI processing tasks in descending order of priority.
  • executing AI processing tasks according to priority includes: executing a preset number of AI processing tasks in descending order of priority, and giving up executing the remaining AI processing tasks.
  • a third embodiment of the present disclosure provides an artificial intelligence application management device, which is arranged on a network device.
  • the device includes: a receiving unit for receiving artificial intelligence AI processing capabilities sent by user equipment; and a configuration unit for Based on the relationship between the AI processing capabilities and the processing capability requirements of the AI processing tasks to be configured, the configuration rules for configuring the AI processing tasks to the user equipment are determined.
  • a fourth embodiment of the present disclosure provides an artificial intelligence application management device, which is applied to user equipment.
  • the device includes: a transceiver unit, configured to send artificial intelligence AI processing capabilities to network equipment, and receive network equipment configuration information. AI processing tasks; execution units, used to perform AI processing tasks.
  • a fifth aspect embodiment of the present disclosure provides a communication device.
  • the communication device includes: a transceiver; a memory; and a processor, respectively connected to the transceiver and the memory, and configured to control the transceiver by executing computer-executable instructions on the memory.
  • wireless signal transceiver and can implement the method as in the embodiment of the first aspect or the embodiment of the second aspect of the present disclosure.
  • a sixth embodiment of the present disclosure provides a computer storage medium, wherein the computer storage medium stores computer-executable instructions; after the computer-executable instructions are executed by a processor, the computer-executable instructions can implement the first embodiment or the third aspect of the present disclosure.
  • the network device can receive the AI processing capability sent by the user device, and determine the configuration to the user device based on the relationship between the AI processing capability and the processing capability requirements of the AI processing task to be configured.
  • the configuration rules of AI processing tasks optimize the scheduling of each task in the communication process. This AI processing task coordination scheme avoids inefficiency or even interruption problems caused by processing tasks exceeding the terminal's processing capabilities.
  • Figure 1 is a schematic flowchart of an artificial intelligence application management method according to an embodiment of the present disclosure
  • Figure 2 is a schematic flowchart of an artificial intelligence application management method according to an embodiment of the present disclosure
  • Figure 3 is a schematic flowchart of an artificial intelligence application management method according to an embodiment of the present disclosure
  • Figure 4 is a schematic flowchart of an artificial intelligence application management method according to an embodiment of the present disclosure
  • Figure 5 is a schematic flowchart of an artificial intelligence application management method according to an embodiment of the present disclosure
  • Figure 6 is a schematic diagram of signaling interaction of an artificial intelligence application management method according to an embodiment of the present disclosure
  • Figure 7 is a schematic block diagram of an artificial intelligence application management device according to an embodiment of the present disclosure.
  • Figure 8 is a schematic block diagram of an artificial intelligence application management device according to an embodiment of the present disclosure.
  • Figure 9 is a schematic block diagram of an artificial intelligence application management device according to an embodiment of the present disclosure.
  • Figure 10 is a schematic block diagram of an artificial intelligence application management device according to an embodiment of the present disclosure.
  • Figure 11 is a schematic structural diagram of a communication device according to an embodiment of the present disclosure.
  • Figure 12 is a schematic structural diagram of a chip provided by an embodiment of the present disclosure.
  • RAN1 in the Third Generation Partnership Project (3GPP) R18 standard protocol has established a research project on artificial intelligence technology in the wireless air interface.
  • This project aims to study how to introduce artificial intelligence technology into wireless air interfaces, and at the same time explore how artificial intelligence technology can assist in improving the transmission technology of wireless air interfaces.
  • AI-based CSI enhancement In wireless AI research, application cases of artificial intelligence in the communication field include, for example: AI-based CSI enhancement, AI-based beam management, AI-based positioning, etc.
  • AI operations There are two important stages involved in AI operations.
  • the first stage is the training stage of the model, that is, the stage of obtaining the model.
  • the second stage is the deployment stage of the model, that is, the inference application stage of the model.
  • the training is basically a model that has been trained in advance.
  • the terminal side will generally participate in inference.
  • the terminal is required to have certain AI processing capabilities, such as storage capacity and/or computing power.
  • the present disclosure proposes an artificial intelligence application management method, device and communication equipment, provides a solution for coordinating artificial intelligence processing tasks, optimizes the scheduling of each task in the communication process, and avoids processing tasks exceeding the terminal processing capacity. This can lead to inefficiency or even stuck problems.
  • Figure 1 shows a schematic flowchart of an artificial intelligence application management method according to an embodiment of the present disclosure. As shown in Figure 1, this method is performed by a network device.
  • the network device is a base station, for example, in a 5G communication scenario, it is a gNB (next Generation Node B).
  • the method may include the following steps.
  • user equipment includes but is not limited to smart terminal devices, cellular phones, wireless devices, handsets, mobile units, vehicles, vehicle-mounted equipment, etc., which are not limited in the present disclosure.
  • the AI processing capabilities sent by the UE include storage capabilities and/or computing capabilities.
  • the storage capacity is used to store the model and intermediate parameters generated during inference
  • the computing capacity is used to calculate the results of inference.
  • the processing capability may also include other parameters used to measure the terminal processing capability, etc., which are not limited here.
  • S102 Determine the configuration rules for configuring the AI processing task to the user device based on the relationship between the AI processing capability and the processing capability requirements of the AI processing task to be configured.
  • the network device will receive the AI processing capabilities reported by the terminal and be able to determine the occupation of the terminal's processing capabilities by each processing task.
  • the network device will determine the configuration rules for configuring AI processing tasks to the user device based on the relationship between the upper limit of the terminal's processing capabilities and the processing tasks. In other words, the network device can determine whether to configure a certain AI processing task for the terminal or determine to configure the AI processing task. time of occurrence, or determine that a certain processing task will be delayed.
  • the network device can receive the AI processing capability sent by the user device, and based on the relationship between the AI processing capability and the processing capability requirements of the AI processing task to be configured , determine the configuration rules for configuring AI processing tasks to user equipment, thereby optimizing the scheduling of each task during the communication process.
  • This coordination solution for AI processing tasks avoids inefficiency or even interruption problems caused by processing tasks exceeding the terminal's processing capabilities.
  • Figure 2 shows a schematic flowchart of an artificial intelligence application management method according to an embodiment of the present disclosure. The method is executed by the network device. Based on the embodiment shown in Figure 1, as shown in Figure 2, the method may include the following steps.
  • S201 Receive the artificial intelligence AI processing capability sent by the user device.
  • the above-mentioned step S201 has the same principle as the step S101 in the embodiment shown in FIG. 1. Reference may be made to the relevant description of the above-mentioned step S101, which will not be described again here.
  • the network device can determine the configuration rule for configuring the AI processing task to the user device based on the relationship between the AI processing capability and the processing capability requirements of the AI processing task to be configured. For example, if the network device wants to configure multiple AI processing tasks for the terminal, it needs to determine whether the processing capabilities of the user device (for example, the terminal) required by the multiple AI processing tasks meet the processing capabilities, that is, multiple AI processing tasks to be configured When the sum of the processing capacity requirements of AI processing tasks is less than or equal to the AI processing capacity reported by the UE, the network device determines that the configuration rule for configuring AI processing tasks to the user equipment is: configure multiple AI processing tasks for the UE.
  • the network device can directly configure the UE to execute multiple AI processing tasks without specifying the number of each AI processing task.
  • the time of occurrence or other parameters are not described in this disclosure.
  • AI processing tasks include but are not limited to AI-based CSI enhancement, AI-based beam management, AI-based positioning, etc., which are not limited in this disclosure.
  • the total processing power requirements for AI processing tasks configured by the network for terminals should not be greater than the AI processing capabilities reported by the terminals.
  • the processing capabilities include storage capabilities, computing capabilities, or other measures used to measure terminal processing. Capability parameters, etc.
  • AI processing capability as storage capability as an example, the total storage size of the terminal for AI tasks is C0, that is, the AI processing capability reported by the UE is C0.
  • the network device is configured to perform AI-based CSI compression tasks
  • the storage requirements are C1
  • the storage requirement of the AI-based beam management task is C2 if C1+C2 ⁇ C0, the AI-based CSI compression task and the AI-based beam management task can be configured for the UE.
  • the network device determines whether the processing capability requirements of multiple AI processing tasks on the user equipment meet the processing capabilities, if the processing capability requirements of the multiple AI processing tasks to be configured are If the sum is greater than the AI processing capability reported by the UE, the network equipment has a variety of configuration methods, which will be discussed on a case-by-case basis below.
  • the network device can directly determine the configuration rule as: prohibiting configuring multiple AI processing tasks to the UE. .
  • the storage requirement is C1
  • the AI-based beam management task has a storage requirement of C2
  • the AI-based positioning task has a storage requirement of C3, if C1+C2 +C3>C0, it can be determined that the configuration rule is: it is prohibited to configure AI-based CSI compression tasks, AI-based beam management tasks, and AI-based positioning tasks to the UE.
  • the network device can make a secondary judgment, that is, further judge the multiple AI processing tasks The relationship between the processing power requirements of two or more AI processing tasks and the AI processing capabilities, and determine at least one of the occurrence time, delay interval and priority of the AI processing tasks to determine the configuration rules accordingly .
  • S204 Determine configuration rules based on at least one of the occurrence time, delay interval and priority of the AI processing task.
  • this step specifically includes: when the processing capacity requirements of two or more AI processing tasks among the multiple AI processing tasks to be configured meet the simultaneous processing conditions, the occurrence time of the two AI processing tasks is configured to be the same.
  • the condition for simultaneous processing is: the sum of the processing power requirements of the two AI processing tasks is less than or equal to the AI processing power.
  • the total processing capacity requirements of the terminal AI processing tasks configured by the network can be greater than the processing capacity reported by the terminal, but the AI tasks processed must meet the preset relationship. For example, the AI tasks processed at the same time cannot be greater than the processing capacity reported by the terminal.
  • the storage requirement is C1
  • the AI-based beam management task has a storage requirement of C2
  • the AI-based positioning task has a storage requirement of C2.
  • it is C3, if C1+C2+C3>C0, but C1+C2 ⁇ C0, it can be determined that the configuration rule is: configure the AI-based CSI compression task and the AI-based beam management task to the UE.
  • this step may also include: when the processing capacity requirements of two or more AI processing tasks among the multiple AI processing tasks to be configured do not meet the simultaneous processing conditions, configure the rules Determined to be any of the following: Prohibiting the configuration of two or more AI processing tasks to user equipment; configuring a delay interval between the occurrence times of two or more AI processing tasks; configuring two or more AI Processing tasks have different priorities.
  • the storage requirement is C1
  • the AI-based beam management task has a storage requirement of C2
  • the AI-based positioning task has a storage requirement of C2.
  • it is C3, if C1+C2+C3>C0, and C1+C2>C0, it can be determined that the configuration rule is: it is prohibited to configure the AI-based CSI compression task and the AI-based beam management task to the UE.
  • the network device may also determine the configuration rule to configure a delay interval between the occurrence times of two or more AI processing tasks.
  • the network configures the total processing capacity requirements for terminal AI processing tasks to be greater than the processing capacity reported by the terminal, but the AI tasks processed need to meet the preset relationship. For example, the AI processing tasks need to be separated by X time units.
  • the configuration rule can be determined as follows: the occurrence time between configuring the AI-based CSI compression task and the AI-based beam management task to the UE has a delay interval.
  • the AI-based CSI compression task and the AI-based beam management task cannot be processed at the same time, but need to be separated by X time units.
  • the terminal can be configured to process the CSI compression task in the time period t0 to t1 and the beam management task in the time period t2 to t3.
  • the delay interval corresponding to a certain AI processing task may be the interval between the AI processing task and the previous AI processing task, or the delay interval between the AI processing task and the first executed AI processing task. spacing, which is not limited in this disclosure.
  • this delay interval should be larger than the processing time of processing AI tasks. That is, the delay interval between the AI processing task processed first and the AI processing task processed later should be greater than the processing time of the AI processing task processed first.
  • the configuration rule can be further determined as: configuring the AI-based CSI compression task and the AI-based beam to the UE Manage tasks and configure delay intervals between AI-based positioning tasks, AI-based CSI compression tasks, and AI-based beam management tasks.
  • the terminal can be configured to process AI-based CSI compression tasks and beam management tasks in the t0 to t1 time period, and to process AI-based positioning tasks in the t2 to t3 interval.
  • the delay interval corresponding to the AI-based positioning task can be understood as the length of the time period between t0-t2.
  • the network device can configure the terminal to process the AI-based CSI compression task in the time period t0 to t1, the beam management task in the time period t2 to t3, and the AI-based positioning task in the time period t4 to t5.
  • the delay interval corresponding to the AI-based positioning task can be understood as the length of the time period between t0 and t4, or the length of the time period between t2 and t4, which is not limited in this disclosure.
  • the delay interval in this disclosure can be a preset value or a network configuration value. This disclosure does not limit the value of the above time unit or delay interval, which depends on the specific situation. It will be appreciated that the delay interval can be measured in units of time. The value of the time unit is not limited in this disclosure.
  • the delay interval can be determined according to the priority of each AI processing task, wherein, in descending order of priority, the delay interval between the previous AI processing task and the subsequent AI processing task is greater than the previous one.
  • the processing time of the AI processing task is determined according to the priority of each AI processing task, wherein, in descending order of priority, the delay interval between the previous AI processing task and the subsequent AI processing task is greater than the previous one.
  • the beam management task is processed in the time period t2 to t3, and the AI-based positioning task is processed in the time period t4 to t5.
  • the delay interval corresponding to the AI-based positioning task is the length of the time period between t2 and t4. , then the length of the time period t2-t4 should be greater than or equal to the processing time required to process the beam management task.
  • the AI-based CSI compression task and beam management task are processed in the t0 ⁇ t1 time period, and the AI-based positioning task is processed in the t2 ⁇ t3 period.
  • the length of the t0-t2 time period should be greater than or equal to the AI-based CSI compression task and beam management task. One of the longest administrative tasks.
  • the network device may not configure the order of executing each AI processing task to the UE, but only configure the time interval between each AI processing task to ensure that multiple AUs may exceed the UE's processing capabilities.
  • the processing tasks can be executed at different times.
  • the network device can also determine the configuration rules to configure two or more AI processing tasks with different priorities.
  • the network configures the total processing capacity requirements for terminal AI processing tasks to be greater than the processing capacity reported by the terminal, but the AI tasks processed need to meet the preset relationship. For example, if the execution priorities of AI processing tasks are different, they can be processed according to the priority. Each AI processing task is executed sequentially.
  • the configuration rule can be determined as follows: configuring the UE with an AI-based CSI compression task and an AI-based beam management task with different priorities.
  • AI-based CSI compression tasks and AI-based beam management tasks cannot be processed at the same time, but need to be processed sequentially in a certain priority order.
  • the priority of the task can be configured as CSI>Beam Management>Positioning, and the terminal can perform the AI-based CSI compression task first and then the AI-based beam management task according to the rules.
  • the priority of each AI task can be configured as CSI>Beam Management>Positioning, then the terminal can perform the AI-based CSI compression task first , then perform the AI-based beam management task, and then perform the AI-based positioning task according to the rules, or only perform the first two tasks and abandon the third task with a lower priority.
  • the network device can preset priorities at this time.
  • the terminal processes AI tasks according to the processing priority or abandons low-priority task processing.
  • the method for configuring the priority of the network device may include: determining the priority of each AI processing task to be configured according to the network protocol between the network device and the user device; or according to the importance of multiple AI processing tasks to be configured. Level or emergency level, configure the priority of each AI processing task.
  • the priority may also be determined in other ways, which are not limited in this disclosure.
  • delay interval between two or more AI processing tasks of different priorities.
  • the delay interval can be explained with reference to the above embodiments.
  • the network device in addition to configuring the time interval between each AI processing task, can also configure the order of executing each AI processing task to the UE to ensure that it may exceed Multiple AU processing tasks of UE processing capabilities are not executed at the same time and are executed in a certain priority order.
  • the configuration rules can be different, that is, multiple configuration rules can be applied at the same time.
  • the first and second tasks can be executed at the same time, and there is a delay interval between them and the third task. Then the third task will be executed after a certain period of time after the first and second tasks are executed. If the priority of the first task is higher than the fourth task, the third task will be executed first and then the fourth task.
  • the network device can determine different configuration rules for configuring AI processing tasks to the user device based on the relationship between the AI processing capabilities and the processing capability requirements of the AI processing tasks to be configured. , such as allowing or disabling or executing at a certain delay interval or executing at a certain priority, etc., fully considers the scheduling and configuration of AI applications in various situations, optimizes the scheduling of each task during the communication process, and avoids processing tasks exceeding the terminal Inefficiency and even interruption problems caused by processing power.
  • FIG. 3 shows a schematic flowchart of an artificial intelligence application management method according to an embodiment of the present disclosure.
  • the method is executed by user equipment.
  • user equipment User Equipment, UE
  • user equipment includes but is not limited to intelligent terminal equipment, cellular phones, wireless devices, handsets, mobile units, vehicles, vehicle-mounted equipment, etc., in No limitations are set forth in this disclosure.
  • the method may include the following steps.
  • S301 Send artificial intelligence AI processing capabilities to the network device.
  • the AI processing capabilities sent by the UE to the network device include storage capabilities and/or computing capabilities.
  • the storage capacity is used to store the model and intermediate parameters generated during inference
  • the computing capacity is used to calculate the results of inference.
  • the processing capability may also include other parameters used to measure the terminal processing capability, etc., which are not limited here.
  • S302 Receive and execute the AI processing task configured by the network device.
  • the UE may receive from the network device the configuration rules for configuring AI processing tasks to the user equipment determined by the network device based on the relationship between the terminal processing capability upper limit and the processing tasks.
  • the network device may determine whether to If the terminal configures a certain AI processing task or determines the occurrence time of the configured AI processing task, or determines that a certain processing task will be delayed, the UE can execute it according to the configuration of the network device.
  • the UE can send artificial intelligence AI processing capabilities to the network device, receive and execute the AI processing tasks configured by the network device, thereby executing various AI processing tasks according to the configuration of the network device, and optimizing By eliminating the scheduling of various tasks during the communication process, this AI processing task coordination solution avoids inefficiency or even interruption problems caused by processing tasks exceeding the terminal's processing capabilities.
  • Figure 4 is a schematic flowchart of an artificial intelligence application management method according to an embodiment of the present disclosure. The method is applied to the UE. Based on the embodiment shown in Figure 3, as shown in Figure 4, the method may include the following steps.
  • S401 Send artificial intelligence AI processing capabilities to the network device.
  • this step includes: sending the storage capability and/or computing capability of the user equipment to the network device.
  • step S401 is the same as that of step S301 in the embodiment shown in FIG. 3. Please refer to the relevant description of S403, which will not be described again here.
  • S402 Receive at least one of the occurrence time, delay interval and priority of the AI processing task.
  • the network device determines that the configuration rule is to prohibit configuring AI processing tasks to the UE, the UE will not receive the relevant AI tasks. Therefore, for the UE On the other hand, this situation is outside the scope of discussion.
  • the network device determines to configure an AI processing task to the UE, the configured AI processing task must meet the judgment conditions of the network side, and the UE can, according to the configuration of the network side, Receive specific configuration rules and the AI processing tasks to be performed to perform one or more AI processing tasks. Among them, the UE can receive the configuration rules determined by the network device, such as the occurrence time, delay interval or priority of executing the AI processing task.
  • S403 Execute the AI processing task according to at least one of the occurrence time, delay interval and priority of the AI processing task.
  • this step specifically includes: executing the corresponding AI processing task at the occurrence time; and/or executing the AI processing task after the delay interval corresponding to the AI processing task; and/or executing the AI in descending order of priority. Process tasks.
  • the delay interval corresponding to a certain AI processing task may be the interval between the AI processing task and the previous AI processing task, or the delay interval between the AI processing task and the first executed AI processing task. spacing, which is not limited in this disclosure.
  • this delay interval should be larger than the processing time of processing AI tasks. That is, the delay interval between the AI processing task processed first and the AI processing task processed later should be greater than the processing time of the AI processing task processed first.
  • the AI processing capabilities reported by the UE are used to assist the network device in determining rules for configuring AI processing tasks for the UE.
  • the network device can determine the configuration rules for configuring the AI processing task to the user device based on the relationship between the AI processing capability and the processing capability requirements of the AI processing task to be configured.
  • the network device determines that the configuration rule for configuring AI processing tasks to the user equipment is: configure multiple AI processing tasks for the UE. Task.
  • the UE can receive and execute these AI processing tasks configured for it by the network device.
  • the network device does not need to configure other scheduling rules for the terminal.
  • the total storage size of the terminal for AI tasks is C0, that is, the AI processing capability reported by the UE is C0.
  • the storage requirements are C1
  • the storage requirement for the AI-based beam management task is C2
  • the requirement for the AI-based positioning task is C3, if C1+C2+C3 ⁇ C0, these three tasks can be configured for the UE.
  • the network device can directly configure the UE to execute multiple AI processing tasks without specifying each AI processing task. The occurrence time of the task or other parameters, so the UE can directly perform multiple AI processing tasks.
  • the network device determines that the configuration rules for configuring AI processing tasks to the user equipment need to be discussed on a case-by-case basis.
  • the UE can be configured with an AI-based CSI compression task and an AI-based beam management task.
  • the UE can receive that the AI processing tasks configured by the network device are AI-based CSI compression tasks and AI-based beam management tasks, and can perform these two tasks at the same time. This situation is similar to the previous example.
  • the network device prohibits configuring the task to the UE; 2) the network device configures the delay interval or priority for executing the task.
  • the UE will not receive information related to performing AI-based positioning tasks.
  • the UE may receive the delay interval or priority sent by the network device to perform the task.
  • the network device is configured to perform the AI-based CSI compression task and the AI-based beam management task first, and then perform the AI-based positioning task after an interval of CSI compression tasks and AI-based beam management tasks, the UE performs them according to this priority.
  • the UE can process the AI-based CSI compression task in the t0-t1 time period and the AI-based positioning task in the t2-t3 period according to the configuration of the network device.
  • the network device can also inform the UE which AI processing task interval the time interval is relative to, and the UE can execute it accordingly.
  • the UE can first perform the AI-based CSI compression task, then perform the AI-based beam management task, and then perform the AI-based positioning task in the priority order of CSI>beam management>positioning. This example has been described in detail in the embodiment shown in FIG. 2 and will not be described again here.
  • the UE may execute a preset number of AI processing tasks in descending order of priority and give up executing the remaining AI processing tasks. For example, according to the priority order of CSI>beam management>positioning, only the first two tasks are executed, and the third task with lower priority is abandoned.
  • the network device can only inform the UE of the delay interval required during execution of each AI processing task, without informing the specific execution order, or the network device can also inform the UE to execute each AI processing task in a certain priority order, and the UE Can be performed based on the configuration of the network device.
  • the delay interval or time unit may be a preset value or a network configuration value. This disclosure does not limit the value of the above time unit or delay interval, which depends on the specific situation.
  • the priority in this disclosure may be determined based on the network protocol between the network device and the user equipment, or may be configured based on the network device according to the importance level or emergency level of multiple AI processing tasks to be configured. No limitations are set forth in this disclosure.
  • the UE can report AI processing capabilities to the network device, thereby assisting the network device in determining the AI processing capability to the user based on the relationship between the AI processing capabilities and the processing capability requirements of the AI processing tasks to be configured.
  • the device configures different configuration rules for AI processing tasks.
  • the UE can receive and execute AI processing tasks according to the configuration of the network device. It fully considers the scheduling and configuration of AI applications in various situations, optimizes the scheduling of each task during the communication process, and avoids Inefficiency or even interruption problems caused by processing tasks exceeding the terminal's processing capabilities.
  • network equipment and user equipment may include hardware structures and software modules to implement the above functions in the form of hardware structures, software modules, or hardware structures plus software modules.
  • a certain function among the above functions can be executed by a hardware structure, a software module, or a hardware structure plus a software module.
  • the present disclosure also provides an artificial intelligence application management device.
  • the application management method corresponds, so the implementation of the artificial intelligence application management method is also applicable to the artificial intelligence application management device provided in this embodiment, and will not be described in detail in this embodiment.
  • FIG. 5 is a schematic structural diagram of an artificial intelligence application management device 500 provided by an embodiment of the present disclosure.
  • the artificial intelligence application management device 500 can be used for network equipment.
  • the device 500 may include: a receiving unit 510, configured to receive the artificial intelligence AI processing capability sent by the user equipment; and a configuring unit 520, configured to configure the processing capability according to the AI processing capability and the AI processing task to be configured.
  • the relationship between the requirements determines the configuration rules for configuring AI processing tasks to user devices.
  • the network device can receive the AI processing capability sent by the user equipment, and determine the direction to the AI processing capability based on the relationship between the AI processing capability and the processing capability requirements of the AI processing task to be configured.
  • User equipment configures configuration rules for AI processing tasks, thereby optimizing the scheduling of each task during the communication process.
  • This AI processing task coordination solution avoids inefficiency or even interruption problems caused by processing tasks exceeding the terminal's processing capabilities.
  • the configuration unit 520 is configured to: when there are multiple AI processing tasks to be configured, and the sum of the processing capacity requirements of the multiple AI processing tasks to be configured is less than or equal to the AI processing capacity, configure the configuration to the user device. Multiple AI processing tasks to be configured.
  • the configuration unit 520 is configured to: determine the occurrence of an AI processing task when there are multiple AI processing tasks to be configured, and the sum of the processing capacity requirements of the multiple AI processing tasks to be configured is greater than the AI processing capacity. At least one of time, delay interval and priority; determine the configuration rule according to at least one of the occurrence time, delay interval and priority of the AI processing task.
  • the configuration unit 520 is configured to: when the processing capacity requirements of two or more AI processing tasks among the multiple AI processing tasks to be configured meet simultaneous processing conditions, determine the configuration rule as: configure two One or more AI processing tasks occur at the same time.
  • the configuration unit 520 is also configured to determine the configuration rules as follows when the processing capacity requirements of two or more AI processing tasks among the multiple AI processing tasks to be configured do not meet the simultaneous processing conditions. Any item: It is prohibited to configure two or more AI processing tasks to user equipment; configure the occurrence time of two or more AI processing tasks with a delay interval; configure two or more AI processing tasks with different priorities Level, where the simultaneous processing condition is: the sum of the processing power requirements of two or more AI processing tasks is less than or equal to the AI processing power.
  • the configuration unit 520 is configured to: determine the priority of each AI processing task to be configured according to the network protocol between the network device and the user device; or according to the importance levels of multiple AI processing tasks to be configured; or Emergency level, configure the priority of each AI processing task.
  • AI processing capabilities include storage capabilities and/or computing capabilities of the user device.
  • the network device can determine different configuration rules for configuring AI processing tasks to the user equipment based on the relationship between the AI processing capabilities and the processing capability requirements of the AI processing tasks to be configured. , such as allowing or disabling or executing at a certain delay interval or executing at a certain priority, etc., fully considers the scheduling and configuration of AI applications in various situations, optimizes the scheduling of each task during the communication process, and avoids processing tasks exceeding the terminal Inefficiency and even interruption problems caused by processing power.
  • FIG. 6 is a schematic structural diagram of an artificial intelligence application management device 600 provided by an embodiment of the present disclosure.
  • the artificial intelligence application management device 600 can be used in user equipment.
  • the apparatus 600 may include: a transceiver unit 610, used to send artificial intelligence AI processing capabilities to the network device and receive AI processing tasks configured by the network device; and an execution unit 620, used to execute the AI processing task.
  • the UE can send artificial intelligence AI processing capabilities to network equipment, receive and execute AI processing tasks configured by the network equipment, thereby executing various AI processing tasks according to the configuration of the network equipment, optimizing the communication process.
  • the scheduling of each task in the AI processing task coordination scheme avoids inefficiency or even interruption problems caused by processing tasks exceeding the terminal processing capacity.
  • the transceiving unit 610 is configured to: send the storage capability and/or computing capability of the user equipment to the network device.
  • the transceiving unit 610 is configured to receive at least one of an occurrence time, a delay interval, and a priority of the AI processing task.
  • the execution unit 620 is configured to: execute the AI processing task according to at least one of an occurrence time, a delay interval, and a priority of the AI processing task.
  • the execution unit 620 is configured to: execute the corresponding AI processing task at the occurrence time; and/or execute the AI processing task after the execution delay interval of the previous AI processing task; and/or execute the AI in descending order of priority. Process tasks.
  • the execution unit 620 is configured to: execute a preset number of AI processing tasks in descending priority order, and give up executing the remaining AI processing tasks.
  • the UE can report AI processing capabilities to the network device, thereby assisting the network device in determining to configure AI to the user device based on the relationship between the AI processing capabilities and the processing capability requirements of the AI processing tasks to be configured. Different configuration rules for processing tasks.
  • the UE can receive and execute AI processing tasks according to the configuration of the network device. It fully considers the scheduling and configuration of AI applications in various situations, optimizes the scheduling of each task during the communication process, and avoids processing tasks exceeding Inefficiency and even interruption problems caused by terminal processing capabilities.
  • Figure 11 is a schematic structural diagram of a communication device 700 provided by an embodiment of the present application.
  • the communication device 700 may be a network device, a user equipment, a chip, a chip system, or a processor that supports network equipment to implement the above method, or a chip, a chip system, or a processor that supports user equipment to implement the above method. Processor etc.
  • the device can be used to implement the method described in the above method embodiment. For details, please refer to the description in the above method embodiment.
  • Communication device 700 may include one or more processors 701.
  • the processor 701 may be a general-purpose processor or a special-purpose processor, or the like.
  • it can be a baseband processor or a central processing unit.
  • the baseband processor can be used to process communication protocols and communication data.
  • the central processor can be used to control communication devices (such as base stations, baseband chips, terminal equipment, terminal equipment chips, DU or CU, etc.) and execute computer programs. , processing data for computer programs.
  • the communication device 700 may also include one or more memories 702, on which a computer program 704 may be stored.
  • the processor 701 executes the computer program 704, so that the communication device 700 executes the method described in the above method embodiment.
  • the memory 702 may also store data.
  • the communication device 700 and the memory 702 can be provided separately or integrated together.
  • the communication device 700 may also include a transceiver 705 and an antenna 706.
  • the transceiver 705 may be called a transceiver unit, a transceiver, a transceiver circuit, etc., and is used to implement transceiver functions.
  • the transceiver 705 may include a receiver and a transmitter.
  • the receiver may be called a receiver or a receiving circuit, etc., used to implement the receiving function;
  • the transmitter may be called a transmitter, a transmitting circuit, etc., used to implement the transmitting function.
  • the communication device 700 may also include one or more interface circuits 707.
  • the interface circuit 707 is used to receive code instructions and transmit them to the processor 701 .
  • the processor 701 executes code instructions to cause the communication device 700 to perform the method described in the above method embodiment.
  • the processor 701 may include a transceiver for implementing receiving and transmitting functions.
  • the transceiver may be a transceiver circuit, an interface, or an interface circuit.
  • the transceiver circuits, interfaces or interface circuits used to implement the receiving and transmitting functions can be separate or integrated together.
  • the above-mentioned transceiver circuit, interface or interface circuit can be used for reading and writing codes/data, or the above-mentioned transceiver circuit, interface or interface circuit can be used for signal transmission or transfer.
  • the processor 701 may store a computer program 703, and the computer program 703 runs on the processor 701, causing the communication device 700 to perform the method described in the above method embodiment.
  • the computer program 703 may be solidified in the processor 701, in which case the processor 701 may be implemented by hardware.
  • the communication device 700 may include a circuit, which may implement the functions of sending or receiving or communicating in the foregoing method embodiments.
  • the processor and transceiver described in this application can be implemented in integrated circuits (ICs), analog ICs, radio frequency integrated circuits RFICs, mixed signal ICs, application specific integrated circuits (ASICs), printed circuit boards ( printed circuit board (PCB), electronic equipment, etc.
  • the processor and transceiver can also be manufactured using various IC process technologies, such as complementary metal oxide semiconductor (CMOS), n-type metal oxide-semiconductor (NMOS), P-type Metal oxide semiconductor (positive channel metal oxide semiconductor, PMOS), bipolar junction transistor (BJT), bipolar CMOS (BiCMOS), silicon germanium (SiGe), gallium arsenide (GaAs), etc.
  • CMOS complementary metal oxide semiconductor
  • NMOS n-type metal oxide-semiconductor
  • PMOS P-type Metal oxide semiconductor
  • BJT bipolar junction transistor
  • BiCMOS bipolar CMOS
  • SiGe silicon germanium
  • GaAs gallium arsenide
  • the communication device described in the above embodiments may be a network device or user equipment, but the scope of the communication device described in this application is not limited thereto, and the structure of the communication device may not be limited by FIG. 10 .
  • the communication device may be a stand-alone device or may be part of a larger device.
  • the communication device can be:
  • the IC collection may also include storage components for storing data and computer programs;
  • the communication device may be a chip or a chip system
  • the communication device may be a chip or a chip system
  • the chip shown in Figure 8 includes a processor 801 and an interface 802.
  • the number of processors 801 may be one or more, and the number of interfaces 802 may be multiple.
  • the chip also includes a memory 803, which is used to store necessary computer programs and data.
  • This application also provides a readable storage medium on which instructions are stored. When the instructions are executed by a computer, the functions of any of the above method embodiments are implemented.
  • This application also provides a computer program product, which, when executed by a computer, implements the functions of any of the above method embodiments.
  • a computer program product includes one or more computer programs.
  • the computer may be a general purpose computer, a special purpose computer, a computer network, or other programmable device.
  • the computer program may be stored in or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer program may be transmitted from a website, computer, server or data center via a wireline (e.g.
  • Coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless means to transmit to another website, computer, server or data center.
  • Computer-readable storage media can be any available media that can be accessed by a computer or a data storage device such as a server, data center, or other integrated media that contains one or more available media. Available media may be magnetic media (e.g., floppy disks, hard disks, tapes), optical media (e.g., high-density digital video discs (DVD)), or semiconductor media (e.g., solid state disks (SSD)) )wait.
  • magnetic media e.g., floppy disks, hard disks, tapes
  • optical media e.g., high-density digital video discs (DVD)
  • semiconductor media e.g., solid state disks (SSD)
  • At least one in this application can also be described as one or more, and the plurality can be two, three, four or more, which is not limited by this application.
  • the technical feature is distinguished by “first”, “second”, “third”, “A”, “B”, “C” and “D”, etc.
  • the technical features described in “first”, “second”, “third”, “A”, “B”, “C” and “D” are in no particular order or order.
  • machine-readable medium and “computer-readable medium” refer to any computer program product, apparatus, and/or means for providing machine instructions and/or data to a programmable processor (for example, magnetic disks, optical disks, memories, programmable logic devices (PLD)), including machine-readable media that receive machine instructions as machine-readable signals.
  • machine-readable signal refers to any signal used to provide machine instructions and/or data to a programmable processor.
  • the systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., A user's computer having a graphical user interface or web browser through which the user can interact with implementations of the systems and technologies described herein), or including such backend components, middleware components, or any combination of front-end components in a computing system.
  • the components of the system may be interconnected by any form or medium of digital data communication (eg, a communications network). Examples of communication networks include: local area network (LAN), wide area network (WAN), and the Internet.
  • Computer systems may include clients and servers.
  • Clients and servers are generally remote from each other and typically interact over a communications network.
  • the relationship of client and server is created by computer programs running on corresponding computers and having a client-server relationship with each other.

Landscapes

  • Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

一种人工智能应用管理方法、装置及通信设备，涉及移动通信技术领域。网络设备能够接收用户设备发送的AI处理能力，并根据该AI处理能力以及待配置的AI处理任务对处理能力的需求之间的关系，确定向用户设备配置AI处理任务的配置规则，从而优化了通信过程中各个任务的调度，该AI处理任务的协调方案避免了处理任务超过终端处理能力而带来的低效率甚至卡断问题。

Description

人工智能应用管理方法、装置及通信设备 技术领域
本公开涉及移动通信技术领域，特别涉及一种人工智能应用管理方法、装置及通信设备。
背景技术
随着移动网络通信技术的不断演进，各应用场景对于网络通信效率的需求越来越高。人工智能(Artificial Intelligence，AI)技术在通信领域取得不断突破，为用户带来丰富的应用体验。然而，由于终端的处理能力有限，对于协调AI处理任务目前尚没有很好的解决方案。
发明内容
本公开提出了一种人工智能应用管理方法、装置及通信设备，提供了一种协调人工智能处理任务的方案，优化了通信过程中各个任务的调度，避免处理任务超过终端处理能力而带来的低效率甚至卡断问题。
本公开的第一方面实施例提供了一种人工智能应用管理方法，该方法由网络设备执行，该方法包括：接收用户设备发送的人工智能AI处理能力；以及根据AI处理能力以及待配置的AI处理任务对处理能力的需求之间的关系，确定向用户设备配置AI处理任务的配置规则。
在本公开的一些实施例中，确定向用户设备配置AI处理任务的配置规则包括：当待配置的AI处理任务为多个，且多个待配置的AI处理任务对处理能力的需求总和小于或等于AI处理能力时，向用户设备配置多个待配置的AI处理任务。
在本公开的一些实施例中，确定向用户设备配置AI处理任务的配置规则包括：当待配置的AI处理任务为多个，且多个待配置的AI处理任务对处理能力的需求总和大于AI处理能力时，确定AI处理任务的发生时间、延迟间隔和优先级中的至少一个；根据AI处理任务的发生时间、延迟间隔和优先级中的至少一个，确定配置规则。
在本公开的一些实施例中，根据AI处理任务的发生时间、延迟间隔和优先级中的至少一个，确定配置规则包括：当多个待配置的AI处理任务中的两个或两个以上AI处理任务对处理能力的需求满足同时处理条件，将配置规则确定为：配置两个或两个以上AI处理任务的发生时间相同。
在本公开的一些实施例中，根据AI处理任务的发生时间、延迟间隔和优先级中的至少一个，确定配置规则包括：当多个待配置的AI处理任务中的两个或两个以上AI处理任务对处理能力的需求不满足同时处理条件，将配置规则确定为以下任一项：禁止向用户设备配置两个或两个以上AI处理任务；配置两个或两个以上AI处理任务的发生时间之间具有延迟间隔；配置两个或两个以上AI处理任务具有不同的优先级，其中，同时处理条件为：两个AI处理任务对处理能力的需求总和小于或等于AI处理能力。
在本公开的一些实施例中，确定AI处理任务的优先级包括：根据网络设备和用户设备之间的网络协议，确定各个待配置的AI处理任务的优先级；或根据多个待配置的AI处理任务的重要等级或紧急等级，对各个AI处理任务配置优先级。
在本公开的一些实施例中，不同优先级的两个或两个以上AI处理任务之间具有延时间隔，按照优先级降序的顺序，前一AI处理任务与后一AI处理任务之间的延时间隔大于前一AI处理任务的处理时间。
在本公开的一些实施例中，AI处理能力包括用户设备的存储能力和/或计算能力。
本公开的第二方面实施例提供了一种人工智能应用管理方法，该方法由用户设备执行，该方法包括：向网络设备发送人工智能AI处理能力；以及接收并执行网络设备配置的AI处理任务。
在本公开的一些实施例中，向网络设备AI处理能力包括：向网络设备发送用户设备的存储能力和/或计算能力。
在本公开的一些实施例中，接收并执行网络设备配置的AI处理任务包括：接收AI处理任务的发生时间、延迟间隔和优先级中的至少一个；根据AI处理任务的发生时间、延迟间隔和优先级中的至少一个，执行AI处理任务。
在本公开的一些实施例中，根据AI处理任务的发生时间、延迟间隔和优先级中的至少一个，执行AI处理任务包括：在发生时间执行对应的AI处理任务；和/或在AI处理任务对应的延迟间隔后执行AI处理任务；和/或以优先级降序顺序执行AI处理任务。
在本公开的一些实施例中，根据优先级，执行AI处理任务包括：以优先级降序顺序，执行预设个数的AI处理任务，放弃执行剩余AI处理任务。
本公开的第三方面实施例提供了一种人工智能应用管理装置，该装置布置于网络设备，该装置包括：接收单元，用于接收用户设备发送的人工智能AI处理能力；以及配置单元，用于根据AI处理能力以及待配置的AI处理任务对处理能力的需求之间的关系，确定向用户设备配置AI处理任务的配置规则。
本公开的第四方面实施例提供了一种人工智能应用管理装置，该装置应用于用户设备，该装置包括：收发单元，用于向网络设备发送人工智能AI处理能力，以及接收网络设备配置的AI处理任务；执行单元，用于执行AI处理任务。
本公开的第五方面实施例提供了一种通信设备，该通信设备包括：收发器；存储器；处理器，分别与收发器及存储器连接，配置为通过执行存储器上的计算机可执行指令，控制收发器的无线信号收发，并能够实现如本公开第一方面实施例或第二方面实施例的方法。
本公开的第六方面实施例提供了一种计算机存储介质，其中，计算机存储介质存储有计算机可执行指令；计算机可执行指令被处理器执行后，能够实现如本公开第一方面实施例或第二方面实施例的方法。
根据本公开的人工智能应用管理方法，网络设备能够接收用户设备发送的AI处理能力，并根据该AI处理能力以及待配置的AI处理任务对处理能力的需求之间的关系，确定向用户设备配置AI处理任务的配置规则，从而优化了通信过程中各个任务的调度，该AI处理任务的协调方案避免了处理任务超过终端处理能力而带来的低效率甚至卡断问题。
本公开附加的方面和优点将在下面的描述中部分给出，部分将从下面的描述中变得明显，或通过本公开的实践了解到。
附图说明
本公开上述的和/或附加的方面和优点从下面结合附图对实施例的描述中将变得明显和容易理解，其中：
图1为根据本公开实施例的一种人工智能应用管理方法的流程示意图；
图2为根据本公开实施例的一种人工智能应用管理方法的流程示意图；
图3为根据本公开实施例的一种人工智能应用管理方法的流程示意图；
图4为根据本公开实施例的一种人工智能应用管理方法的流程示意图；
图5为根据本公开实施例的一种人工智能应用管理方法的流程示意图；
图6为根据本公开实施例的一种人工智能应用管理方法的信令交互示意图；
图7为根据本公开实施例的一种人工智能应用管理装置的示意框图；
图8为根据本公开实施例的一种人工智能应用管理装置的示意框图；
图9为根据本公开实施例的一种人工智能应用管理装置的示意框图；
图10为根据本公开实施例的一种人工智能应用管理装置的示意框图；
图11为根据本公开实施例的一种通信装置的结构示意图；
图12为本公开实施例提供的一种芯片的结构示意图。
具体实施方式
下面详细描述本公开的实施例，实施例的示例在附图中示出，其中自始至终相同或类似的标号表示相同或类似的元件或具有相同或类似功能的元件。下面通过参考附图描述的实施例是示例性的，旨在用于解释本公开，而不能理解为对本公开的限制。
移动网络通信技术在现实生活中应用场景的边界在不断扩展，例如面向未来的增强现实(Augmented Reality，AR)、虚拟现实(Virtual Reality，VR)、以及更多新型互联网应用(诸如车联网、物联网等)等应用场景的涌现，各应用场景对于网络通信效率、速度、时延、带宽等能力的需求越来越高。另一方面，人工智能(Artificial Intelligence，AI)技术在通信领域取得不断突破，为用户带来丰富的应用体验。例如智能语音、计算机视觉等应用涉及教育、交通、家居、医疗、零售、安防等多个领域，给人们生活带来便利同时，也在促进各个行业进行产业升级。
第三代合作伙伴计划(Third Generation Partnership Project，3GPP)R18标准协议中的RAN1设立了关于人工智能技术在无线空口中的研究项目。该项目旨在研究如何在无线空口中引入人工智能技术，同时探讨人工智能技术如何对无线空口的传输技术进行辅助提高。
在无线AI的研究中，人工智能在通信领域的应用案例包括例如：基于AI的CSI增强、基于AI的波束管理、基于AI的定位等。在AI操作中涉及两个重要的阶段，第一个阶段是模型的训练阶段，即获得模型的阶段，第二个阶段是模型的部署阶段，即模型的推理应用阶段，在目前的讨论中AI的训练基本上是提前训练好的模型，终端侧一般会参与到推理中去，而对于终端侧的推理，要求终端具有一定的AI处理能力，例如存储能力和/或计算能力。
然而，由于无线AI的使用场景可以部署AI模型在终端侧，这些AI任务会共享终端的AI处理能力，对于终端来说，终端的AI处理能力是一定的，那么当存在多个AI处理任务时，此时需要协调任务处理，目前对于协调AI处理任务尚没有很好的解决方案。
为此，本公开提出了一种人工智能应用管理方法、装置及通信设备，提供了一种协调人工智能处理任务的方案，优化了通信过程中各个任务的调度，避免处理任务超过终端处理能力而带来的低效率甚至卡断问题。
可以理解的是，本公开提供的方案可以用于第五代移动通信技术(Fifth Generation，5G)及其后续通信技术，诸如第五代移动通信技术演进(5G-advanced)、第六代移动通信技术(SixthGeneration，6G)等，在本公开中不予限制。
下面结合附图对本申请所提供的人工智能应用管理方案进行详细介绍。
图1示出了根据本公开实施例的一种人工智能应用管理方法的流程示意图。如图1所示，该方法由网络设备执行。在本公开的实施例中，网络设备为基站，例如，在5G通信场景中为gNB(next Generation Node B)。
该方法可以包括以下步骤。
S101，接收用户设备发送的人工智能AI处理能力。
在本公开的实施例中，用户设备(User Equipment，UE)包括但不限于智能终端设备、蜂窝电话、无线设备、手持机、移动单元、车辆、车载设备等，在本公开中不予限制。
本公开中，UE发送的AI处理能力包括存储能力和/或计算能力。其中，存储能力用以存储模型以及推理中产生的中间参数，计算能力用以计算出推理的结果。本公开中该处理能力还可以包括其他用于衡量终端处理能力的参数等，在此不予限制。
S102，根据AI处理能力以及待配置的AI处理任务对处理能力的需求之间的关系，确定向用户设备配置AI处理任务的配置规则。
在本公开的实施例中，网络设备将收到终端上报的AI处理能力，并能够确定每个处理任务对终端处理能力的占据情况。网络设备会根据终端处理能力上限与处理任务之间的关系，确定向用户设备配置AI处理任务的配置规则，换言之，网络设备可以判断是否要给终端配置某个AI处理任务或者确定配置AI处理任务的发生时间，或者确定某个处理任务会出现延迟等信息。
综上，根据本公开实施例提供的人工智能应用管理方法，网络设备能够接收用户设备发送的AI处理能力，并根据该AI处理能力以及待配置的AI处理任务对处理能力的需求之间的关系，确定向用户设备配置AI处理任务的配置规则，从而优化了通信过程中各个任务的调度，该AI处理任务的协调方案避免了处理任务超过终端处理能力而带来的低效率甚至卡断问题。
图2示出了根据本公开实施例的一种人工智能应用管理方法的流程示意图。该方法由网络设备执行，基于图1所示实施例，如图2所示，该方法可以包括以下步骤。
S201，接收用户设备发送的人工智能AI处理能力。
在本公开的实施例中，上述步骤S201与图1所示实施例中的步骤S101原理相同，可参照上述步骤S101的相关说明，在此不再赘述。
S202，当待配置的AI处理任务为多个，且多个待配置的AI处理任务对处理能力的需求总和小于或等于AI处理能力时，向用户设备配置多个待配置的AI处理任务。
在本公开的一些实施例中，网络设备能够根据AI处理能力以及待配置的AI处理任务对处理能力的需求之间的关系，确定向用户设备配置AI处理任务的配置规则。例如，如果网络设备想要为终端配 置多个AI处理任务，则需要判断多个AI处理任务对用户设备(例如，终端)的处理能力的需求是否满足处理能力，也即，多个待配置的AI处理任务对处理能力的需求总和小于或等于UE上报的AI处理能力时，则网络设备确定向用户设备配置AI处理任务的配置规则为：为UE配置多个AI处理任务。
在该实施例中，由于多个AI处理任务对UE处理能力的需求并未超过UE上报的处理能力，因此网络设备可以直接向UE配置执行多个AI处理任务，可以不指定各个AI处理任务的发生时间或其他参数，对此本公开中不予描述。
在本公开中，AI处理任务包括但不限于基于AI的CSI增强、基于AI的波束管理、基于AI的定位等，在本公开中不予限制。
例如，为了避免终端处理能力紧缺，网络配置给终端进行的AI处理任务对处理能力的需求总和应当不大于终端上报的AI处理能力，该处理能力包括存储能力、计算能力或其他用于衡量终端处理能力的参数等。以AI处理能力为存储能力为例，终端对AI任务总的存储大小为C0，即，UE上报的AI处理能力为C0，当网络设备配置进行基于AI的CSI压缩任务对存储的需求为C1、基于AI的波束管理任务对存储的需求为C2时，如果C1+C2<C0，则可以向UE配置基于AI的CSI压缩任务以及基于AI的波束管理任务。
S203，当待配置的AI处理任务为多个，且多个待配置的AI处理任务对处理能力的需求总和大于AI处理能力时，确定AI处理任务的发生时间、延迟间隔和优先级中的至少一个。
在本公开中，作为一种可选的实施例，网络设备判断多个AI处理任务对用户设备的处理能力的需求是否满足处理能力时，如果多个待配置的AI处理任务对处理能力的需求总和大于UE上报的AI处理能力，则网络设备有多种配置方式，下文将分情况予以讨论。
在一种可选实施例中，如果多个待配置的AI处理任务对处理能力的需求总和大于UE上报的AI处理能力，网络设备可以直接确定配置规则为：禁止向UE配置多个AI处理任务。
例如，当网络设备配置进行基于AI的CSI压缩任务对存储的需求为C1、基于AI的波束管理任务对存储的需求为C2、基于AI的定位任务对存储的需求为C3时，如果C1+C2+C3＞C0，可以确定配置规则为：禁止向UE配置基于AI的CSI压缩任务、基于AI的波束管理任务以及基于AI的定位任务。
在本公开的可选实施例中，如果多个待配置的AI处理任务对处理能力的需求总和大于UE上报的AI处理能力，网络设备可以进行二级判断，即，进一步判断多个AI处理任务中的两个或两个以上AI处理任务对处理能力的需求与AI处理能力之间的关系，并确定AI处理任务的发生时间、延迟间隔和优先级中的至少一个，以据此确定配置规则。
S204，根据AI处理任务的发生时间、延迟间隔和优先级中的至少一个，确定配置规则。
在本公开中，该步骤中具体包括：当多个待配置的AI处理任务中的两个或两个以上AI处理任务对处理能力的需求满足同时处理条件，将两个AI处理任务的发生时间配置为相同。
其中，同时处理条件为：两个AI处理任务对处理能力的需求总和小于或等于AI处理能力。
换言之，网络配置给终端AI处理任务对处理能力的需求总和可以大于终端上报的处理能力，但是处理的AI任务需满足预设关系。例如，同时处理的AI任务不能大于终端上报的处理能力。
举例而言，如上述的示例中，当网络设备配置进行基于AI的CSI压缩任务对存储的需求为C1、基于AI的波束管理任务对存储的需求为C2、基于AI的定位任务对存储的需求为C3时，如果C1+C2+ C3＞C0，但C1+C2＜C0，则可以确定配置规则为：向UE配置基于AI的CSI压缩任务以及基于AI的波束管理任务。
在一种可选的实施例中，该步骤还可以包括：当多个待配置的AI处理任务中的两个或两个以上AI处理任务对处理能力的需求不满足同时处理条件，将配置规则确定为下述任一项：禁止向用户设备配置两个或两个以上AI处理任务；配置两个或两个以上AI处理任务的发生时间之间具有延迟间隔；配置两个或两个以上AI处理任务具有不同的优先级。
举例而言，如上述的示例中，当网络设备配置进行基于AI的CSI压缩任务对存储的需求为C1、基于AI的波束管理任务对存储的需求为C2、基于AI的定位任务对存储的需求为C3时，如果C1+C2+C3＞C0，且C1+C2＞C0，则可以确定配置规则为：禁止向UE配置基于AI的CSI压缩任务以及基于AI的波束管理任务。
作为一种另一种可选方式，网络设备还可以确定配置规则为：配置两个或两个以上AI处理任务的发生时间之间具有延迟间隔。换言之，网络配置给终端AI处理任务对处理能力的需求总和可以大于终端上报的处理能力，但是处理的AI任务需满足预设关系，例如，AI处理任务需间隔X个时间单元。
如上示例，当网络设备配置进行基于AI的CSI压缩任务对存储的需求为C1、基于AI的波束管理任务对存储的需求为C2、基于AI的定位任务对存储的需求为C3时，如果C1+C2+C3＞C0，且C1+C2＞C0，则可以配置规则可以确定为：向UE配置基于AI的CSI压缩任务以及基于AI的波束管理任务之间的发生时间具有延迟间隔。换言之，基于AI的CSI压缩任务以及基于AI的波束管理任务不能同时处理，而是需要间隔X个时间单元。例如，可以配置终端在t0～t1时间段处理CSI压缩任务，在t2～t3时间段处理波束管理任务。
应当理解的是，某一项AI处理任务对应的延迟间隔可以是该AI处理任务与前一AI处理任务之间的间隔，也可以是该AI处理任务与首个执行的AI处理任务之间的间隔，对此本公开中不予限制。此外，该延迟间隔应当大于处理AI任务的处理时间。即，先处理的AI处理任务与后处理的AI处理任务之间的延时间隔应大于先处理的AI处理任务的处理时间。
按照上述描述，在另一个可选示例中，如果C1+C2+C3＞C0，但C1+C2＜C0，可以将配置规则进一步确定为：向UE配置基于AI的CSI压缩任务以及基于AI的波束管理任务，并配置基于AI的定位任务与基于AI的CSI压缩任务以及基于AI的波束管理任务之间具有延迟间隔。例如，可以配置终端在t0～t1时间段处理基于AI的CSI压缩任务和波束管理任务，在t2～t3区间处理基于AI的定位任务。换言之，基于AI的定位任务对应的延迟间隔可以理解为t0-t2之间的时间段长度。
再例如，网络设备可以配置终端在t0～t1时间段处理基于AI的CSI压缩任务，在t2～t3时间段处理波束管理任务，在t4～t5时间段处理基于AI的定位任务。则基于AI的定位任务对应的延迟间隔可以理解为t0-t4之间的时间段长度，或t2-t4之间的时间段长度，在本公开中不予限制。
本公开中的延迟间隔可以是一个预设的值，也可以是网络配置的值，本公开并不限制上述时间单元或延迟间隔的数值，其依据具体情况而定。可以理解的是，该延迟间隔可以用时间单元衡量。时间单元的值在本公开中不予限制。
在一种可选方案中，延迟间隔可以根据各个AI处理任务的优先级确定，其中，按照优先级降序的顺序，前一AI处理任务与后一AI处理任务之间的延时间隔大于前一AI处理任务的处理时间。
例如，在上述示例中，在t2～t3时间段处理波束管理任务，在t4～t5时间段处理基于AI的定位任务，基于AI的定位任务对应的延迟间隔为t2-t4之间的时间段长度，则t2-t4的时间段长度应当大于等于处理波束管理任务所需要的处理时间。
在t0～t1时间段处理基于AI的CSI压缩任务和波束管理任务，在t2～t3区间处理基于AI的定位任务，则t0-t2的时间段长度应当大于等于处理基于AI的CSI压缩任务和波束管理任务中时间最长的一个。
可以理解的是，在该实施例中，网络设备可以不向UE配置执行各个AI处理任务的顺序，而只配置各个AI处理任务之间的时间间隔，以保证可能超过UE处理能力的多个AU处理任务不同时执行即可。
作为一种可选方式，网络设备还可以确定配置规则为：配置两个或两个以上AI处理任务具有不同的优先级。
换言之，网络配置给终端AI处理任务对处理能力的需求总和可以大于终端上报的处理能力，但是处理的AI任务需满足预设关系，例如，AI处理任务之间的执行优先级不同，可以按照优先级顺序执行各个AI处理任务。
如上示例，当网络设备配置进行基于AI的CSI压缩任务对存储的需求为C1、基于AI的波束管理任务对存储的需求为C2、基于AI的定位任务对存储的需求为C3时，如果C1+C2+C3＞C0，且C1+C2＞C0，则可以配置规则可以确定为：向UE配置基于AI的CSI压缩任务以及基于AI的波束管理任务具有不同优先级。换言之，基于AI的CSI压缩任务以及基于AI的波束管理任务不能同时处理，而是需要按照一定的优先级顺序依次处理。例如，可以配置任务的优先级为CSI>波束管理>定位，则终端可以按照先执行基于AI的CSI压缩任务，然后再执行基于AI的波束管理任务的规则进行。
按照上述描述，在另一个可选示例中，如果C1+C2+C3＞C0，可以配置各个AI任务的优先级为CSI>波束管理>定位，则则终端可以按照先执行基于AI的CSI压缩任务，再执行基于AI的波束管理任务，然后再执行基于AI的定位任务的规则进行，或者仅执行前两个任务，放弃优先级较低的第三项任务。
换言之，在本实施例中，对于可能出现处理能力不足的情况，此时网络设备可以预设优先级。终端根据处理优先级对AI任务进行处理或者放弃低优先级的任务处理。
具体而言，网络设备配置优先级的方式可以包括：根据网络设备和用户设备之间的网络协议，确定各个待配置的AI处理任务的优先级；或根据多个待配置的AI处理任务的重要等级或紧急等级，对各个AI处理任务配置优先级。优先级的确定方式还可以是其他方式，在本公开中不予限制。
在本公开中，不同优先级的两个或两个以上AI处理任务之间具有延时间隔，该延时间隔可以参照上述实施例中的解释。
可以理解的是，相较于上述示例而言，在该实施例中，网络设备除了配置各个AI处理任务之间的时间间隔，还可以向UE配置执行各个AI处理任务的顺序，以保证可能超过UE处理能力的多个AU处理任务不同时执行，并且按照一定的优先级顺序执行。与上述示例相同的是，先执行的处理任务与后执行的处理任务之间具有延迟间隔，从而保证AI处理任务在执行过程中不超过UE的处理能力。
应当理解的是，在对多个AI处理任务进行配置时，配置规则可以不同，即可以同时适用多种配置规则。例如，可以同时执行第一项和第二项任务，并与第三项任务之间具有一延迟间隔，则执行第一项 和第二项任务后间隔一定时间后执行第三项任务，第三项任务的优先级高于第四项任务，则先执行第三项任务再执行第四项任务。
综上，根据本公开的人工智能应用管理方法，网络设备能够根据AI处理能力以及待配置的AI处理任务对处理能力的需求之间的关系，确定向用户设备配置AI处理任务的不同的配置规则，例如允许或禁用或以某一延迟间隔执行或以某一优先级执行等，充分考虑了各个情况下对AI应用的调度和配置，优化了通信过程中各个任务的调度，避免处理任务超过终端处理能力而带来的低效率甚至卡断问题。
图3示出了根据本公开实施例的一种人工智能应用管理方法的流程示意图。该方法由用户设备执行，在本公开的实施例中，用户设备(User Equipment，UE)包括但不限于智能终端设备、蜂窝电话、无线设备、手持机、移动单元、车辆、车载设备等，在本公开中不予限制。
如图3所示，该方法可以包括以下步骤。
S301，向网络设备发送人工智能AI处理能力。
本公开中，UE向网络设备发送的AI处理能力包括存储能力和/或计算能力。其中，存储能力用以存储模型以及推理中产生的中间参数，计算能力用以计算出推理的结果。本公开中该处理能力还可以包括其他用于衡量终端处理能力的参数等，在此不予限制。
S302，接收并执行网络设备配置的AI处理任务。
在本公开的实施例中，UE可以从网络设备接收网络设备根据终端处理能力上限与处理任务之间的关系确定的向用户设备配置AI处理任务的配置规则，换言之，网络设备可以判断是否要给终端配置某个AI处理任务或者确定配置AI处理任务的发生时间，或者确定某个处理任务会出现延迟等信息，则UE能够按照网络设备的配置进行执行。
综上，根据本公开的人工智能应用管理方法，UE能够向网络设备发送人工智能AI处理能力，接收并执行网络设备配置的AI处理任务，从而按照网络设备的配置执行各项AI处理任务，优化了通信过程中各个任务的调度，该AI处理任务的协调方案避免了处理任务超过终端处理能力而带来的低效率甚至卡断问题。
图4为根据本公开实施例的一种人工智能应用管理方法的流程示意图。该方法应用于UE，基于图3所示的实施例，如图4所示，该方法可以包括以下步骤。
S401，向网络设备发送人工智能AI处理能力。
具体地，在本公开的实施例中，该步骤包括：向网络设备发送用户设备的存储能力和/或计算能力。
步骤S401与图3所示的实施例中步骤S301原理相同，可参照S403的相关描述，在此不再赘述。
S402，接收AI处理任务的发生时间、延迟间隔和优先级中的至少一个。
在本公开的实施例中，根据图1和图2所示的实施例，如果网络设备确定配置规则为禁止配置AI处理任务给UE，则UE将不会接收到相关AI任务，因此，对于UE侧，该种情况则不在讨论范围内。
进一步地，根据图1和图2所示的实施例，如果网络设备确定向UE配置AI处理任务，则所配置的AI处理任务势必满足网络侧的判断条件，则UE可以根据网络侧的配置，接收到具体的配置规则以及将要执行的AI处理任务，以执行一个或多个AI处理任务。其中，UE可以接收到网络设备确定的配置规则，例如执行AI处理任务的发生时间、延迟间隔或优先级。
S403，根据AI处理任务的发生时间、延迟间隔和优先级中的至少一个，执行AI处理任务。
在本公开的实施例中，该步骤具体包括：在发生时间执行对应的AI处理任务；和/或在AI处理任务对应的延迟间隔后执行AI处理任务；和/或以优先级降序顺序执行AI处理任务。
应当理解的是，某一项AI处理任务对应的延迟间隔可以是该AI处理任务与前一AI处理任务之间的间隔，也可以是该AI处理任务与首个执行的AI处理任务之间的间隔，对此本公开中不予限制。此外，该延迟间隔应当大于处理AI任务的处理时间。即，先处理的AI处理任务与后处理的AI处理任务之间的延时间隔应大于先处理的AI处理任务的处理时间。
相对于如图2所示实施例，下面对该实施例进行对照描述。
在本公开的一些实施例中，UE上报的AI处理能力用于辅助网络设备确定对UE配置AI处理任务的规则。换言之，网络设备能够根据AI处理能力以及待配置的AI处理任务对处理能力的需求之间的关系，确定向用户设备配置AI处理任务的配置规则。
例如，多个待配置的AI处理任务对处理能力的需求总和小于或等于UE上报的AI处理能力时，则网络设备确定向用户设备配置AI处理任务的配置规则为：为UE配置多个AI处理任务。
此时，由于UE的处理能力大于将要配置的AI处理任务的需求，则UE可以接收到网络设备为其配置的这些AI处理任务，并执行。在这种情况下，网络设备无需为终端配置其他调度规则。
以AI处理能力为存储能力为例，终端对AI任务总的存储大小为C0，即，UE上报的AI处理能力为C0，当网络设备配置进行基于AI的CSI压缩任务对存储的需求为C1、基于AI的波束管理任务对存储的需求为C2、基于AI的定位任务的需求为C3时，如果C1+C2+C3<C0，则可以向UE配置这三项任务。
换言之，在该实施例中，由于多个AI处理任务对UE处理能力的需求并未超过UE上报的处理能力，因此网络设备可以直接向UE配置执行多个AI处理任务，可以不指定各个AI处理任务的发生时间或其他参数，因此UE可以直接执行多个AI处理任务。
当多个待配置的AI处理任务对处理能力的需求总和大于UE上报的AI处理能力时，则网络设备确定向用户设备配置AI处理任务的配置规则需要分情况讨论。
在上述示例中，如果C1+C2+C3＞C0，但C1+C2<C0，则可以向UE配置基于AI的CSI压缩任务以及基于AI的波束管理任务。
此时，UE可以接收到网络设备向其配置的AI处理任务为基于AI的CSI压缩任务以及基于AI的波束管理任务，并且可以同时执行这两项任务。该情况与上一示例类似。
在该情况下，对于基于AI的定位任务而言，有两种情况：1)网络设备禁止向UE配置该任务；2)网络设备配置该任务执行的延迟间隔或优先级。
其中，对于上述情况1)，UE将不会接收到关于执行基于AI的定位任务相关的信息。对于上述情况2)，UE可能接收到网络设备发送的执行该任务的延迟间隔或优先级。例如，网络设备配置先执行基于AI的CSI压缩任务以及基于AI的波束管理任务，间隔X个时间单元之后执行基于AI的定位任务，或者，配置基于AI的定位任务的优先级低于基于AI的CSI压缩任务以及基于AI的波束管理任务，则UE根据该优先级执行。
例如，UE可以根据网络设备的配置，在t0～t1时间段处理基于AI的CSI压缩任务，在t2～t3区间处理基于AI的定位任务。
对于UE而言，只需要根据网络设备的配置执行即可。在本公开的实施例中，网络设备还可以告知UE该时间间隔是相对于哪一个AI处理任务的间隔，则UE可以对应执行。再例如，UE可以根据网络设备的配置，按照CSI>波束管理>定位的优先级顺序，先执行基于AI的CSI压缩任务，再执行基于AI的波束管理任务，然后再执行基于AI的定位任务。关于该示例在图2所述的实施例中已经予以详述，在此不再赘述。
在一种可选的实施例中，UE可以以优先级降序顺序，执行预设个数的AI处理任务，放弃执行剩余AI处理任务。例如，按照CSI>波束管理>定位的优先级顺序，仅执行前两个任务，放弃优先级较低的第三项任务。
换言之，网络设备可以只告知UE各个AI处理任务的在执行时需要间隔的延迟间隔，而无需告知具体的执行顺序，或者网络设备还可以告知UE以一定的优先级顺序执行各个AI处理任务，UE可以根据网络设备的配置进行执行。
其中，延迟间隔或时间单元可以是一个预设的值，也可以是网络配置的值，本公开并不限制上述时间单元或延迟间隔的数值，其依据具体情况而定。
另外，本公开中的优先级可以是根据网络设备和用户设备之间的网络协议确定的，也可以是根据网络设备根据多个待配置的AI处理任务的重要等级或紧急等级进行配置的，在本公开中不予限制。
综上，根据本公开的人工智能应用管理方法，UE能够向网络设备上报AI处理能力，从而辅助网络设备根据AI处理能力以及待配置的AI处理任务对处理能力的需求之间的关系确定向用户设备配置AI处理任务的不同的配置规则，UE能够接收并按照网络设备的配置执行AI处理任务，充分考虑了各个情况下对AI应用的调度和配置，优化了通信过程中各个任务的调度，避免处理任务超过终端处理能力而带来的低效率甚至卡断问题。
上述本申请提供的实施例中，分别从网络设备和用户设备的角度对本申请实施例提供的方法进行了介绍。为了实现上述本申请实施例提供的方法中的各功能，网络设备和用户设备可以包括硬件结构、软件模块，以硬件结构、软件模块、或硬件结构加软件模块的形式来实现上述各功能。上述各功能中的某个功能可以以硬件结构、软件模块、或者硬件结构加软件模块的方式来执行。
与上述几种实施例提供的人工智能应用管理方法相对应，本公开还提供一种人工智能应用管理装置，由于本公开实施例提供的人工智能应用管理装置与上述几种实施例提供的人工智能应用管理方法相对应，因此人工智能应用管理方法的实施方式也适用于本实施例提供的人工智能应用管理装置，在本实施例中不再详细描述。
图5为本公开实施例提供的一种人工智能应用管理装置500的结构示意图，该人工智能应用管理装置500可用于网络设备。
如图5所示，该装置500可以包括：接收单元510，用于接收用户设备发送的人工智能AI处理能力；以及配置单元520，用于根据AI处理能力以及待配置的AI处理任务对处理能力的需求之间的关系，确定向用户设备配置AI处理任务的配置规则。
根据本公开实施例提供的人工智能应用管理装置，网络设备能够接收用户设备发送的AI处理能力，并根据该AI处理能力以及待配置的AI处理任务对处理能力的需求之间的关系，确定向用户设备配置 AI处理任务的配置规则，从而优化了通信过程中各个任务的调度，该AI处理任务的协调方案避免了处理任务超过终端处理能力而带来的低效率甚至卡断问题。
在一些实施例中，配置单元520用于：当待配置的AI处理任务为多个，且多个待配置的AI处理任务对处理能力的需求总和小于或等于AI处理能力时，向用户设备配置多个待配置的AI处理任务。
在一些实施例中，配置单元520用于：当待配置的AI处理任务为多个，且多个待配置的AI处理任务对处理能力的需求总和大于AI处理能力时，确定AI处理任务的发生时间、延迟间隔和优先级中的至少一个；根据AI处理任务的发生时间、延迟间隔和优先级中的至少一个，确定配置规则。
在一些实施例中，配置单元520用于：当多个待配置的AI处理任务中的两个或两个以上AI处理任务对处理能力的需求满足同时处理条件，将配置规则确定为：配置两个或两个以上AI处理任务的发生时间相同。
在一些实施例中，配置单元520还用于，当多个待配置的AI处理任务中的两个或两个以上AI处理任务对处理能力的需求不满足同时处理条件，将配置规则确定为以下任一项：禁止向用户设备配置两个或两个以上AI处理任务；配置两个或两个以上AI处理任务的发生时间具有延迟间隔；配置两个或两个以上AI处理任务具有不同的优先级，其中，同时处理条件为：两个或两个以上AI处理任务对处理能力的需求总和小于或等于AI处理能力。
在一些实施例中，配置单元520用于：根据网络设备和用户设备之间的网络协议，确定各个待配置的AI处理任务的优先级；或根据多个待配置的AI处理任务的重要等级或紧急等级，对各个AI处理任务配置优先级。
在一些实施例中，不同优先级的两个或两个以上AI处理任务之间具有延时间隔，且按照优先级降序的顺序，前一AI处理任务与后一AI处理任务之间的延时间隔大于前一AI处理任务的处理时间。
在一些实施例中，AI处理能力包括用户设备的存储能力和/或计算能力。
综上，根据本公开的人工智能应用管理装置，网络设备能够根据AI处理能力以及待配置的AI处理任务对处理能力的需求之间的关系，确定向用户设备配置AI处理任务的不同的配置规则，例如允许或禁用或以某一延迟间隔执行或以某一优先级执行等，充分考虑了各个情况下对AI应用的调度和配置，优化了通信过程中各个任务的调度，避免处理任务超过终端处理能力而带来的低效率甚至卡断问题。
图6为本公开实施例提供的一种人工智能应用管理装置600的结构示意图。该人工智能应用管理装置600可用于用户设备。
如图6所示，该装置600可以包括：收发单元610，用于向网络设备发送人工智能AI处理能力，以及接收网络设备配置的AI处理任务；以及执行单元620，用于执行AI处理任务。
根据本公开的人工智能应用管理装置，UE能够向网络设备发送人工智能AI处理能力，接收并执行网络设备配置的AI处理任务，从而按照网络设备的配置执行各项AI处理任务，优化了通信过程中各个任务的调度，该AI处理任务的协调方案避免了处理任务超过终端处理能力而带来的低效率甚至卡断问题。
在一些实施例中，收发单元610用于：向网络设备发送用户设备的存储能力和/或计算能力。
在一些实施例中，收发单元610用于：接收AI处理任务的发生时间、延迟间隔和优先级中的至少一个。
在一些实施例中，执行单元620用于：根据AI处理任务的发生时间、延迟间隔和优先级中的至少一个，执行AI处理任务。
在一些实施例中，执行单元620用于：在发生时间执行对应的AI处理任务；和/或在前一AI处理任务执行延迟间隔后执行AI处理任务；和/或以优先级降序顺序执行AI处理任务。
在一些实施例中，执行单元620用于：以优先级降序顺序，执行预设个数的AI处理任务，放弃执行剩余AI处理任务。
根据本公开的人工智能应用管理装置，UE能够向网络设备上报AI处理能力，从而辅助网络设备根据AI处理能力以及待配置的AI处理任务对处理能力的需求之间的关系确定向用户设备配置AI处理任务的不同的配置规则，UE能够接收并按照网络设备的配置执行AI处理任务，充分考虑了各个情况下对AI应用的调度和配置，优化了通信过程中各个任务的调度，避免处理任务超过终端处理能力而带来的低效率甚至卡断问题。
请参见图7，图11是本申请实施例提供的一种通信装置700的结构示意图。通信装置700可以是网络设备，也可以是用户设备，也可以是支持网络设备实现上述方法的芯片、芯片系统、或处理器等，还可以是支持用户设备实现上述方法的芯片、芯片系统、或处理器等。该装置可用于实现上述方法实施例中描述的方法，具体可以参见上述方法实施例中的说明。
通信装置700可以包括一个或多个处理器701。处理器701可以是通用处理器或者专用处理器等。例如可以是基带处理器或中央处理器。基带处理器可以用于对通信协议以及通信数据进行处理，中央处理器可以用于对通信装置(如，基站、基带芯片，终端设备、终端设备芯片，DU或CU等)进行控制，执行计算机程序，处理计算机程序的数据。
可选的，通信装置700中还可以包括一个或多个存储器702，其上可以存有计算机程序704，处理器701执行计算机程序704，以使得通信装置700执行上述方法实施例中描述的方法。可选的，存储器702中还可以存储有数据。通信装置700和存储器702可以单独设置，也可以集成在一起。
可选的，通信装置700还可以包括收发器705、天线706。收发器705可以称为收发单元、收发机、或收发电路等，用于实现收发功能。收发器705可以包括接收器和发送器，接收器可以称为接收机或接收电路等，用于实现接收功能；发送器可以称为发送机或发送电路等，用于实现发送功能。
可选的，通信装置700中还可以包括一个或多个接口电路707。接口电路707用于接收代码指令并传输至处理器701。处理器701运行代码指令以使通信装置700执行上述方法实施例中描述的方法。
在一种实现方式中，处理器701中可以包括用于实现接收和发送功能的收发器。例如该收发器可以是收发电路，或者是接口，或者是接口电路。用于实现接收和发送功能的收发电路、接口或接口电路可以是分开的，也可以集成在一起。上述收发电路、接口或接口电路可以用于代码/数据的读写，或者，上述收发电路、接口或接口电路可以用于信号的传输或传递。
在一种实现方式中，处理器701可以存有计算机程序703，计算机程序703在处理器701上运行，可使得通信装置700执行上述方法实施例中描述的方法。计算机程序703可能固化在处理器701中，该种情况下，处理器701可能由硬件实现。
在一种实现方式中，通信装置700可以包括电路，该电路可以实现前述方法实施例中发送或接收或者通信的功能。本申请中描述的处理器和收发器可实现在集成电路(integrated circuit，IC)、模拟IC、 射频集成电路RFIC、混合信号IC、专用集成电路(application specific integrated circuit，ASIC)、印刷电路板(printed circuit board，PCB)、电子设备等上。该处理器和收发器也可以用各种IC工艺技术来制造，例如互补金属氧化物半导体(complementary metal oxide semiconductor，CMOS)、N型金属氧化物半导体(nMetal-oxide-semiconductor，NMOS)、P型金属氧化物半导体(positive channel metal oxide semiconductor，PMOS)、双极结型晶体管(bipolar junction transistor，BJT)、双极CMOS(BiCMOS)、硅锗(SiGe)、砷化镓(GaAs)等。
以上实施例描述中的通信装置可以是网络设备或者用户设备，但本申请中描述的通信装置的范围并不限于此，而且通信装置的结构可以不受图10的限制。通信装置可以是独立的设备或者可以是较大设备的一部分。例如该通信装置可以是：
(1)独立的集成电路IC，或芯片，或，芯片系统或子系统；
(2)具有一个或多个IC的集合，可选的，该IC集合也可以包括用于存储数据，计算机程序的存储部件；
(3)ASIC，例如调制解调器(Modem)；
(4)可嵌入在其他设备内的模块；
(5)接收机、终端设备、智能终端设备、蜂窝电话、无线设备、手持机、移动单元、车载设备、网络设备、云设备、人工智能设备等等；
(6)其他等等。
对于通信装置可以是芯片或芯片系统的情况，可参见图8所示的芯片的结构示意图。图8所示的芯片包括处理器801和接口802。其中，处理器801的数量可以是一个或多个，接口802的数量可以是多个。
可选的，芯片还包括存储器803，存储器803用于存储必要的计算机程序和数据。
本领域技术人员还可以了解到本申请实施例列出的各种说明性逻辑块(illustrative logical block)和步骤(step)可以通过电子硬件、电脑软件，或两者的结合进行实现。这样的功能是通过硬件还是软件来实现取决于特定的应用和整个系统的设计要求。本领域技术人员可以对于每种特定的应用，可以使用各种方法实现的功能，但这种实现不应被理解为超出本申请实施例保护的范围。
本申请还提供一种可读存储介质，其上存储有指令，该指令被计算机执行时实现上述任一方法实施例的功能。
本申请还提供一种计算机程序产品，该计算机程序产品被计算机执行时实现上述任一方法实施例的功能。
在上述实施例中，可以全部或部分地通过软件、硬件、固件或者其任意组合来实现。当使用软件实现时，可以全部或部分地以计算机程序产品的形式实现。计算机程序产品包括一个或多个计算机程序。在计算机上加载和执行计算机程序时，全部或部分地产生按照本申请实施例的流程或功能。计算机可以是通用计算机、专用计算机、计算机网络、或者其他可编程装置。计算机程序可以存储在计算机可读存储介质中，或者从一个计算机可读存储介质向另一个计算机可读存储介质传输，例如，计算机程序可以从一个网站站点、计算机、服务器或数据中心通过有线(例如同轴电缆、光纤、数字用户线(digital subscriber line，DSL))或无线(例如红外、无线、微波等)方式向另一个网站站点、计算机、服务器 或数据中心进行传输。计算机可读存储介质可以是计算机能够存取的任何可用介质或者是包含一个或多个可用介质集成的服务器、数据中心等数据存储设备。可用介质可以是磁性介质(例如，软盘、硬盘、磁带)、光介质(例如，高密度数字视频光盘(digital video disc，DVD))、或者半导体介质(例如，固态硬盘(solid state disk，SSD))等。
本领域普通技术人员可以理解：本申请中涉及的第一、第二等各种数字编号仅为描述方便进行的区分，并不用来限制本申请实施例的范围，也表示先后顺序。
本申请中的至少一个还可以描述为一个或多个，多个可以是两个、三个、四个或者更多个，本申请不做限制。在本申请实施例中，对于一种技术特征，通过“第一”、“第二”、“第三”、“A”、“B”、“C”和“D”等区分该种技术特征中的技术特征，该“第一”、“第二”、“第三”、“A”、“B”、“C”和“D”描述的技术特征间无先后顺序或者大小顺序。
如本文使用的，术语“机器可读介质”和“计算机可读介质”指的是用于将机器指令和/或数据提供给可编程处理器的任何计算机程序产品、设备、和/或装置(例如，磁盘、光盘、存储器、可编程逻辑装置(PLD))，包括，接收作为机器可读信号的机器指令的机器可读介质。术语“机器可读信号”指的是用于将机器指令和/或数据提供给可编程处理器的任何信号。
可以将此处描述的系统和技术实施在包括后台部件的计算系统(例如，作为数据服务器)、或者包括中间件部件的计算系统(例如，应用服务器)、或者包括前端部件的计算系统(例如，具有图形用户界面或者网络浏览器的用户计算机，用户可以通过该图形用户界面或者该网络浏览器来与此处描述的系统和技术的实施方式交互)、或者包括这种后台部件、中间件部件、或者前端部件的任何组合的计算系统中。可以通过任何形式或者介质的数字数据通信(例如，通信网络)来将系统的部件相互连接。通信网络的示例包括：局域网(LAN)、广域网(WAN)和互联网。
计算机系统可以包括客户端和服务器。客户端和服务器一般远离彼此并且通常通过通信网络进行交互。通过在相应的计算机上运行并且彼此具有客户端-服务器关系的计算机程序来产生客户端和服务器的关系。
应该理解，可以使用上面所示的各种形式的流程，重新排序、增加或删除步骤。例如，本公开中记载的各步骤可以并行地执行也可以顺序地执行也可以不同的次序执行，只要能够实现本公开公开的技术方案所期望的结果，本文在此不进行限制。
此外，应该理解，本申请的各种实施例可以单独实施，也可以在方案允许的情况下与其他实施例组合实施。
本领域普通技术人员可以意识到，结合本文中所公开的实施例描述的各示例的单元及算法步骤，能够以电子硬件、或者计算机软件和电子硬件的结合来实现。这些功能究竟以硬件还是软件方式来执行，取决于技术方案的特定应用和设计约束条件。专业技术人员可以对每个特定的应用来使用不同方法来实现所描述的功能，但是这种实现不应认为超出本申请的范围。
所属领域的技术人员可以清楚地了解到，为描述的方便和简洁，上述描述的系统、装置和单元的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。
以上，仅为本申请的具体实施方式，但本申请的保护范围并不局限于此，任何熟悉本技术领域的技术人员在本申请揭露的技术范围内，可轻易想到变化或替换，都应涵盖在本申请的保护范围之内。因此，本申请的保护范围应以权利要求的保护范围为准。

Claims (17)

  1. 一种人工智能应用管理方法，其特征在于，所述方法由网络设备执行，所述方法包括：
    接收用户设备发送的人工智能AI处理能力；以及
    根据所述AI处理能力以及待配置的AI处理任务对处理能力的需求之间的关系，确定向所述用户设备配置AI处理任务的配置规则。
  2. 根据权利要求1所述的方法，其特征在于，所述确定向所述用户设备配置AI处理任务的配置规则包括：
    当所述待配置的AI处理任务为多个，且所述多个待配置的AI处理任务对处理能力的需求总和小于或等于所述AI处理能力时，向所述用户设备配置所述多个待配置的AI处理任务。
  3. 根据权利要求1或2所述的方法，其特征在于，所述确定向所述用户设备配置AI处理任务的配置规则包括：
    当所述待配置的AI处理任务为多个，且所述多个待配置的AI处理任务对处理能力的需求总和大于所述AI处理能力时，确定AI处理任务的发生时间、延迟间隔和优先级中的至少一个；
    根据所述AI处理任务的发生时间、延迟间隔和优先级中的至少一个，确定所述配置规则。
  4. 根据权利要求3所述的方法，其特征在于，所述根据所述AI处理任务的发生时间、延迟间隔和优先级中的至少一个，确定所述配置规则包括：
    当所述多个待配置的AI处理任务中的两个或两个以上AI处理任务对处理能力的需求满足同时处理条件，将所述配置规则确定为：配置所述两个或两个以上AI处理任务的发生时间相同。
  5. 根据权利要求3或4所述的方法，其特征在于，所述根据所述AI处理任务的发生时间、延迟间隔和优先级中的至少一个，确定所述配置规则包括：
    当所述多个待配置的AI处理任务中的两个或两个以上AI处理任务对处理能力的需求不满足同时处理条件，将所述配置规则确定为以下任一项：
    禁止向所述用户设备配置所述两个或两个以上AI处理任务；
    配置所述两个或两个以上AI处理任务的发生时间之间具有所述延迟间隔；
    配置所述两个或两个以上AI处理任务具有不同的优先级，
    其中，所述同时处理条件为：所述两个或两个以上AI处理任务对处理能力的需求总和小于或等于所述AI处理能力。
  6. 根据权利要求3所述的方法，其特征在于，所述确定AI处理任务的优先级包括：
    根据所述网络设备和所述用户设备之间的网络协议，确定各个待配置的AI处理任务的优先级；或
    根据所述多个待配置的AI处理任务的重要等级或紧急等级，对各个AI处理任务配置优先级。
  7. 根据权利要求3至6中任一项所述的方法，其特征在于，
    不同优先级的两个或两个以上AI处理任务之间具有所述延时间隔，
    按照优先级降序的顺序，前一AI处理任务与后一AI处理任务之间的延时间隔大于前一AI处理任务的处理时间。
  8. 根据权利要求1至7中任一项所述的方法，其特征在于，所述AI处理能力包括所述用户设备的存储能力和/或计算能力。
  9. 一种人工智能应用管理方法，其特征在于，所述方法由用户设备执行，所述方法包括：
    向网络设备发送人工智能AI处理能力；以及
    接收并执行所述网络设备配置的AI处理任务。
  10. 根据权利要求9所述的方法，其特征在于，所述向所述网络设备所述AI处理能力包括：
    向所述网络设备发送所述用户设备的存储能力和/或计算能力。
  11. 根据权利要求9或10所述的方法，其特征在于，所述接收并执行所述网络设备配置的AI处理任务包括：
    接收所述AI处理任务的发生时间、延迟间隔和优先级中的至少一个；
    根据所述AI处理任务的发生时间、延迟间隔和优先级中的至少一个，执行所述AI处理任务。
  12. 根据权利要求11所述的方法，其特征在于，所述根据所述AI处理任务的发生时间、延迟间隔和优先级中的至少一个，执行所述AI处理任务包括：
    在所述发生时间执行对应的AI处理任务；和/或
    在所述AI处理任务对应的延迟间隔后执行所述AI处理任务；和/或
    以优先级降序顺序执行所述AI处理任务。
  13. 根据权利要求11所述的方法，其特征在于，所述根据所述优先级，执行所述AI处理任务包括：
    以优先级降序顺序，执行预设个数的AI处理任务，放弃执行剩余AI处理任务。
  14. 一种人工智能应用管理装置，其特征在于，所述装置布置于网络设备，所述装置包括：
    接收单元，用于接收用户设备发送的人工智能AI处理能力；以及
    配置单元，用于根据所述AI处理能力以及待配置的AI处理任务对处理能力的需求之间的关系，确定向所述用户设备配置AI处理任务的配置规则。
  15. 一种人工智能应用管理装置，其特征在于，所述装置应用于用户设备，所述装置包括：
    收发单元，用于向网络设备发送人工智能AI处理能力，以及接收所述网络设备配置的AI处理任务；以及
    执行单元，用于执行所述AI处理任务。
  16. 一种通信设备，其中，包括：收发器；存储器；处理器，分别与所述收发器及所述存储器连接，配置为通过执行所述存储器上的计算机可执行指令，控制所述收发器的无线信号收发，并能够实现权利要求1-12中任一项所述的方法。
  17. 一种计算机存储介质，其中，所述计算机存储介质存储有计算机可执行指令；所述计算机可执行指令被处理器执行后，能够实现权利要求1-12中任一项所述的方法。
PCT/CN2022/103169 2022-06-30 2022-06-30 人工智能应用管理方法、装置及通信设备 WO2024000533A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
PCT/CN2022/103169 WO2024000533A1 (zh) 2022-06-30 2022-06-30 人工智能应用管理方法、装置及通信设备
CN202280002428.2A CN117642720A (zh) 2022-06-30 2022-06-30 人工智能应用管理方法、装置及通信设备

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2022/103169 WO2024000533A1 (zh) 2022-06-30 2022-06-30 人工智能应用管理方法、装置及通信设备

Publications (1)

Publication Number Publication Date
WO2024000533A1 true WO2024000533A1 (zh) 2024-01-04

Family

ID=89383865

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/103169 WO2024000533A1 (zh) 2022-06-30 2022-06-30 人工智能应用管理方法、装置及通信设备

Country Status (2)

Country Link
CN (1) CN117642720A (zh)
WO (1) WO2024000533A1 (zh)

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6757897B1 (en) * 2000-02-29 2004-06-29 Cisco Technology, Inc. Apparatus and methods for scheduling and performing tasks
CN108563500A (zh) * 2018-05-08 2018-09-21 深圳市零度智控科技有限公司 基于云平台的任务调度方法、云平台及计算机存储介质
US20210185515A1 (en) * 2019-12-16 2021-06-17 Qualcomm Incorporated Neural network configuration for wireless communication system assistance
WO2021212657A1 (zh) * 2020-04-24 2021-10-28 平安科技（深圳）有限公司 终端和服务器之间的任务调度方法、装置和调度器
CN113687950A (zh) * 2021-08-31 2021-11-23 平安医疗健康管理股份有限公司 基于优先级的任务分配方法、装置、设备及存储介质
WO2021248423A1 (zh) * 2020-06-12 2021-12-16 华为技术有限公司 人工智能资源的调度方法、装置、存储介质和芯片
CN113918314A (zh) * 2020-07-08 2022-01-11 中兴通讯股份有限公司 任务处理方法、边缘计算设备、计算机设备、介质
WO2022082658A1 (zh) * 2020-10-22 2022-04-28 华为技术有限公司 通信方法及装置

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6757897B1 (en) * 2000-02-29 2004-06-29 Cisco Technology, Inc. Apparatus and methods for scheduling and performing tasks
CN108563500A (zh) * 2018-05-08 2018-09-21 深圳市零度智控科技有限公司 基于云平台的任务调度方法、云平台及计算机存储介质
US20210185515A1 (en) * 2019-12-16 2021-06-17 Qualcomm Incorporated Neural network configuration for wireless communication system assistance
WO2021212657A1 (zh) * 2020-04-24 2021-10-28 平安科技（深圳）有限公司 终端和服务器之间的任务调度方法、装置和调度器
WO2021248423A1 (zh) * 2020-06-12 2021-12-16 华为技术有限公司 人工智能资源的调度方法、装置、存储介质和芯片
CN113918314A (zh) * 2020-07-08 2022-01-11 中兴通讯股份有限公司 任务处理方法、边缘计算设备、计算机设备、介质
WO2022082658A1 (zh) * 2020-10-22 2022-04-28 华为技术有限公司 通信方法及装置
CN113687950A (zh) * 2021-08-31 2021-11-23 平安医疗健康管理股份有限公司 基于优先级的任务分配方法、装置、设备及存储介质

Also Published As

Publication number Publication date
CN117642720A (zh) 2024-03-01

Similar Documents

Publication Publication Date Title
WO2023130322A1 (zh) 确定共享信道占用时间的方法及其装置
WO2023000149A1 (zh) 一种中继终端设备测量上报的方法及其装置
WO2023206179A1 (zh) 一种确定传输配置指示状态的方法及装置
WO2023245521A1 (zh) 一种控制资源的位置信息的确定方法及其装置
WO2023019411A1 (zh) 一种下行控制信息的对齐方法及其装置
WO2021244264A1 (zh) 通信方法、装置和系统
WO2024050776A1 (zh) 一种信息确定方法/装置/设备及存储介质
WO2024000533A1 (zh) 人工智能应用管理方法、装置及通信设备
WO2023206034A1 (zh) 混合自动重传请求harq反馈的处理方法及其装置
WO2017177438A1 (zh) 一种控制信息的传输方法及装置
WO2022266861A1 (zh) 寻呼处理方法、通信装置和存储介质
WO2024016244A1 (zh) 支持非激活态组播业务接收的配置方法及装置
WO2024098265A1 (zh) 一种上行数据传输方法及装置
WO2023035202A1 (zh) 一种跟踪参考信号周期的确定方法及其装置
WO2023283782A1 (zh) 一种信道状态反馈的方法及其装置
WO2024045039A1 (zh) 一种能力交互方法/装置/设备及存储介质
WO2023201755A1 (zh) 一种移动性管理的配置方法及装置
WO2022236622A1 (zh) 一种寻呼方法及其装置
WO2024000329A1 (zh) 一种信道占用信息的发送方法及其装置
WO2023245681A1 (zh) 基于轨道角动量的上行传输配置方法、装置及系统
WO2022266963A1 (zh) 资源分配方法及其装置
WO2024077501A1 (zh) 侧行链路的功率确定方法及装置
WO2024031272A1 (zh) 一种上报方法、装置、设备及存储介质
WO2024060143A1 (zh) 一种上报方法/装置/设备及存储介质
WO2024045078A1 (zh) 一种终端处理能力的上报方法、数据处理方法及其装置

Legal Events

Date Code Title Description
WWE Wipo information: entry into national phase

Ref document number: 202280002428.2

Country of ref document: CN

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22948635

Country of ref document: EP

Kind code of ref document: A1