WO2023020432A1 - 算力请求、算力分配、算力执行方法、终端及网络侧设备 - Google Patents

算力请求、算力分配、算力执行方法、终端及网络侧设备 Download PDF

Info

Publication number
WO2023020432A1
WO2023020432A1 PCT/CN2022/112497 CN2022112497W WO2023020432A1 WO 2023020432 A1 WO2023020432 A1 WO 2023020432A1 CN 2022112497 W CN2022112497 W CN 2022112497W WO 2023020432 A1 WO2023020432 A1 WO 2023020432A1
Authority
WO
WIPO (PCT)
Prior art keywords
computing power
task
target
information
terminal
Prior art date
Application number
PCT/CN2022/112497
Other languages
English (en)
French (fr)
Inventor
吕华章
Original Assignee
维沃移动通信有限公司
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 维沃移动通信有限公司 filed Critical 维沃移动通信有限公司
Publication of WO2023020432A1 publication Critical patent/WO2023020432A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/44Arrangements for executing specific programs
    • G06F9/445Program loading or initiating
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00Network arrangements or protocols for supporting network services or applications
    • H04L67/01Protocols
    • H04L67/10Protocols in which an application is distributed across nodes in the network
    • H04L67/1001Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
    • H04L67/1004Server selection for load balancing
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00Network arrangements or protocols for supporting network services or applications
    • H04L67/01Protocols
    • H04L67/10Protocols in which an application is distributed across nodes in the network
    • H04L67/1097Protocols in which an application is distributed across nodes in the network for distributed storage of data in networks, e.g. transport arrangements for network file system [NFS], storage area networks [SAN] or network attached storage [NAS]

Definitions

  • the present application belongs to the technical field of terminal communication, and specifically relates to a computing power request, computing power distribution, computing power execution method, terminal and network side equipment.
  • Computing power can be understood as the computing speed or computing power of a certain device, such as the computing speed or computing power of a server, a central processing unit (Central Processing Unit, CPU), a graphics processing unit (Graphics Processing Unit, GPU), a terminal, etc., the above-mentioned Devices contain computing units or computing capabilities.
  • computing power can be expressed by how many calculations are completed per second, such as the number of floating-point operations per second (FLOPS) performed per second, which is generally used to measure the parallel computing power of GPU.
  • FLOPS floating-point operations per second
  • hash rate is a measure of the processing power of the Bitcoin network. That is, the speed at which the computer (CPU) calculates the output of the hash function.
  • the Bitcoin network must perform intensive mathematical and cryptography-related operations for security purposes.
  • Computing power tasks can be understood as tasks to be completed that require a certain amount of computing power. For example, to complete a face recognition task for a certain group of photos, 10 GPUs will be consumed in computing power. Computing power requirements can be understood as what is required to complete the computing power task Computing resources consumed.
  • a terminal when a terminal has a computing power task, for example, there is a computing power task inside the terminal, or an application in the terminal has a computing power task to be executed, the terminal can use its own computing power resources to perform the computing power task locally.
  • the terminal when the terminal executes computing power tasks locally, there may be a problem of insufficient computing power, that is, the computing power requirements of the terminal computing power tasks or application computing power tasks are greater than the computing power capabilities of the terminal, thus affecting the computing power.
  • the execution efficiency of the task may even lead to the failure of the execution of the computing power task.
  • Embodiments of the present application provide a computing power request, computing power distribution, computing power execution method, terminal and network side equipment, which can solve the problem that the terminal cannot normally perform computing power tasks when the computing power is insufficient.
  • a computing power request method which includes:
  • the terminal sends a computing power request to the network side device, and the computing power request is used to request computing power for computing power tasks in the terminal;
  • the target computing power device includes a device or a device cluster with computing power capability
  • the computing power task is sent to the target computing power device based on the device information, and the computing power task is executed by the target computing power device.
  • a computing power request device including:
  • a first sending module configured to send a computing power request to a network side device, where the computing power request is used to request computing power for a computing power task in the terminal;
  • a receiving module configured to receive device information of a target computing power device returned by the network side device, where the target computing power device includes a device or a device cluster with computing power capability;
  • the second sending module is configured to send the computing power task to the target computing power device based on the device information, and the target computing power device executes the computing power task.
  • a computing power distribution method including:
  • the network side device receives a computing power request from the terminal, where the computing power request is used to request computing power for a computing power task in the terminal;
  • the target computing power device includes a device or a device cluster with computing power capability
  • the terminal sends the computing power task to the target computing power device based on the device information, and the target computing power device executes the Computational tasks.
  • a computing power distribution device including:
  • a receiving module configured to receive a computing power request from a terminal, where the computing power request is used to request computing power for a computing power task in the terminal;
  • a determining module configured to determine a target computing power device based on the computing power request, where the target computing power device includes a device or a device cluster with computing power capability;
  • a sending module configured to send the device information of the target computing power device to the terminal, the terminal sends the computing power task to the target computing power device based on the device information, and the target The computing power device performs the computing power task.
  • a computing power execution method including:
  • the target computing power device receives a computing power task from the terminal, and the computing power task is sent by the terminal based on the device information of the target computing power device, and the device information is sent by the network side device after receiving the computing power of the terminal After requesting and determining the target computing power device and returning it to the terminal, the computing power request is used to request computing power for the computing power task in the terminal, and the target computing power device includes a device with computing power capability or device cluster;
  • a computing power execution device including:
  • a receiving module configured to receive a computing power task from a terminal, the computing power task is sent by the terminal based on the device information of the target computing power device, and the device information is received by the network side device after receiving the computing power task of the terminal request and determine the target computing power device and return it to the terminal, the computing power request is used to request computing power for the computing power task in the terminal, and the target computing power device includes devices with computing power capabilities or device clusters;
  • a task execution module configured to execute the computing power task.
  • a terminal in a seventh aspect, includes a processor, a memory, and a program or instruction stored in the memory and operable on the processor, when the program or instruction is executed by the processor. The steps of the method described in the first aspect are realized.
  • a terminal including a processor and a communication interface, wherein the communication interface is used to send a computing power request to a network side device, and the computing power request is used to provide computing power tasks in the terminal request computing power; receive the device information of the target computing power device returned by the network side device, the target computing power device includes a device or device cluster with computing power capability; send the computing power task to The target computing power device is used to execute the computing power task by the target computing power device.
  • a network-side device includes a processor, a memory, and a program or instruction stored in the memory and operable on the processor, and the program or instruction is executed by the
  • the processor implements the steps of the method described in the third aspect when executing.
  • a network side device including a processor and a communication interface, wherein the communication interface is used to receive a computing power request from a terminal, and the computing power request is used to provide computing power for the terminal.
  • the task requests computing power
  • the processor is used to determine a target computing power device based on the computing power request
  • the target computing power device includes a device or a device cluster with computing power capability
  • the communication interface is used to transfer the target computing power to
  • the device information of the power device is sent to the terminal, and the terminal sends the computing power task to the target computing power device based on the device information, and the target computing power device executes the computing power task.
  • a target computing power device in an eleventh aspect, includes a processor, a memory, and a program or instruction stored on the memory and operable on the processor, the program or instruction When executed by the processor, the steps of the method according to the fifth aspect are realized.
  • a target computing power device including a processor and a communication interface, wherein the communication interface is used to receive a computing power task from a terminal, and the computing power task is determined by the terminal based on the target
  • the device information of the computing power device is sent, and the device information is returned to the terminal by the network side device after receiving the computing power request from the terminal and determining the target computing power device, and the computing power request is used for all
  • the computing power task in the terminal requests computing power
  • the target computing power device includes a device or device cluster with computing power capability; the processor is used to execute the computing power task.
  • a thirteenth aspect provides a readable storage medium, on which a program or instruction is stored, and when the program or instruction is executed by a processor, the steps of the method as described in the first aspect are implemented, or The steps of the method described in the third aspect, or implementing the steps of the method described in the fifth aspect.
  • a chip in a fourteenth aspect, there is provided a chip, the chip includes a processor and a communication interface, the communication interface is coupled to the processor, the processor is used to run programs or instructions, and implement the method described in the first aspect method, or realize the method as described in the third aspect, or realize the steps of the method as described in the fifth aspect.
  • a computer program/program product is provided, the computer program/program product is stored in a non-volatile storage medium, and the program/program product is executed by at least one processor to implement the first
  • the terminal when the terminal has a computing power task, it requests computing power from the network side device, and the network side device determines the target computing power device with computing power according to the terminal computing power request, and sends the target computing power
  • the computing power of the power device is allocated to the terminal, so that the terminal can execute the computing power task with the help of the computing power of the target computing power device. Therefore, in the case of insufficient computing power of the terminal, the normal execution of the computing power task can be guaranteed.
  • FIG. 1 is a schematic diagram of a wireless communication system according to an embodiment of the present application.
  • Fig. 2 is a schematic flowchart of a computing power request method according to an embodiment of the present application
  • Fig. 3 is a schematic flow chart of a computing power distribution method according to an embodiment of the present application.
  • Fig. 4 is a schematic flow chart of a computing power execution method according to an embodiment of the present application.
  • Fig. 5 is a schematic structural diagram of a computing power request device according to an embodiment of the present application.
  • Fig. 6 is a schematic structural diagram of a computing power distribution device according to an embodiment of the present application.
  • Fig. 7 is a schematic structural diagram of a computing power execution device according to an embodiment of the present application.
  • FIG. 8 is a schematic structural diagram of a communication device according to an embodiment of the present application.
  • FIG. 9 is a schematic structural diagram of a terminal according to an embodiment of the present application.
  • Fig. 10 is a schematic structural diagram of a network side device according to an embodiment of the present application.
  • first, second and the like in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific sequence or sequence. It is to be understood that the terms so used are interchangeable under appropriate circumstances such that the embodiments of the application are capable of operation in sequences other than those illustrated or described herein and that "first" and “second” distinguish objects. It is usually one category, and the number of objects is not limited. For example, there may be one or more first objects.
  • “and/or” in the description and claims means at least one of the connected objects, and the character “/” generally means that the related objects are an "or” relationship.
  • LTE Long Term Evolution
  • LTE-Advanced LTE-Advanced
  • LTE-A Long Term Evolution-Advanced
  • CDMA Code Division Multiple Access
  • TDMA Time Division Multiple Access
  • FDMA Frequency Division Multiple Access
  • OFDMA Orthogonal Frequency Division Multiple Access
  • SC-FDMA Single-carrier Frequency-Division Multiple Access
  • system and “network” in the embodiments of the present application are often used interchangeably, and the described technology can be used for the above-mentioned system and radio technology, and can also be used for other systems and radio technologies.
  • the following description describes the New Radio (New Radio, NR) system for illustrative purposes, and uses NR terminology in most of the following descriptions, but these techniques can also be applied to applications other than NR system applications, such as the 6th generation (6 th Generation , 6G) communication system.
  • 6th generation 6 th Generation
  • Fig. 1 shows a block diagram of a wireless communication system to which the embodiment of the present application is applicable.
  • the wireless communication system includes a terminal 11 , a network side device 12 and a computing power device 13 .
  • the terminal 11 can also be called a terminal device or a user terminal (User Equipment, UE), and the terminal 11 can be a mobile phone, a tablet computer (Tablet Personal Computer), a laptop computer (Laptop Computer) or a notebook computer, a personal digital Assistant (Personal Digital Assistant, PDA), handheld computer, netbook, ultra-mobile personal computer (UMPC), mobile Internet device (Mobile Internet Device, MID), wearable device (Wearable Device) or vehicle-mounted device (VUE), Pedestrian Terminal (PUE) and other terminal-side devices, wearable devices include: smart watches, bracelets, earphones, glasses, etc.
  • the network side device 12 can be understood as a computing power resource management platform or a computing power management network element, which can be a 5G core network element, such as: session management function (Session Management Function, SMF), access management function (Access Management Function, AMF), network data analysis function (Network Data Analytics Function, NWDAF), computing power control function (Computing Resource Control Function, CSCF), etc., can be a base station or a core network, where the base station can be called a node B, an evolved node B , access point, base transceiver station (Base Transceiver Station, BTS), radio base station, radio transceiver, basic service set (Basic Service Set, BSS), extended service set (Extended Service Set, ESS), B node, evolution eNB, Home NodeB, Home Evolved NodeB, WLAN access point, WiFi node, TransmittingReceiving Point (TRP) or some other 5G core network element, such as: session management function (Session Management Function, SMF),
  • the computing device 13 may be a device or a device cluster with computing power, and may be a terminal, a network element of a core network, a server, an edge cloud, a central cloud, and the like.
  • the embodiment of the present application provides a computing power request method 200, which can be executed by a terminal, in other words, the method can be executed by software or hardware installed in the terminal, and the method includes the following steps.
  • the terminal sends a computing power request to the network side device, where the computing power request is used to request computing power for a computing power task in the terminal.
  • the terminal can send a computing power request to the network side device to request computing power for the computing power task in the terminal.
  • the terminal can send a computing power request to the network-side device.
  • the "face recognition on 1 million pictures" is the computing power task in the terminal.
  • the terminal when the terminal sends a computing power request to the network side device, it can be implemented in at least one of the following ways:
  • the first way the terminal sends the computing power demand information of the computing power task to the network side device.
  • the computing power requirement information is used by the network side device to determine the target computing power device that matches the computing power requirement.
  • the computing power requirement information may include at least one of the following: the total amount of computing power required by the computing power task, the computing power required by the computing power task The type, expected completion time of the computing power task, and the algorithm model information required for the computing power task.
  • the total amount of computing power required by the computing power task can be how many CPUs, how many GPUs, or how many servers are required for the computing power task
  • the type of computing power required by the computing power task can be what kind of computing power task requires Computing power, such as the need for CPU, GPU, FPGA or server, etc.
  • the expected completion time of the computing power task can be how much time is expected to complete the computing power task, for example, it can be 1 hour, 2 hours, etc.
  • the algorithm model information required by the computing power task may be the information of the model specifically required by the computing power task, for example, it may be the specific name of the algorithm model, the detailed parameters required by the algorithm model, and the like.
  • the second method the terminal sends computing power task description information to the network side device, and the computing power task description information is used to determine the above computing power requirement information.
  • the computing power task description information can be used by network side devices to analyze and obtain computing power requirement information.
  • the computing power task description information may include at least one of the following: the type of computing power task, the type of algorithm model required by the computing power task, the description of the task amount of the computing power task, the expected completion time range of the computing power task, and the expected completion time of the computing power task Completion success rate.
  • the computing power task type can include at least one of the following: artificial intelligence (AI) business, data processing type business, image processing type, audio processing type business, and the type of algorithm model can be the algorithm model required to perform the computing power task
  • AI artificial intelligence
  • Each type of algorithm model can include multiple algorithm models, and the network-side device can determine which algorithm model is specifically required for the computing power task based on the type of algorithm model, and then determine the corresponding algorithm model information.
  • the task amount description of the computing power task may be a detailed description of the computing power task amount.
  • the expected completion time range of the computing power task can be the time range within which the computing power task is expected to be executed, for example, it can be 1-2 hours, 2-3 hours, etc.
  • the network side device can be based on the expected completion time of the computing power task
  • the scope determines the expected completion time of the computing power task.
  • the expected completion success rate of the computing power task can be the success rate of the expected computing power task being executed after the computing power task is executed. For example, face recognition is performed on 1 million pictures, and the expected recognition success rate is above 99%.
  • the computing power task description information can be: the terminal needs a CPU, the AI model of face recognition needs to be used, and the face recognition of 1 million pictures is required.
  • the CPU is the type of computing power task, and the face recognition type
  • the AI model is the algorithm model required for the computing power task, and the face recognition of 1 million pictures is the task description of the computing power task.
  • the computing power task in the terminal may be a computing power task of a target application installed in the terminal, or may be a computing power task of the terminal itself (that is, a computing power task of a non-target application).
  • the computing power task of the target application may be, for example, the computing power task that the target application for video processing needs to process several videos.
  • the target application in the terminal can send the computing power request to the terminal. That is to say, when the target application in the terminal has a computing power requirement, the target application can send a computing power request to the terminal, and then the terminal sends the computing power request to the network side device.
  • the terminal described here may specifically refer to a terminal (Operating System, OS) operating system sending.
  • the target application when the target application sends a computing power request to the terminal, it can also be implemented in any of the above two ways, that is, the target application can send at least one of computing power demand information and computing power task description information to the terminal one item.
  • the target application can send at least one of computing power demand information and computing power task description information to the terminal one item.
  • the target application sends the computing power task description information to the terminal
  • the terminal when the terminal sends a computing power request to the network side device, it can directly send the computing power task description information to the network side device, or it can first update the computing power
  • the task description information is analyzed to obtain the computing power demand information, and then the computing power demand information is sent to the network side device.
  • the terminal sends a computing power request to the network side device, which may include at least one of the following four situations:
  • the first case the terminal itself has a computing power task, and the terminal sends the computing power demand information of the computing power task to the network side device;
  • the second case the terminal itself has a computing power task, and the terminal sends computing power task description information to the network side device;
  • the third case the target application in the terminal has a computing power task, the target application sends the computing power demand information of the computing power task to the terminal, and the terminal sends the computing power demand information to the network side device;
  • Case 4 The target application in the terminal has a computing power task, and the target application sends computing power task description information to the terminal, and the terminal sends the computing power task description information to the network side device, or the terminal determines the computing power based on the computing power task description information. Power demand information, and send the computing power demand information to the network side device.
  • the network-side device may receive at least one of computing power requirement information and computing power task description information from the terminal.
  • S204 Receive the device information of the target computing power device returned by the network side device, where the target computing power device includes a device or a device cluster with computing power capability.
  • the network side device After receiving the computing power request from the terminal, the network side device can determine the target computing power device from multiple computing power devices based on the computing power request, that is, find the target computing power device that can match the computing power requirement based on the computing power request, and then set the The device information of the target computing power device is sent to the terminal.
  • the multiple computing power devices and the target computing power device are devices or device clusters with computing power capabilities.
  • the target computing power device can be used to perform computing power tasks of the terminal. For the specific implementation manner of determining the target computing power device by the network side device, refer to the embodiment of the computing power distribution method shown in FIG. 3 , which will not be repeated here.
  • S206 Send the computing power task to the target computing power device based on the device information, and the target computing power device executes the computing power task.
  • the terminal After receiving the device information of the target computing power device sent by the network side device, the terminal can locate the target computing power device based on the device information, and send the computing power task to the target computing power device, so that the target computing power device can execute the computing power. power task.
  • the device information of the target computing power device may specifically include at least one of the following: device address information and device identifier.
  • the device address information may specifically be an Internet Protocol (Internet Protocol, IP) address, a Uniform Resource Locator (Uniform Resource Locator, URL) address, a Fully Qualified Domain Name (Fully Qualified Domain Name, FQDN), etc.
  • the device identifier may be a target The equipment identification (ID) of the computing power equipment, etc.
  • the terminal When the terminal sends a computing power task to the target computing power device, first, it can establish a connection with the target computing power device based on at least one of the device address information and the device identifier, such as an IP connection, etc., if the connection is successfully established , the terminal can send the computing power task to the target computing power device.
  • the terminal can send computing power data and algorithms when sending computing power tasks. For example, if the computing power task is to perform face recognition on 1 million pictures, then the terminal can send 1 million pictures (computing power data) and the corresponding face recognition algorithm to the target computing power device.
  • the number of target computing power devices may be one or more.
  • the target computing power device can independently execute the computing power task of the terminal, or the target computing power device can jointly perform computing power tasks with other computing power devices. That is to say, when there is only one target computing power device, the terminal can send the computing power task to this target computing power device, and the target computing power device can execute the computing power alone after receiving the computing power task.
  • Computing power tasks can also be split and assigned to other computing power tasks, so as to jointly perform computing power tasks with other computing power equipment. Among them, how to split the computing power tasks can be determined according to actual needs, and there is no specific limitation here.
  • the other computing power devices mentioned above can be determined by the network side device, and the specific implementation method can refer to the embodiment of the computing power distribution method shown in FIG. 3 , which will not be described in detail here.
  • the network side device determines the target computing power device and other computing power devices, it can send the device information of the target computing power device to the terminal, and send the device information of other computing power devices to the target computing power device, so that the terminal can The device information of the device sends the computing power task to the target computing power device.
  • the target computing power device receives the computing power task and splits the computing power task, it can split the computing power based on the device information of other computing power devices.
  • Tasks are assigned to other computing devices.
  • the target computing power device here can be called the main device, and other computing power devices can be called auxiliary devices. Participate in the execution of computing power tasks of the terminal, and only execute the computing power tasks assigned to him by the master device.
  • the terminal when the terminal sends a computing power task, it can first split the computing power task to obtain multiple computing power subtasks, and then send the multiple computing power subtasks to multiple target computing device.
  • different target computing power devices are used to perform different computing power subtasks.
  • the multiple target computing power devices here can be understood as a collection of the above-mentioned primary devices and secondary devices. That is to say, when the network side device determines multiple target computing power
  • the device information of multiple target computing power devices can also be sent to the terminal, so that the terminal can split the computing power task and distribute it to multiple The target computing power device executes.
  • the terminal When the terminal has a computing power task, it can request computing power from the network side device, and the network side device determines the target computing power device with computing power capability, and distributes the computing power of the target computing power device to the terminal, so that the terminal can use Therefore, in the case of insufficient terminal computing power, the normal execution of computing power tasks can be guaranteed.
  • the terminal after the terminal sends the computing power task to the target computing power device, it can also receive the computing power task execution result returned by the target computing power device.
  • the execution result of the computing power task received by the terminal is the execution result of the computing power task by the one target computing power device. If the number of the target computing power device is one and the target computing power device and other computing power devices jointly execute the computing power task, the execution result of the computing power task received by the terminal is the calculation result of the target computing power device and other computing power devices.
  • the joint execution results of power tasks in which the execution results of the target computing power device can be directly sent to the terminal, and the execution results of other computing power devices can be sent to the target computing power device first, and then sent to the terminal by the target computing power device. If there are multiple target computing power devices, the computing power task execution result received by the terminal is the execution result of the computing power task by the multiple target computing power devices.
  • the terminal After receiving the execution result of the computing power task, the terminal can judge whether the execution result of the computing power task meets the preset conditions or determine whether the computing power task is successfully executed based on the execution result of the computing power task.
  • meeting the preset conditions can mean that the execution result of the computing power task meets the preset execution effect, for example, the accuracy of the execution result is higher than the set threshold, the execution time of the computing power task meets the requirements, etc.
  • the successful execution of the computing power task can be The computing power task was successfully executed without considering the execution effect.
  • the terminal can send a notification message to the network side device, which is used to tell the network side device the computing power task.
  • the execution results meet expectations, and notify the network side device to release the computing power reserved for the computing power task of the terminal, that is, release the computing power reserved by the target computing power device for the computing power task, otherwise, the target computing power device can continue to Execute the computing power task until the execution result of the computing power task meets the preset conditions or it is determined that the computing power task is successfully executed based on the execution result of the computing power task.
  • the target computing power Since the computing power reserved for the computing power task can be released by the network side device when the execution result of the computing power task of the terminal meets the preset conditions or the computing power task is successfully executed, on the one hand, the target computing power will not be occupied for a long time.
  • the computing power of the power device affects the normal use of the target computing power device. On the other hand, it can facilitate the target computing power device to reserve computing power for other computing power tasks after releasing the computing power, making the use of computing power more flexible.
  • the embodiment of the present application provides a computing power distribution method 300, which can be executed by the network side device, in other words, the method can be executed by software or hardware installed on the network side device, the method includes the following step.
  • the network side device receives a computing power request from the terminal, where the computing power request is used to request computing power for a computing power task in the terminal.
  • the terminal can send a computing power request to the network side device to request computing power for the computing power task in the terminal, and the network side device can receive the computing power request from the terminal.
  • the specific implementation manner of sending the computing power request by the terminal can refer to the embodiment of the computing power request method shown in FIG. 2 , which will not be described repeatedly.
  • the computing power request received by the network side device from the terminal may include at least one of the following:
  • the computing power demand information includes at least one of the following: the total amount of computing power required by the computing power task, the type of computing power required by the computing power task, and the expected completion of the computing power task time;
  • the computing power task description information is used to determine the computing power demand information, and the computing power task description information includes at least one of the following: computing power task type, algorithm model required by the computing power task, computing power Amount description of the task.
  • the computing power request received by the network side device may specifically be at least one of computing power requirement information and computing power task description information.
  • computing power requirement information and computing power task description information please refer to The embodiment shown in FIG. 2 will not be repeated here.
  • S304 Determine a target computing power device based on the computing power request, where the target computing power device includes a device or a device cluster with computing power capability.
  • the network side device can determine the target computing power device from multiple computing power devices based on the computing power request, that is, find the target computing power device that can match the computing power demand based on the computing power request, and Computing resources that match the computing power requirements are reserved in the power equipment.
  • the network side device When the network side device determines the target computing power device based on the computing power request, it may specifically determine the target computing power device based on the computing power demand information of the computing power task. Wherein, if the computing power request received by the network side device includes computing power requirement information, the target computing power device may be determined directly based on the computing power requirement information of the computing power task.
  • the computing power task description information can be analyzed first (it can be executed by the network side device, or by an additional computing power Power task description information analysis network element), this analysis process is mainly based on the computing power task description information, after analysis and judgment, to obtain the computing power demand information corresponding to the computing power task, and then determine the target computing power based on the computing power demand information. power equipment.
  • the computing power status information of the computing power device may include at least one of the following: total computing power, computing power usage, computing power type, and available computing power.
  • the computing power state information of multiple computing power devices can be registered in the network side device by multiple computing power devices before establishing a protocol data unit (Protocol Data Unit, PDU) session with the network side device. The information is carried in the registration request and reported to the network side device, or, after registration in the network side device, when the PDU session is established with the network side device, the computing power status information is reported to the network side device, or it can also be Report to the network-side device through other methods, which will not be described here one by one.
  • PDU Protocol Data Unit
  • the network side device After receiving the computing power status information of multiple computing power devices, the network side device can store the computing power information. The computing power status information of the power device.
  • S3042 Match the computing power demand information with the computing power state information of multiple computing power devices, and determine at least one matching candidate computing power device.
  • the computing power state information of multiple computing power devices can be matched with the computing power demand information of the computing power task, and one or more matching computing power devices can be used as candidate computing power according to the matching results equipment.
  • the match here can be at least one of the following: the computing power type of the computing power device includes the computing power type required by the computing power task, the available computing power of the computing power device is greater than or equal to the total computing power required by the computing power task, The task execution efficiency of the computing power device meets or exceeds the requirements of the computing power task, and the distance between the computing power device and the terminal is less than the preset distance (meeting the delay requirement).
  • the matching may also refer to matching other information of the computing power device with the terminal or the computing power task of the terminal, which can be set according to the actual situation, and is not specifically limited here.
  • S3043 Determine a target computing power device based on at least one candidate computing power device.
  • a computing power reservation request can be sent to the candidate computing power device.
  • the computing power reservation request is used to request the candidate computing power device to reserve a matching computing power manpower resources.
  • the candidate computing power device After the candidate computing power device receives the computing power reservation request, it can reserve the available computing power or the total computing power that matches the computing power resources in the computing power reservation request. If the reservation is successful, it can return the reserved If the reservation fails, it can return the failure information of the reserved computing power or return no information.
  • the candidate computing power device receives the reserved computing power success information, it can determine the candidate computing power device as the target computing power device, and thus obtain a target computing power device.
  • the receiving of the reserved computing power success message is taken as an example for illustration.
  • one or more candidate computing power devices may be selected as the target computing power device, and the specific implementation method is as follows.
  • the network-side device can send computing power reservation requests to multiple candidate computing power devices. For any candidate computing power Determined as the target computing power device, if the candidate computing power device has not received the reserved computing power success information or received the reserved computing power failure information returned by the candidate computing power device, then the candidate computing power device may not be determined as the target computing power device. power equipment. Thus, one or more target computing power devices can be determined.
  • the network side device may also send a computing power reservation request to at least one candidate computing power device based on the descending order of computing power device priorities, and then transfer the first candidate computing power among at least one candidate computing power device to The device is determined as the target computing power device, and the first candidate computing power device is the computing power device that receives the computing power reservation request and returns the success information of the computing power reserved. That is to say, the network-side device can first send a computing power reservation request to the candidate computing power device with the highest priority.
  • the target computing power device If it is determined as the target computing power device, if it does not receive the success information of the reserved computing power of the candidate computing power device or receives the reserved computing power failure information returned by the candidate computing power device, it can send the candidate computing power of the second priority
  • the power device sends computing power reservation information, ..., and so on, until receiving the success information of one of the candidate computing power devices, and determining the candidate computing power device as the target computing power device.
  • computing power reservation information may no longer be sent to other candidate computing power devices, thereby determining a target computing power device, or continuing to send other candidate computing power devices Send computing power reservation information, so that multiple target computing power devices can be determined.
  • the priority of the above-mentioned computing power equipment is determined based on at least one of the following: the total computing power of the computing power device, the matching degree between the total computing power and the computing power task, and the routing distance between the computing power device and the terminal.
  • the greater the total computing power of the computing power equipment, the higher the priority of the computing power equipment, the higher the matching degree between the total computing power and the computing power task, the higher the priority of the computing power equipment, the higher the priority of the computing power equipment, the higher the priority of the computing power equipment and the terminal The closer the routing distance between (that is, the smaller the delay), the higher the priority of computing power equipment.
  • the matching degree between the total amount of computing power and the computing power task can represent the execution effect of the computing power task (such as the completion time of the computing power task, the accuracy of the execution result, etc.), the matching degree of the computing power type, etc.
  • S306 Send the device information of the target computing power device to the terminal, and the terminal sends the computing power task to the target computing power device based on the device information, and the target computing power device executes the computing power task.
  • the network side device After the network side device determines the target computing power device, it can send the device information of the target computing power device to the terminal.
  • the device information of the target computing power device may be that the target computing power device reports the computing power status information to the network side device at the same time, and the device information may include at least one of the following: device address information, device identification symbol.
  • the network side device when the network side device sends the device information of the target computing power device, if the number of the target computing power device is one, it can directly send the device information of the target computing power device to the terminal.
  • the device information of the target computing power device can be sent in the following two ways.
  • the first method Send the device information of multiple target computing power devices to the terminal.
  • the terminal can split the computing power task into multiple computing power subtasks, and based on the device information of multiple target computing power devices, the terminal Send multiple computing power subtasks to multiple target computing power devices respectively, and different target computing power devices execute different computing power subtasks.
  • This method depends on that the terminal may split the computing power task into computing power sub-tasks, and then send them to different target computing power devices.
  • the second method determine the main device and the auxiliary device among multiple target computing power devices, send the device information of the main device to the terminal, and send the device information of the auxiliary device to the main device.
  • the main device can be a target computing power device among multiple target computing power devices, and the auxiliary device can be other target computing power devices other than the main device among multiple target computing power devices.
  • the main device is used to receive the computing power tasks of the terminal And split the computing power task, and then assign it to the auxiliary equipment to perform the computing power task together with the auxiliary equipment.
  • the auxiliary equipment is used to participate in the execution of the computing power task, and only executes the part of the computing power task assigned to it by the main device.
  • the network side device can determine which computing power device among multiple target computing power devices is the main device according to the actual situation, and the remaining computing power devices are the auxiliary devices.
  • the device information of the master device can be sent to the terminal, and the device information of the slave device can be sent to the master device, so that the terminal can send the computing power task to the master device based on the device information of the master device.
  • the device After the device receives the computing power task, it splits the computing power task and assigns the split computing power task to the auxiliary device based on the device information of the auxiliary device to jointly perform the computing power task.
  • the terminal can send the computing power task to the target computing power device based on the received device information, so that the computing power device can execute the computing power
  • the computing power request method shown in FIG. 2 please refer to the embodiment of the computing power request method shown in FIG. 2 , which will not be repeated here.
  • the terminal When the terminal has a computing power task, it can request computing power from the network side device, and the network side device determines the target computing power device with computing power capability, and distributes the computing power of the target computing power device to the terminal, so that the terminal can use Therefore, in the case of insufficient terminal computing power, the normal execution of computing power tasks can be guaranteed.
  • the network side device may also release the computing power reserved by the target computing power device for the computing power task of the terminal when receiving the notification information sent by the terminal.
  • the specific implementation is as follows.
  • the target computing power device executes the computing power task and sends the execution result of the computing power task to the terminal.
  • the execution result of the power task determines whether the computing power task is successfully executed. If the execution result of the computing power task meets the preset conditions or the execution of the computing power task is determined to be successful based on the execution result of the computing power task, the terminal will send a notification message to the network-side device. After receiving the notification message, the network-side device can trigger as the terminal The resources of the target computing power device that executes the computing power task are released, and a computing power release request is sent to the target computing power device. After receiving the computing power release request, the target computing power device can release the computing power reserved for the terminal computing power task. force.
  • the network side device can release the computing power reserved by the target computing power device for the computing power task. Therefore, on the one hand, there will be no long-term Occupying the computing power of the target computing power device will affect the normal use of the target computing power device. On the other hand, it can facilitate the target computing power device to reserve computing power for other computing power tasks after releasing the computing power, making the use of computing power more efficient. to be flexible.
  • the embodiment of the present application provides a computing power execution method 400, which can be executed by the target computing power device, in other words, the method can be executed by software or hardware installed on the target computing power device, the method Including the following steps.
  • the target computing power device receives the computing power task from the terminal, the computing power task is sent by the terminal based on the device information of the target computing power device, and the device information is sent by the network side device after receiving the computing power request from the terminal and determining the target computing power device Returning to the terminal, the computing power request is used to request computing power for the computing power task in the terminal, and the target computing power device includes a device or a device cluster with computing power capability.
  • the terminal can send a computing power request to the network side device to request computing power for the computing power task in the terminal.
  • the network side device can determine the target computing power device based on the computing power request, and send the device information of the target computing power device to the terminal.
  • the terminal can send the computing power task to the target computing power device based on the device information, and the target computing power device can receive the computing power task from the terminal.
  • the specific implementation of the terminal sending a computing power request to the network-side device and sending a computing power task to the target computing power device can refer to the specific implementation of the corresponding steps in the embodiment shown in Figure 2.
  • the specific implementation of the network-side device to determine the target computing power device For the implementation manner, reference may be made to the specific implementation of corresponding steps in the embodiment shown in FIG. 3 , which will not be repeated here.
  • the number of target computing power devices determined by the network side device can be one or more.
  • the execution subject of the method shown in Figure 4 is For one target computing power device, when there are multiple target computing power devices, if the network side device sends all the device information of the multiple target computing power devices to the terminal, the execution of the method shown in Figure 4 The subject is one of the target computing power devices. If the network side device only sends the device information of one of the multiple target computing power devices (ie, the main device) to the terminal, then the execution subject of the method shown in Figure 4 as the master device.
  • the target computing power device After receiving the computing power task, the target computing power device can execute the computing power task.
  • this target computing power device can independently execute the computing power tasks. If the number of target computing power devices determined by the network side device is multiple, and the device information of the multiple target computing power devices is sent to the terminal, then when executing the computing power task, for each target computing power device, It can execute the computing power subtask assigned to it by the terminal.
  • the computing power task can be split and assigned to other computing power equipment (ie, auxiliary equipment) to jointly perform the computing power task.
  • the terminal When the terminal has a computing power task, it can request computing power from the network side device, and the network side device determines the target computing power device with computing power capability, and distributes the computing power of the target computing power device to the terminal, so that the terminal can use Therefore, in the case of insufficient terminal computing power, the normal execution of computing power tasks can be guaranteed.
  • the target computing power device after the target computing power device completes the computing power task, it can return the computing power task execution result to the terminal.
  • the auxiliary device can first send the execution result of the computing power task to the main device, and then the main device sends it to the terminal.
  • the terminal After receiving the execution result of the computing power task, the terminal can send notification information to the network-side device when it determines that the execution result of the computing power task meets the preset conditions or determines that the computing power task is successfully executed based on the execution result of the computing power task.
  • the device After the device receives the notification information, it can send a computing power release request to the target computing power device to request the release of the computing power reserved for the computing power task.
  • the target computing power device can release the computing power reserved for the computing power task after receiving the computing power release request
  • the target computing power Since the computing power reserved for the computing power task can be released by the network side device when the execution result of the computing power task of the terminal meets the preset conditions or the computing power task is successfully executed, on the one hand, the target computing power will not be occupied for a long time.
  • the computing power of the power device affects the normal use of the target computing power device. On the other hand, it can facilitate the target computing power device to reserve computing power for other computing power tasks after releasing the computing power, making the use of computing power more flexible.
  • the execution subject may be a computing power requesting device, or a control module in the computing power requesting device for executing the computing power requesting method.
  • the computing power requesting device implemented by the computing power requesting device is taken as an example to illustrate the computing power requesting device provided in the embodiment of the present application.
  • the computing power distribution method provided in the embodiment of the present application may be executed by a computing power distribution device, or a control module in the computing power distribution device for executing the computing power distribution method.
  • the computing power distribution method performed by the computing power distribution device is taken as an example to illustrate the computing power distribution device provided in the embodiment of the present application.
  • the execution subject may be a computing power execution device, or a control module in the computing power execution device for executing the computing power execution method.
  • the computing power implementing device is used as an example to illustrate the computing power implementing device provided in the embodiment of the present application.
  • Fig. 5 is a schematic structural diagram of a device for requesting computing power according to an embodiment of the present application, and the device may correspond to a terminal in other embodiments. As shown in FIG. 5 , the device 500 includes the following modules.
  • the first sending module 501 is configured to send a computing power request to a network side device, where the computing power request is used to request computing power for a computing power task in the terminal;
  • the receiving module 502 is configured to receive the device information of the target computing power device returned by the network side device, and the target computing power device includes a device or a device cluster with computing power capability;
  • the second sending module 503 is configured to send the computing power task to the target computing power device based on the device information, and the target computing power device executes the computing power task.
  • the first sending module 501 is also used for at least one of the following:
  • the computing power demand information is used to determine the target computing power equipment matching the computing power demand, and the computing power demand information includes at least one of the following: computing power task The total amount of computing power required, the type of computing power required for the computing power task, the expected completion time of the computing power task, and the algorithm model information required for the computing power task;
  • the computing power task description information is used to determine the computing power demand information
  • the computing power task description information includes at least one of the following: computing power task type, computing power task The type of the required algorithm model, the description of the task amount of the computing power task, the expected completion time range of the computing power task, and the expected completion success rate of the computing power task.
  • the first sending module 501 is further configured to: send the computing power request to the terminal;
  • the computing power request sent by the target application includes the computing power task description information
  • the computing power request sent by the terminal includes the computing power demand information
  • the computing power request is determined based on the computing power task description information. Describe computing power demand information
  • the device information includes at least one of the following: device address information, device identifier; wherein, the second sending module 503 is further configured to:
  • the computing power task is sent to the target computing power device.
  • the number of target computing devices is one or more;
  • the target computing power device when the number of the target computing power device is one, the target computing power device is used to independently execute the computing power task, or the target computing power device is used to execute the computing power task After splitting, it is allocated to other computing power devices to jointly perform the computing power task, and the other computing power devices are determined by the network side device;
  • the second sending module 503 is further configured to:
  • the multiple computing power subtasks are sent to multiple target computing power devices, and different target computing power devices are used to execute different computing power subtasks.
  • the receiving module 502 is also configured to receive the computing power task execution result returned by the target computing power device;
  • the first sending module 501 is further configured to send the network side
  • the device sends notification information, where the notification information is used by the network side device to release the computing power reserved for the computing power task.
  • the device 500 according to the embodiment of the present application can refer to the process of the method 200 corresponding to the embodiment of the present application, and each unit/module in the device 500 and the above-mentioned other operations and/or functions are respectively in order to realize the corresponding process in the method 200, And can achieve the same or equivalent technical effect, for the sake of brevity, no more details are given here.
  • Fig. 6 is a schematic structural diagram of a device for distributing computing power according to an embodiment of the present application, and the device may correspond to network-side devices in other embodiments. As shown in FIG. 6 , the device 600 includes the following modules.
  • a receiving module 601 configured to receive a computing power request from a terminal, where the computing power request is used to request computing power for a computing power task in the terminal;
  • a determining module 602 configured to determine a target computing power device based on the computing power request, where the target computing power device includes a device or a device cluster with computing power capability;
  • the sending module 603 is configured to send the device information of the target computing power device to the terminal, and the terminal sends the computing power task to the target computing power device based on the device information, and the The target computing power device executes the computing power task.
  • the receiving module 601 is also used for at least one of the following:
  • Computing power requirement information of the computing power task is used to determine a target computing power device matching the computing power requirement, and the computing power requirement information includes at least one of the following: computing power The total amount of computing power required for the task, the type of computing power required for the computing power task, the expected completion time of the computing power task, and the algorithm model information required for the computing power task;
  • the computing power task description information is used to determine the computing power demand information, and the computing power task description information includes at least one of the following: computing power task type, computing power task The type of algorithm model required, the task amount description of the computing power task, the expected completion time range of the computing power task, and the expected completion success rate of the computing power task.
  • the determining module 602 is also used for at least one of the following:
  • the computing power request includes the computing power requirement information, determine the target computing power device based on the computing power requirement information;
  • the computing power requirement information is determined based on the computing power task description information; and the target computing power device is determined based on the computing power requirement information.
  • the determining module 602 is further configured to:
  • the computing power status information includes at least one of the following: total computing power, computing power usage, computing power type, and available computing power;
  • the target computing power device is determined based on the at least one candidate computing power device.
  • the determining module 602 is further configured to:
  • the first candidate computing power device is a computing power device that receives the computing power reservation request and returns the reserved computing power success information; the priority of the computing power device is based on at least one of the following Determine: the total amount of computing power of the computing power device, the matching degree between the total computing power and the computing power task, and the routing distance between the computing power device and the terminal.
  • the number of target computing devices is one or more;
  • the sending module 603 is further configured to:
  • determining the primary device and the secondary device among the plurality of target computing power devices sending the device information of the primary device to the terminal, and sending the device information of the secondary device to the primary device, and the The terminal sends the computing task to the primary device based on the device information of the primary device, and the primary device splits the computing task and assigns it to the secondary device based on the device information of the secondary device.
  • devices to jointly perform the computing tasks to jointly perform the computing tasks.
  • the receiving module 601 is further configured to receive notification information from the terminal, the notification information is received by the terminal from the computing power task execution result returned by the target computing power device Afterwards, when it is determined that the execution result of the computing power task satisfies the preset condition or it is determined that the execution of the computing power task is successful based on the execution result of the computing power task;
  • the sending module 603 is further configured to send a computing power release request to the target computing power device, so as to release the computing power reserved by the target computing power device for the computing power task.
  • the device 600 according to the embodiment of the present application can refer to the process of the method 300 corresponding to the embodiment of the present application, and each unit/module in the device 600 and the above-mentioned other operations and/or functions are respectively in order to realize the corresponding process in the method 300, And can achieve the same or equivalent technical effect, for the sake of brevity, no more details are given here.
  • Fig. 7 is a schematic structural diagram of a computing power execution device according to an embodiment of the present application, which may correspond to the target computing power device in other embodiments. As shown in FIG. 7 , the device 700 includes the following modules.
  • the receiving module 701 is configured to receive a computing power task from a terminal, the computing power task is sent by the terminal based on the device information of the target computing power device, and the device information is received by the network side device after receiving the terminal's computing power request and determine the target computing power device and return it to the terminal, the computing power request is used to request computing power for the computing power task in the terminal, the target computing power device includes a device or a cluster of devices;
  • the number of target computing devices is one or more, wherein:
  • the task execution module 702 is further configured to:
  • the computing power task is split and distributed to other computing power devices to jointly perform the computing power task, the other computing power devices are determined by the network side device, and the other computing power devices complete the computing power task After that, the execution result of the computing power task is sent to the target computing power device;
  • the task execution module 702 is further configured to:
  • the task execution module 702 is also configured to send the computing power task execution result to the terminal;
  • the receiving module 701 is further configured to receive a computing power release request from the network side device, the computing power release request is sent by the network side device after receiving notification information from the terminal, and the notification information is sent by After receiving the execution result of the computing power task, the terminal sends the result to the terminal when it is determined that the execution result of the computing power task satisfies a preset condition or determines that the execution of the computing power task is successful based on the execution result of the computing power task.
  • the network side device release the computing power reserved for the computing power task based on the computing power release request.
  • the device 700 according to the embodiment of the present application can refer to the process of the method 400 corresponding to the embodiment of the present application, and each unit/module in the device 700 and the above-mentioned other operations and/or functions are respectively in order to realize the corresponding process in the method 400, And can achieve the same or equivalent technical effect, for the sake of brevity, no more details are given here.
  • the computing power requesting device, computing power distributing device, and computing power executing device in the embodiments of this application may be devices, devices with operating systems or electronic equipment, or components, integrated circuits, or chips in terminals.
  • the apparatus or electronic equipment may be a mobile terminal or a non-mobile terminal.
  • the mobile terminal may include but not limited to the types of terminals 11 listed above, and the non-mobile terminal may be a server, a network attached storage (Network Attached Storage, NAS), a personal computer (personal computer, PC), a television ( television, TV), teller machines or self-service machines, etc., are not specifically limited in this embodiment of the present application.
  • the computing power requesting device, computing power distributing device, and computing power executing device provided in the embodiments of the present application can realize the various processes realized by the method embodiments in Fig. 2 to Fig. 4 respectively, and achieve the same technical effect. In order to avoid repetition, it is not repeated here Let me repeat.
  • this embodiment of the present application further provides a communication device 800, including a processor 801, a memory 802, and programs or instructions stored in the memory 802 and operable on the processor 801,
  • a communication device 800 including a processor 801, a memory 802, and programs or instructions stored in the memory 802 and operable on the processor 801
  • the communication device 800 is a terminal
  • the program or instruction is executed by the processor 801
  • each process of the above computing power request method embodiment can be realized, and the same technical effect can be achieved.
  • the communication device 800 is a network-side device
  • the program or instruction is executed by the processor 801
  • each process of the above computing power distribution method embodiment can be achieved, and the same technical effect can be achieved.
  • the communication device 800 is the target computing power device, when the program or instruction is executed by the processor 801, the various processes of the above computing power execution method embodiments can be achieved, and the same technical effect can be achieved. To avoid repetition, details are not repeated here.
  • the embodiment of the present application also provides a terminal, including a processor and a communication interface, the communication interface is used to send a computing power request to a network side device, and the computing power request is used to request computing power for a computing power task in the terminal; Receive the device information of the target computing power device returned by the network side device, the target computing power device includes a device or device cluster with computing power capability; send the computing power task to the target computing power based on the device information power device, and the target computing power device executes the computing power task.
  • This terminal embodiment corresponds to the above-mentioned terminal-side method embodiment, and each implementation process and implementation mode of the above-mentioned method embodiment can be applied to this terminal embodiment, and can achieve the same technical effect.
  • FIG. 9 is a schematic diagram of a hardware structure of a terminal implementing an embodiment of the present application.
  • the terminal 900 includes, but is not limited to: a radio frequency unit 901, a network module 902, an audio output unit 903, an input unit 904, a sensor 905, a display unit 906, a user input unit 907, an interface unit 908, a memory 909, and a processor 910, etc. at least some of the components.
  • the terminal 900 can also include a power supply (such as a battery) for supplying power to various components, and the power supply can be logically connected to the processor 910 through the power management system, so as to manage charging, discharging, and power consumption through the power management system. Management and other functions.
  • a power supply such as a battery
  • the terminal structure shown in FIG. 9 does not constitute a limitation on the terminal, and the terminal may include more or fewer components than shown in the figure, or combine some components, or arrange different components, which will not be repeated here.
  • the input unit 904 may include a graphics processor (Graphics Processing Unit, GPU) 9041 and a microphone 9042, and the graphics processor 9041 is used for the image capture device (such as the image data of the still picture or video obtained by the camera) for processing.
  • the display unit 906 may include a display panel 9061, and the display panel 9061 may be configured in the form of a liquid crystal display, an organic light emitting diode, or the like.
  • the user input unit 907 includes a touch panel 9071 and other input devices 9072 .
  • the touch panel 9071 is also called a touch screen.
  • the touch panel 9071 may include two parts, a touch detection device and a touch controller.
  • Other input devices 9072 may include, but are not limited to, physical keyboards, function keys (such as volume control buttons, switch buttons, etc.), trackballs, mice, and joysticks, which will not be repeated here.
  • the radio frequency unit 901 receives the downlink data from the network side device, and processes it to the processor 910; in addition, sends the uplink data to the network side device.
  • the radio frequency unit 901 includes, but is not limited to, an antenna, at least one amplifier, a transceiver, a coupler, a low noise amplifier, a duplexer, and the like.
  • the memory 909 can be used to store software programs or instructions as well as various data.
  • the memory 909 may mainly include a program or instruction storage area and a data storage area, wherein the program or instruction storage area may store an operating system, at least one application or instruction required by a function (such as a sound playback function, an image playback function, etc.) and the like.
  • the memory 909 may include a high-speed random access memory, and may also include a non-volatile memory, wherein the non-volatile memory may be a read-only memory (Read-Only Memory, ROM), a programmable read-only memory (Programmable ROM, PROM) ), erasable programmable read-only memory (ErasablePROM, EPROM), electrically erasable programmable read-only memory (Electrically EPROM, EEPROM) or flash memory.
  • ROM Read-Only Memory
  • PROM programmable read-only memory
  • ErasablePROM ErasablePROM
  • EPROM electrically erasable programmable read-only memory
  • EEPROM electrically erasable programmable read-only memory
  • flash memory for example at least one magnetic disk storage device, flash memory device, or other non-volatile solid-state storage device.
  • the processor 910 may include one or more processing units; optionally, the processor 910 may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface, application programs or instructions, etc., Modem processors mainly handle wireless communications, such as baseband processors. It can be understood that the foregoing modem processor may not be integrated into the processor 910 .
  • the radio frequency unit 910 is used to send a computing power request to the network side device, and the computing power request is used to request computing power for the computing power task in the terminal; the device receiving the target computing power device returned by the network side device information, the target computing power device includes a device or a device cluster with computing power capability; based on the device information, the computing power task is sent to the target computing power device, and the target computing power device executes the computing power power task.
  • the terminal When the terminal has a computing power task, it can request computing power from the network side device, and the network side device determines the target computing power device with computing power capability, and distributes the computing power of the target computing power device to the terminal, so that the terminal can use Therefore, in the case of insufficient terminal computing power, the normal execution of computing power tasks can be guaranteed.
  • the terminal 900 provided in the embodiment of the present application can also implement the various processes in the above embodiment of the computing power request method, and can achieve the same technical effect. To avoid repetition, details are not repeated here.
  • the embodiment of the present application also provides a network side device, including a processor and a communication interface, the communication interface is used to receive a computing power request from a terminal, and the computing power request is used to request computing power for a computing power task in the terminal
  • the processor is used to determine a target computing power device based on the computing power request, and the target computing power device includes a device or a device cluster with computing power capabilities;
  • the communication interface is also used to send the device information of the target computing power device to The terminal sends the computing power task to the target computing power device based on the device information, and the target computing power device executes the computing power task.
  • the network-side device embodiment corresponds to the above-mentioned network-side device method embodiment, and each implementation process and implementation mode of the above-mentioned method embodiment can be applied to this network-side device embodiment, and can achieve the same technical effect.
  • the embodiment of the present application also provides a network side device.
  • the network device 1000 includes: an antenna 101 , a radio frequency device 102 , and a baseband device 103 .
  • the antenna 101 is connected to the radio frequency device 102 .
  • the radio frequency device 102 receives information through the antenna 101, and sends the received information to the baseband device 103 for processing.
  • the baseband device 103 processes the information to be sent and sends it to the radio frequency device 102
  • the radio frequency device 102 processes the received information and sends it out through the antenna 101 .
  • the foregoing frequency band processing device may be located in the baseband device 103 , and the method executed by the network side device in the above embodiments may be implemented in the baseband device 103 , and the baseband device 103 includes a processor 104 and a memory 105 .
  • the baseband device 103 may include at least one baseband board, for example, a plurality of chips are arranged on the baseband board, as shown in FIG.
  • the baseband device 103 may also include a network interface 106 for exchanging information with the radio frequency device 102, such as a common public radio interface (CPRI for short).
  • a network interface 106 for exchanging information with the radio frequency device 102, such as a common public radio interface (CPRI for short).
  • CPRI common public radio interface
  • the network side device in the embodiment of the present invention also includes: instructions or programs stored in the memory 105 and operable on the processor 104, and the processor 104 calls the instructions or programs in the memory 105 to execute the modules shown in FIG. 6 To avoid duplication, the method of implementation and to achieve the same technical effect will not be repeated here.
  • the embodiment of the present application also provides a readable storage medium.
  • the readable storage medium stores programs or instructions.
  • the program or instructions are executed by the processor, the various processes in the above embodiments of the computing power request method are realized, or the above-mentioned
  • the processor is the processor in the terminal described in the foregoing embodiments.
  • the readable storage medium includes computer readable storage medium, such as computer read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disk or optical disk, etc.
  • the embodiment of the present application further provides a chip, the chip includes a processor and a communication interface, the communication interface is coupled to the processor, and the processor is used to run programs or instructions to implement the above embodiment of the computing power request method
  • the chip 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 above embodiment of the computing power request method
  • the chip mentioned in the embodiment of the present application may also be called a system-on-chip, a system-on-chip, a system-on-a-chip, or a system-on-a-chip.
  • the embodiment of the present application further provides a computer program/program product, the computer program/program product is stored in a non-volatile storage medium, and the program/program product is executed by at least one processor to realize the above computing power
  • the various processes of the embodiment of the request method, or the various processes of the above embodiment of the computing power distribution method, or the various processes of the above embodiment of the computing power execution method and can achieve the same technical effect, in order to avoid repetition, no more details are given here. .
  • the term “comprising”, “comprising” or any other variation thereof is intended to cover a non-exclusive inclusion such that a process, method, article or apparatus comprising a set of elements includes not only those elements, It also includes other elements not expressly listed, or elements inherent in the process, method, article, or device. Without further limitations, an element defined by the phrase “comprising a " does not preclude the presence of additional identical elements in the process, method, article, or apparatus comprising that element.
  • the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved. Functions are performed, for example, the described methods may be performed in an order different from that described, and various steps may also be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.
  • the methods of the above embodiments can be implemented by means of software plus a necessary general-purpose hardware platform, and of course also by hardware, but in many cases the former is better implementation.
  • the technical solution of the present application can be embodied in the form of computer software products, which are stored in a storage medium (such as ROM/RAM, magnetic disk, etc.) , CD-ROM), including several instructions to make a terminal (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) execute the methods described in the various embodiments of the present application.

Landscapes

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

Abstract

本申请公开了一种算力请求、算力分配、算力执行方法、终端及网络侧设备,属于终端通信技术领域,本申请实施例的算力请求方法包括:终端向网络侧设备发送算力请求,算力请求用于为终端中的算力任务请求算力;接收网络侧设备返回的目标算力设备的设备信息,目标算力设备包括具有算力能力的设备或设备集群;基于设备信息将算力任务发送给目标算力设备,由目标算力设备执行算力任务。

Description

算力请求、算力分配、算力执行方法、终端及网络侧设备
交叉引用
本发明要求在2021年08月18日提交中国专利局、申请号为202110950241.2、发明名称为“算力请求、算力分配、算力执行方法、终端及网络侧设备”的中国专利申请的优先权,该申请的全部内容通过引用结合在本发明中。
技术领域
本申请属于终端通信技术领域,具体涉及一种算力请求、算力分配、算力执行方法、终端及网络侧设备。
背景技术
算力可以理解为某设备的计算速度或计算能力,比如,服务器、中央处理器(Central Processing Unit,CPU)、图形处理器(Graphics Processing Unit,GPU)、终端等的计算速度或计算能力,上述设备都含有计算单元或计算能力。通常算力可以用每秒钟完成多少次计算来表示,比如每秒所执行的浮点运算次数(floating-point operations per second,FLOPS),一般用作衡量GPU的并行计算能力。再比如,哈希率是比特币网络处理能力的度量单位。即为计算机(CPU)计算哈希函数输出的速度。比特币网络必须为了安全目的而进行密集的数学和加密相关操作。例如,当网络达到10Th/s的哈希率时,意味着它可以每秒进行10万亿次计算。算力任务可以理解为需要消耗一定算力的待完成任务,比如,完成对某一组照片的人脸识别任务,将消耗算力10个GPU,算力需求可以理解为完成算力任务所需要消耗的算力资源。
通常,当终端有算力任务时,比如,终端内部有算力任务,或终端中的应用有算力任务待执行时,终端可以利用自身的算力资源在本地执行算力任务。然而,在实际应用中,终端在本地执行算力任务时,可能存在算力不足 的问题,即终端算力任务或应用算力任务的算力需求大于终端具有的算力能力,从而影响算力任务的执行效率,甚至导致算力任务执行失败。
发明内容
本申请实施例提供一种算力请求、算力分配、算力执行方法、终端及网络侧设备,能够解决终端在算力不足的情况下无法正常执行算力任务的问题。
第一方面,提供了一种算力请求方法,该方法包括:
终端向网络侧设备发送算力请求,所述算力请求用于为所述终端中的算力任务请求算力;
接收所述网络侧设备返回的目标算力设备的设备信息,所述目标算力设备包括具有算力能力的设备或设备集群;
基于所述设备信息将所述算力任务发送给所述目标算力设备,由所述目标算力设备执行所述算力任务。
第二方面,提供了一种算力请求装置,包括:
第一发送模块,用于向网络侧设备发送算力请求,所述算力请求用于为所述终端中的算力任务请求算力;
接收模块,用于接收所述网络侧设备返回的目标算力设备的设备信息,所述目标算力设备包括具有算力能力的设备或设备集群;
第二发送模块,用于基于所述设备信息将所述算力任务发送给所述目标算力设备,由所述目标算力设备执行所述算力任务。
第三方面,提供了一种算力分配方法,包括:
网络侧设备接收来自终端的算力请求,所述算力请求用于为所述终端中的算力任务请求算力;
基于所述算力请求确定目标算力设备,所述目标算力设备包括具有算力能力的设备或设备集群;
将所述目标算力设备的设备信息发送给所述终端,由所述终端基于所述 设备信息将所述算力任务发送给所述目标算力设备,并由所述目标算力设备执行所述算力任务。
第四方面,提供了一种算力分配装置,包括:
接收模块,用于接收来自终端的算力请求,所述算力请求用于为所述终端中的算力任务请求算力;
确定模块,用于基于所述算力请求确定目标算力设备,所述目标算力设备包括具有算力能力的设备或设备集群;
发送模块,用于将所述目标算力设备的设备信息发送给所述终端,由所述终端基于所述设备信息将所述算力任务发送给所述目标算力设备,并由所述目标算力设备执行所述算力任务。
第五方面,提供了一种算力执行方法,包括:
目标算力设备接收来自终端的算力任务,所述算力任务由所述终端基于所述目标算力设备的设备信息发送,所述设备信息由网络侧设备在接收到所述终端的算力请求并确定所述目标算力设备后返回给所述终端,所述算力请求用于为所述终端中的算力任务请求算力,所述目标算力设备包括具有算力能力的设备或设备集群;
执行所述算力任务。
第六方面,提供了一种算力执行装置,包括:
接收模块,用于接收来自终端的算力任务,所述算力任务由所述终端基于所述目标算力设备的设备信息发送,所述设备信息由网络侧设备在接收到所述终端的算力请求并确定所述目标算力设备后返回给所述终端,所述算力请求用于为所述终端中的算力任务请求算力,所述目标算力设备包括具有算力能力的设备或设备集群;
任务执行模块,用于执行所述算力任务。
第七方面,提供了一种终端,该终端包括处理器、存储器及存储在所述存储器上并可在所述处理器上运行的程序或指令,所述程序或指令被所述处 理器执行时实现如第一方面所述的方法的步骤。
第八方面,提供了一种终端,包括处理器及通信接口,其中,所述通信接口用于向网络侧设备发送算力请求,所述算力请求用于为所述终端中的算力任务请求算力;接收所述网络侧设备返回的目标算力设备的设备信息,所述目标算力设备包括具有算力能力的设备或设备集群;基于所述设备信息将所述算力任务发送给所述目标算力设备,由所述目标算力设备执行所述算力任务。
第九方面,提供了一种网络侧设备,该网络侧设备包括处理器、存储器及存储在所述存储器上并可在所述处理器上运行的程序或指令,所述程序或指令被所述处理器执行时实现如第三方面所述的方法的步骤。
第十方面,提供了一种网络侧设备,包括处理器及通信接口,其中,所述通信接口用于接收来自终端的算力请求,所述算力请求用于为所述终端中的算力任务请求算力,所述处理器用于基于所述算力请求确定目标算力设备,所述目标算力设备包括具有算力能力的设备或设备集群,所述通信接口用于将所述目标算力设备的设备信息发送给所述终端,由所述终端基于所述设备信息将所述算力任务发送给所述目标算力设备,并由所述目标算力设备执行所述算力任务。
第十一方面,提供了一种目标算力设备,该目标算力设备包括处理器、存储器及存储在所述存储器上并可在所述处理器上运行的程序或指令,所述程序或指令被所述处理器执行时实现如第五方面所述的方法的步骤。
第十二方面,提供了一种目标算力设备,包括处理器及通信接口,其中,所述通信接口用于接收来自终端的算力任务,所述算力任务由所述终端基于所述目标算力设备的设备信息发送,所述设备信息由网络侧设备在接收到所述终端的算力请求并确定所述目标算力设备后返回给所述终端,所述算力请求用于为所述终端中的算力任务请求算力,所述目标算力设备包括具有算力能力的设备或设备集群;所述处理器用于执行所述算力任务。
第十三方面,提供了一种可读存储介质,所述可读存储介质上存储程序或指令,所述程序或指令被处理器执行时实现如第一方面所述的方法的步骤,或者实现如第三方面所述的方法的步骤,或者实现如第五方面所述的方法的步骤。
第十四方面,提供了一种芯片,所述芯片包括处理器和通信接口,所述通信接口和所述处理器耦合,所述处理器用于运行程序或指令,实现如第一方面所述的方法,或实现如第三方面所述的方法,或者实现如第五方面所述的方法的步骤。
第十五方面,提供了一种计算机程序/程序产品,所述计算机程序/程序产品被存储在非易失的存储介质中,所述程序/程序产品被至少一个处理器执行以实现如第一方面所述的方法的步骤,或实现如第三方面所述的方法,或者实现如第五方面所述的方法的步骤。
在本申请实施例中,终端在有算力任务的情况下,通过向网络侧设备请求算力,由网络侧设备根据终端算力请求确定具有算力能力的目标算力设备,并将目标算力设备的算力分配给终端,使得终端可以借助于目标算力设备的算力执行算力任务,因此,在终端算力不足的情况下,可以保证算力任务的正常执行。
附图说明
图1是根据本申请实施例的无线通信系统的示意图;
图2是根据本申请实施例的算力请求方法的示意性流程图;
图3是根据本申请实施例的算力分配方法的示意性流程图;
图4是根据本申请实施例的算力执行方法的示意性流程图;
图5是根据本申请实施例的算力请求装置的结构示意图;
图6是根据本申请实施例的算力分配装置的结构示意图;
图7是根据本申请实施例的算力执行装置的结构示意图;
图8是根据本申请实施例的通信设备的结构示意图;
图9是根据本申请实施例的终端的结构示意图;
图10是根据本申请实施例的网络侧设备的结构示意图。
具体实施方式
下面将结合本申请实施例中的附图,对本申请实施例中的技术方案进行清楚描述,显然,所描述的实施例是本申请一部分实施例,而不是全部的实施例。基于本申请中的实施例,本领域普通技术人员所获得的所有其他实施例,都属于本申请保护的范围。
本申请的说明书和权利要求书中的术语“第一”、“第二”等是用于区别类似的对象,而不用于描述特定的顺序或先后次序。应该理解这样使用的术语在适当情况下可以互换,以便本申请的实施例能够以除了在这里图示或描述的那些以外的顺序实施,且“第一”、“第二”所区别的对象通常为一类,并不限定对象的个数,例如第一对象可以是一个,也可以是多个。此外,说明书以及权利要求中“和/或”表示所连接对象的至少其中之一,字符“/”一般表示前后关联对象是一种“或”的关系。
值得指出的是,本申请实施例所描述的技术不限于长期演进型(Long Term Evolution,LTE)/LTE的演进(LTE-Advanced,LTE-A)系统,还可用于其他无线通信系统,诸如码分多址(Code Division Multiple Access,CDMA)、时分多址(Time Division Multiple Access,TDMA)、频分多址(Frequency Division Multiple Access,FDMA)、正交频分多址(Orthogonal Frequency Division Multiple Access,OFDMA)、单载波频分多址(Single-carrier Frequency-Division Multiple Access,SC-FDMA)和其他系统。本申请实施例中的术语“系统”和“网络”常被可互换地使用,所描述的技术既可用于以上提及的系统和无线电技术,也可用于其他系统和无线电技术。以下描述出于示例目的描述了新空口(NewRadio,NR)系统,并且在以下大部分描述 中使用NR术语,但是这些技术也可应用于NR系统应用以外的应用,如第6代(6 thGeneration,6G)通信系统。
图1示出本申请实施例可应用的一种无线通信系统的框图。无线通信系统包括终端11、网络侧设备12和算力设备13。其中,终端11也可以称作终端设备或者用户终端(User Equipment,UE),终端11可以是手机、平板电脑(Tablet Personal Computer)、膝上型电脑(Laptop Computer)或称为笔记本电脑、个人数字助理(Personal Digital Assistant,PDA)、掌上电脑、上网本、超级移动个人计算机(ultra-mobile personal computer,UMPC)、移动上网装置(Mobile Internet Device,MID)、可穿戴式设备(Wearable Device)或车载设备(VUE)、行人终端(PUE)等终端侧设备,可穿戴式设备包括:智能手表、手环、耳机、眼镜等。需要说明的是,在本申请实施例并不限定终端11的具体类型。网络侧设备12可以理解为算力资源的管理平台或算力管理网元,可以是5G核心网网元,比如:会话管理功能(Session Management Function,SMF),接入管理功能(Access Management Function,AMF),网络数据分析功能(Network Data Analytics Function,NWDAF),算力控制功能(Computing Resource Control Function,CSCF)等,可以是基站或核心网,其中,基站可被称为节点B、演进节点B、接入点、基收发机站(Base Transceiver Station,BTS)、无线电基站、无线电收发机、基本服务集(Basic Service Set,BSS)、扩展服务集(Extended Service Set,ESS)、B节点、演进型B节点(eNB)、家用B节点、家用演进型B节点、WLAN接入点、WiFi节点、发送接收点(TransmittingReceivingPoint,TRP)或所述领域中其他某个合适的术语,只要达到相同的技术效果,所述基站不限于特定技术词汇,需要说明的是,在本申请实施例中仅以NR系统中的基站为例,但是并不限定基站的具体类型。算力设备13可以是具有算力能力的设备或设备集群,可以是终端、核心网的网元、服务器、边缘云、中心云等。
下面结合附图,通过一些实施例及其应用场景对本申请实施例提供的算 力请求、算力分配方法、终端及网络侧设备进行详细地说明。
如图2所示,本申请实施例提供一种算力请求方法200,该方法可以由终端执行,换言之,该方法可以由安装在终端的软件或硬件来执行,该方法包括如下步骤。
S202:终端向网络侧设备发送算力请求,算力请求用于为终端中的算力任务请求算力。
在终端中有算力任务的情况下,终端可以向网络侧设备发送算力请求,以为终端中的算力任务请求算力。比如,当终端需要对100万张图片进行人脸识别时,终端可以向网络侧设备发送算力请求,该“对100万张图片进行人脸识别”即为终端中的算力任务。
可选的,终端在向网络侧设备发送算力请求时,可以通过以下至少一种方式实现:
第一种方式:终端向网络侧设备发送算力任务的算力需求信息。
也就是说,终端在发送算力请求时,具体是发送算力需求信息。算力需求信息用于网络侧设备确定匹配算力需求的目标算力设备,算力需求信息可以包括以下至少一项:算力任务所需的算力总量、算力任务所需的算力类型、算力任务的期望完成时间、算力任务所需的算法模型信息。其中,算力任务所需的算力总量可以是算力任务总共需要多少个CPU,多少个GPU,或者多少服务器等,算力任务所需的算力类型可以是算力任务需要什么样的算力,比如需要CPU,GPU,FPGA还是服务器等,算力任务的期望完成时间可以是期望在多少时间内执行完成算力任务,比如可以是1小时、2小时等。算力任务所需的算法模型信息可以是算力任务具体需要的模型的信息,比如,可以是算法模型的具体名称、算法模型所需的详细参数等。
第二种方式:终端向网络侧设备发送算力任务描述信息,算力任务描述信息用于确定上述算力需求信息。
也就是说,终端在发送算力请求时,具体是发送算力任务描述信息,而 不是发送具体的算力需求。该算力任务描述信息可以用于网络侧设备解析得到算力需求信息。算力任务描述信息可以包括以下至少一项:算力任务类型、算力任务所需的算法模型的类型、算力任务的任务量描述、算力任务的期望完成时间范围、算力任务的期望完成成功率。算力任务类型可以包括以下至少一项:人工智能(Artificial Intelligence,AI)业务、数据处理类型业务、图像处理类、音频处理类业务,算法模型的类型可以是执行算力任务所需的算法模型的类别,每种类型的算法模型可以包括多个算法模型,网络侧设备可以基于算法模型的类型确定算力任务具体需要哪个算法模型,进而确定对应的算法模型信息。算力任务的任务量描述可以是对算力任务量的详细描述。算力任务的期望完成时间范围可以是期望在哪个时间范围内执行完成算力任务,比如可以是1~2个小时、2~3个小时等,网络侧设备可以基于算力任务的期望完成时间范围确定算力任务的期望完成时间。算力任务的期望完成成功率可以是执行完成算力任务后,期望算力任务被执行的成功率,比如对100万张图片进行人脸识别,期望识别成功率在99%以上。
比如,算力任务描述信息可以是:终端需要CPU、需要使用人脸识别类的AI模型、需要对100万张图片进行人脸识别,其中的CPU即为算力任务类型,人脸识别类的AI模型即为算力任务所需的算法模型,对100万张图片进行人脸识别即为算力任务的任务量描述。
本实施例中,终端中的算力任务可以是终端中已安装的目标应用的算力任务,也可以是终端自己的算力任务(即非目标应用的算力任务)。其中,目标应用的算力任务比如可以是,用于进行视频处理的目标应用需要对若干视频进行处理的算力任务。
在终端中的算力任务为目标应用的算力任务的情况下,终端在向网络侧设备发送算力请求之前,终端中的目标应用可以向终端发送算力请求。也就是说,在终端中的目标应用有算力需求的情况下,目标应用可以向终端发送算力请求,然后由终端将算力请求发送给网络侧设备。此处所述的终端具体 可以指终端(Operating System,OS)操作系统发送。
可选的,目标应用在向终端发送算力请求时,也可以通过上述两种方式中的任一种方式实现,即目标应用可以向终端发送算力需求信息和算力任务描述信息中的至少一项。其中,若目标应用向终端发送的是算力任务描述信息,则,终端在向网络侧设备发送算力请求时,可以直接将算力任务描述信息发送给网络侧设备,也可以先对算力任务描述信息进行解析,得到算力需求信息,然后将算力需求信息发送给网络侧设备。
总的来说,终端向网络侧设备发送算力请求,可以包括以下四种情况中的至少一项:
第一种情况:终端自己有算力任务,终端向网络侧设备发送算力任务的算力需求信息;
第二种情况:终端自己有算力任务,终端向网络侧设备发送算力任务描述信息;
第三种情况:终端中的目标应用有算力任务,目标应用向终端发送算力任务的算力需求信息,终端将该算力需求信息发送给网络侧设备;
第四种情况:终端中的目标应用有算力任务,目标应用向终端发送算力任务描述信息,终端将该算力任务描述信息发送给网络侧设备,或终端基于算力任务描述信息确定算力需求信息,将算力需求信息发送给网络侧设备。
针对网络侧设备而言,网络侧设备可以接收到来自终端的算力需求信息和算力任务描述信息中的至少一项。
S204:接收网络侧设备返回的目标算力设备的设备信息,目标算力设备包括具有算力能力的设备或设备集群。
网络侧设备在接收到终端的算力请求后,可以基于算力请求从多个算力设备中确定目标算力设备,即基于算力请求寻找能够匹配算力需求的目标算力设备,然后将目标算力设备的设备信息发送给终端,多个算力设备和目标算力设备均为具有算力能力的设备或设备集群,目标算力设备可以用于执行 终端的算力任务。网络侧设备确定目标算力设备的具体实现方式可以参见图3所示的算力分配方法的实施例,这里不再重复说明。
S206:基于设备信息将算力任务发送给目标算力设备,由目标算力设备执行算力任务。
终端在接收到由网络侧设备发送的目标算设力备的设备信息后,可以基于设备信息定位目标算力设备,并将算力任务发送给目标算力设备,以便由目标算力设备执行算力任务。
目标算力设备的设备信息具体可以包括以下至少一项:设备地址信息、设备标识符。其中,设备地址信息具体可以是互联网协议(Internet Protocol,IP)地址、统一资源定位符(Uniform Resource Locator,URL)地址、全限定域名(Fully Qualified Domain Name,FQDN)等,设备标识符可以是目标算力设备的设备标识(ID)等。终端在向目标算力设备发送算力任务时,首先,可以基于设备地址信息和设备标识符中的至少一项与目标算力设备建立连接,比如,IP连接等,在成功建立连接的情况下,终端可以将算力任务发送给目标算力设备。其中,终端在发送算力任务时,可以发送算力数据和算法等。比如,算力任务为对100万张图片进行人脸识别,那么,终端可以将100万张图片(算力数据)和相应的人脸识别的算法发送给目标算力设备。
本实施例中,目标算力设备的个数可以是一个也可以是多个。其中,在目标算力设备的个数为一个的情况下,该一个目标算力设备可以独立执行终端的算力任务,或者,该目标算力设备可以与其他算力设备共同执行算力任务。也就是说,在目标算力设备的个数为一个的情况下,终端可以将算力任务发送给这一个目标算力设备,目标算力设备在接收到算力任务后,可以单独执行该算力任务,也可以将算力任务进行拆分后分配给其他算力任务,以与其他算力设备共同执行算力任务。其中,如何拆分算力任务可以根据实际需求确定,这里不做具体限定。
上述其他算力设备可以由网络侧设备确定,具体实现方式可以参见图3 所示的算力分配方法的实施例,这里不再详细说明。网络侧设备在确定目标算力设备和其他算力设备后,可以将目标算力设备的设备信息发送给终端,将其他算力设备的设备信息发送给目标算力设备,以便终端基于目标算力设备的设备信息,将算力任务发送给目标算力设备,目标算力设备在接收到算力任务并拆分算力任务后,可以基于其他算力设备的设备信息将拆分后的算力任务分配给其他算力设备。其中,这里的目标算力设备可以称为主设备,其他算力设备可以称为辅设备,主设备用于接收终端的算力任务、拆分算力任务然后分发给辅设备,辅设备用于参与终端的算力任务执行,且只执行主设备分配给他的算力任务。
在目标算力设备的个数为多个的情况下,终端在发送算力任务时,可以先将算力任务拆分得到多个算力子任务,然后将多个算力子任务发送给多个目标算力设备。其中,不同的目标算力设备用于执行不同的算力子任务。需要说明的是,这里的多个目标算力设备可以理解为上述主设备和辅设备的集合,也就是说,网络侧设备在确定多个目标算力设备的情况下,除了上述的将主设备的设备信息发送给终端,将辅设备的设备信息发送给主设备外,还可以将多个目标算力设备的设备信息均发送给终端,以便终端可以将算力任务拆分后分发给多个目标算力设备执行。
由于终端在有算力任务时,可以向网络侧设备请求算力,由网络侧设备确定具有算力能力的目标算力设备,并将目标算力设备的算力分配给终端,使得终端可以借助于目标算力设备的算力执行算力任务,因此,在终端算力不足的情况下,可以保证算力任务的正常执行。
可选的,终端在将算力任务发送给目标算力设备后,还可以接收目标算力设备返回的算力任务执行结果。其中,若目标算力设备的个数为一个且这一个目标算力设备单独执行算力任务,则终端接收到的算力任务执行结果为该一个目标算力设备对算力任务的执行结果。若目标算力设备的个数为一个且这一个目标算力设备和其他算力设备共同执行算力任务,则终端接收到的 算力任务执行结果为目标算力设备和其他算力设备对算力任务的共同执行结果,其中,目标算力设备的执行结果可以直接发送给终端,其他算力设备的执行结果可以先发送给目标算力设备,再由目标算力设备发送给终端。若目标算力设备的个数为多个,则终端接收到的算力任务执行结果为该多个目标算力设备对算力任务的执行结果。
终端在接收到算力任务执行结果后,可以判断算力任务执行结果是否满足预设条件或基于算力任务执行结果确定算力任务是否执行成功。其中,满足预设条件可以是算力任务执行结果满足预设的执行效果,比如,执行结果的准确率高于设定阈值,算力任务的执行时间满足要求等,算力任务执行成功可以是在不考虑执行效果的前提下,算力任务被成功执行。若判断结果为算力任务执行结果满足预设条件或基于算力任务执行结果确定算力任务执行成功,则终端可以向网络侧设备发送通知信息,该通知信息用于告诉网络侧设备算力任务所执行的结果符合预期,并通知网络侧设备释放为终端的算力任务预留的算力,即释放目标算力设备为算力任务预留的算力,否则,可以继续由目标算力设备执行算力任务,直至算力任务执行结果满足预设条件或基于算力任务执行结果确定算力任务执行成功。
由于在终端的算力任务执行结果满足预设条件或在算力任务执行成功的情况下,可以由网络侧设备释放为算力任务预留的算力,因此,一方面不会长期占用目标算力设备的算力进而影响目标算力设备的正常使用,另一方面可以便于目标算力设备在释放算力后可以为其他算力任务预留算力,使得对算力的使用更为灵活。
如图3所示,本申请实施例提供一种算力分配方法300,该方法可以由网络侧设备执行,换言之,该方法可以由安装在网络侧设备的软件或硬件来执行,该方法包括如下步骤。
S302:网络侧设备接收来自终端的算力请求,算力请求用于为终端中的算力任务请求算力。
在终端中有算力任务的情况下,终端可以向网络侧设备发送算力请求,以为终端中的算力任务请求算力,网络侧设备可以接收到来自终端的算力请求。其中,终端发送算力请求的具体实现方式可以参见图2所示的算力请求方法的实施例,这不再重复说明。
本实施例中,网络侧设备接收来自终端算力请求,可以包括以下至少一项:
接收来自终端的算力任务的算力需求信息,算力需求信息包括以下至少一项:算力任务所需的算力总量、算力任务所需的算力类型、算力任务的期望完成时间;
接收来自终端的算力任务描述信息,算力任务描述信息用于确定算力需求信息,算力任务描述信息包括以下至少一项:算力任务类型、算力任务所需的算法模型、算力任务的任务量描述。
也就是说,网络侧设备接收到的算力请求具体可以是算力需求信息和算力任务描述信息中的至少一项,其中,针对算力需求信息和算力任务描述信息的详细介绍可以参见图2所示的实施例,这里不再重复说明。
S304:基于算力请求确定目标算力设备,目标算力设备包括具有算力能力的设备或设备集群。
网络侧设备在接收到算力请求后,可以基于算力请求从多个算力设备中确定目标算力设备,即基于算力请求寻找能够匹配算力需求的目标算力设备,并在目标算力设备中预留匹配算力需求的算力资源。
网络侧设备在基于算力请求确定目标算力设备时,具体可以是基于算力任务的算力需求信息确定目标算力设备。其中,若网络侧设备接收到的算力请求包括算力需求信息,则可以直接基于算力任务的算力需求信息确定目标算力设备。若网络侧设备接收到的算力请求包括算力任务描述信息,则在确定目标算力设备时,可首先对算力任务描述信息进行解析(可以由网络侧设备执行,也可以由额外的算力任务描述信息分析网元),这一解析过程,主要 是根据算力任务描述信息,经过分析和判断,得到该算力任务所对应的算力需求信息,然后基于算力需求信息确定目标算力设备。
在基于算力需求信息确定目标算力设备时,可以包括以下步骤:
S3041:预先获得多个算力设备的算力状态信息。
算力设备的算力状态信息可以包括以下至少一项:算力总量、算力使用量、算力类型、可用算力。多个算力设备的算力状态信息可以由多个算力设备在与网络侧设备建立协议数据单元(Protocol Data Unit,PDU)会话之前,在网络侧设备中进行注册时,通过将算力状态信息携带在注册请求中上报给网络侧设备,或者,也可以是在网络侧设备中进行注册后,在与网络侧设备建立PDU会话时,将算力状态信息上报给网络侧设备,或者也可以通过其他方式上报给网络侧设备,这里不再一一举例说明。
网络侧设备在接收到多个算力设备的算力状态信息后,可以将算力信息进行存储,当需要根据终端的算力需求信息确定目标算力设备时,可以获取预先获取的多个算力设备的算力状态信息。
S3042:将算力需求信息与多个算力设备的算力状态信息进行匹配,确定相匹配的至少一个候选算力设备。
本实施例中,可以将多个算力设备的算力状态信息分别与算力任务的算力需求信息进行匹配,根据匹配结果,可以将相匹配的一个或多个算力设备作为候选算力设备。这里的相匹配可以是以下至少一项:算力设备的算力类型包括算力任务所需的算力类型、算力设备的可用算力大于或等于算力任务所需的算力总量、算力设备任务执行效率符合或高于算力任务的需求、算力设备和终端之间的距离小于预设距离(满足时延要求)。可选地,这里的相匹配还可以是算力设备的其他信息与终端或终端的算力任务相匹配,具体可以根据实际情况进行设置,这里不做具体限定。
S3043:基于至少一个候选算力设备确定目标算力设备。
若候选算力设备的个数为一个,则可以向该一个候选算力设备发送算力 预留请求,算力预留请求用于请求候选算力设备预留匹配算力预留请求中的算力资源。候选算力设备在接收到算力预留请求后,可以将与算力预留请求中的算力资源匹配的可用算力或算力总量预留出来,若预留成功,则可以返回预留算力成功信息,若预留失败,则可以返回预留算力失败信息或不返回任何信息。针对网络侧设备而言,若接收到候选算力设备返回的预留算力成功信息,则可以将候选算力设备确定为目标算力设备,由此可以得到一个目标算力设备,若没有接收到候选算力设备的预留算力成功信息或接收到候选算力设备返回的预留算力失败信息,则可以确定没有目标算力设备,说明本算力设备预留算力失败,不可再使用该设备执行算力任务,本实施例以接收到预留算力成功信息为例进行说明。
若候选算力设备的个数为多个,则可以从中选择一个或多个候选算力设备作为目标算力设备,具体实现方式如下。
网络侧设备可以向多个候选算力设备均发送算力预留请求,针对任一候选算力设备,若接收到候选算力设备返回的预留算力成功信息,则可以将候选算力设备确定为目标算力设备,若没有接收到候选算力设备的预留算力成功信息或接收到候选算力设备返回的预留算力失败信息,则可以不将候选算力设备确定为目标算力设备。由此,可以确定一个或多个目标算力设备。
或者,网络侧设备也可以基于算力设备的优先级从高到低的顺序,向至少一个候选算力设备发送算力预留请求,然后将至少一个候选算力设备中的第一候选算力设备确定为目标算力设备,该第一候选算力设备为接收到算力预留请求并返回预留算力成功信息的算力设备。也就是说,网络侧设备可以先向优先级最高的候选算力设备发送算力预留请求,若接收到该候选算力设备返回的预留算力成功信息,则可以将该候选算力设备确定为目标算力设备,若未接收到该候选算力设备的预留算力成功信息或接收到该候选算力设备返回的预留算力失败信息,则可以向第二优先级的候选算力设备发送算力预留信息,……,如此循环,直至接收到其中一个候选算力设备的预留算力成功 信息,并将该候选算力设备确定为目标算力设备。可选地,在确定一个目标算力设备后,可以不再向其他候选算力设备发送算力预留信息,由此可以确定一个目标算力设备,或者,也可以继续向其他候选算力设备发送算力预留信息,由此可以确定多个目标算力设备。
上述算力设备的优先级基于以下至少一项确定:算力设备的算力总量、算力总量与算力任务的匹配程度、算力设备与终端之间的路由距离。其中,算力设备的算力总量越大,算力设备的优先级越高,算力总量与算力任务的匹配程度越高,算力设备的优先级越高,算力设备与终端之间的路由距离越近(即时延越小),算力设备的优先级越高。算力总量与算力任务的匹配程度可以表征算力任务的执行效果(比如算力任务的完成时间,执行结果的准确率等),算力类型的匹配程度等。
S306:将目标算力设备的设备信息发送给终端,由终端基于设备信息将算力任务发送给目标算力设备,并由目标算力设备执行算力任务。
网络侧设备在确定目标算力设备后,可以将目标算力设备的设备信息发送给终端。其中,目标算力设备的设备信息可以是目标算力设备在将算力状态信息上报给网络侧设备时同时上报给网络侧设备,该设备信息可以包括以下至少一项:设备地址信息、设备标识符。
可选的,网络侧设备在发送目标算力设备的设备信息时,在目标算力设备的个数为一个的情况下,可以将目标算力设备的设备信息直接发送给终端,在目标算力设备的个数为多个的情况下,可以通过以下两种方式发送目标算力设备的设备信息。
第一种方式:将多个目标算力设备的设备信息均发送给终端。
在第一种方式下,终端在接收到多个目标算力设备的设备信息后,可以将算力任务拆分为多个算力子任务后,基于多个目标算力设备的设备信息,终端将多个算力子任务分别发送给多个目标算力设备,不同的目标算力设备执行不同的算力子任务。此方式依赖于,终端可能会对算力任务进行拆分, 拆分为算力子任务,然后分别发送至不同的目标算力设备。
第二种方式:确定多个目标算力设备中的主设备和辅设备,将主设备的设备信息发送给终端,以及将辅设备的设备信息发送给主设备。
主设备可以是多个目标算力设备中的一个目标算力设备,辅设备可以是多个目标算力设备中除主设备以外的其他目标算力设备,主设备用于接收终端的算力任务并拆分算力任务,然后分配给辅设备,以与辅设备共同执行算力任务,辅设备用于参与算力任务的执行,且只执行主设备分配给他的那部分算力任务。网络侧设备可以根据实际情况确定多个目标算力设备中哪个算力设备为主设备,剩余的算力设备即为辅设备。在确定主设备和辅设备后,可以将主设备的设备信息发送给终端,将辅设备的设备信息发送给主设备,以便终端可以基于主设备的设备信息将算力任务发送给主设备,主设备在接收到算力任务后,将算力任务拆分,并基于辅设备的设备信息将拆分后的算力任务分配给辅设备,以共同执行算力任务。
本实施例中,网络侧设备在将目标算力设备的设备信息发送给终端后,终端可以基于接收到的设备信息,将算力任务发送给目标算力设备,以便由算力设备执行算力任务,具体实现方式可以参见图2所示的算力请求方法的实施例,这里不再重复说明。
由于终端在有算力任务时,可以向网络侧设备请求算力,由网络侧设备确定具有算力能力的目标算力设备,并将目标算力设备的算力分配给终端,使得终端可以借助于目标算力设备的算力执行算力任务,因此,在终端算力不足的情况下,可以保证算力任务的正常执行。
可选的,网络侧设备在将目标算力设备的设备信息发送给终端后,还可以在接收到终端发送的通知信息时,释放目标算力设备为终端的算力任务预留的算力,具体实现方式如下。
终端在将算力任务发送给目标算力设备后,目标算力设备执行算力任务,并将算力任务执行结果发送给终端,终端可以确定算力任务执行结果是否满 足预设条件或基于算力任务执行结果确定算力任务是否执行成功。若算力任务执行结果满足预设条件或基于算力任务执行结果确定算力任务执行成功,则终端会向网络侧设备发送通知信息,网络侧设备在接收到通知信息后,可以触发为该终端执行算力任务的目标算力设备的资源释放,并向目标算力设备发送算力释放请求,目标算力设备在接收到算力释放请求后,可以释放为终端的算力任务预留的算力。
由于在终端的算力任务执行结果满足预设条件或在算力任务执行成功的情况下,网络侧设备可以释放目标算力设备为算力任务预留的算力,因此,一方面不会长期占用目标算力设备的算力进而影响目标算力设备的正常使用,另一方面可以便于目标算力设备在释放算力后可以为其他算力任务预留算力,使得对算力的使用更为灵活。
如图4所示,本申请实施例提供一种算力执行方法400,该方法可以由目标算力设备执行,换言之,该方法可以由安装在目标算力设备的软件或硬件来执行,该方法包括如下步骤。
S402:目标算力设备接收来自终端的算力任务,算力任务由终端基于目标算力设备的设备信息发送,设备信息由网络侧设备在接收到终端的算力请求并确定目标算力设备后返回给终端,算力请求用于为终端中的算力任务请求算力,目标算力设备包括具有算力能力的设备或设备集群。
在终端中有算力任务的情况下,终端可以向网络侧设备发送算力请求,以为终端中的算力任务请求算力。网络侧设备在接收到算力请求后,可以基于算力请求确定目标算力设备,并将目标算力设备的设备信息发送给终端。终端接收到设备信息后,可以基于设备信息将算力任务发送给目标算力设备,目标算力设备可以接收到来自终端的算力任务。其中,终端向网络侧设备发送算力请求、向目标算力设备发送算力任务的具体实现方式可以参见图2所示实施例中相应步骤的具体实现,网络侧设备确定目标算力设备的具体实现方式可以参见图3所示实施例中相应步骤的具体实现,这里都不再重复说明。
需要说明的是,网络侧设备确定的目标算力设备的个数可以是一个,也可以是多个,在目标算力设备的个数为一个的情况下,图4所示方法的执行主体为该一个目标算力设备,在目标算力设备的个数为多个的情况下,若网络侧设备将该多个目标算力设备的设备信息均发送给终端,则图4所示方法的执行主体为其中一个目标算力设备,若网络侧设备仅将该多个目标算力设备中的一个目标算力设备(即主设备)的设备信息发送给终端,则图4所示方法的执行主体为主设备。
S404:执行算力任务。
目标算力设备在接收到算力任务后,可以执行算力任务。
可选的,若网络侧设备确定的目标算力设备的个数为一个,则在执行算力任务时,可以由这一个目标算力设备独立执行算力任务。若网络侧设备确定的目标算力设备的个数为多个,且将该多个目标算力设备的设备信息均发送给终端,则在执行算力任务时,针对每个目标算力设备,可以执行终端分配给它的算力子任务。若网络侧设备确定的目标算力设备的个数为多个,且仅将该多个目标算力设备中一个目标算力设备(即主设备)的设备信息发送给终端,则在执行算力任务时,可以将算力任务进行拆分后分配给其他算力设备(即辅设备),以共同执行算力任务。
由于终端在有算力任务时,可以向网络侧设备请求算力,由网络侧设备确定具有算力能力的目标算力设备,并将目标算力设备的算力分配给终端,使得终端可以借助于目标算力设备的算力执行算力任务,因此,在终端算力不足的情况下,可以保证算力任务的正常执行。
可选的,目标算力设备在执行完成算力任务后,可以将算力任务执行结果返回给终端。其中,若算力任务是由主设备和辅设备共同执行,则辅设备可以先将算力任务执行结果发送给主设备,再由主设备发送给终端。
终端在接收到算力任务执行结果后,在确定算力任务执行结果满足预设条件或基于算力任务执行结果确定算力任务执行成功的情况下,可以向网络 侧设备发送通知信息,网络侧设备接收到通知信息后,可以向目标算力设备发送算力释放请求,以请求释放为算力任务预留的算力。目标算力设备在接收到算力释放请求后,可以释放为算力任务预留的算力
由于在终端的算力任务执行结果满足预设条件或在算力任务执行成功的情况下,可以由网络侧设备释放为算力任务预留的算力,因此,一方面不会长期占用目标算力设备的算力进而影响目标算力设备的正常使用,另一方面可以便于目标算力设备在释放算力后可以为其他算力任务预留算力,使得对算力的使用更为灵活。
需要说明的是,本申请实施例提供的算力请求方法,执行主体可以为算力请求装置,或者,该算力请求装置中的用于执行算力请求方法的控制模块。本申请实施例中以算力请求装置执行算力请求方法为例,说明本申请实施例提供的算力请求装置。本申请实施例提供的算力分配方法,执行主体可以为算力分配装置,或者,该算力分配装置中的用于执行算力分配方法的控制模块。本申请实施例中以算力分配装置执行算力分配方法为例,说明本申请实施例提供的算力分配装置。本申请实施例提供的算力执行方法,执行主体可以为算力执行装置,或者,该算力执行装置中的用于执行算力执行方法的控制模块。本申请实施例中以算力执行装置执行算力执行方法为例,说明本申请实施例提供的算力执行装置。
图5是根据本申请实施例的算力请求装置的结构示意图,该装置可以对应于其他实施例中的终端。如图5所示,装置500包括如下模块。
第一发送模块501,用于向网络侧设备发送算力请求,所述算力请求用于为所述终端中的算力任务请求算力;
接收模块502,用于接收所述网络侧设备返回的目标算力设备的设备信息,所述目标算力设备包括具有算力能力的设备或设备集群;
第二发送模块503,用于基于所述设备信息将所述算力任务发送给所述目标算力设备,由所述目标算力设备执行所述算力任务。
可选地,作为一个实施例,所述第一发送模块501,还用于以下至少一项:
向网络侧设备发送所述算力任务的算力需求信息,所述算力需求信息用于确定匹配算力需求的目标算力设备,所述算力需求信息包括以下至少一项:算力任务所需的算力总量、算力任务所需的算力类型、算力任务的期望完成时间、算力任务所需的算法模型信息;
向网络侧设备发送算力任务描述信息,所述算力任务描述信息用于确定所述算力需求信息,所述算力任务描述信息包括以下至少一项:算力任务类型、算力任务所需的算法模型的类型、算力任务的任务量描述、算力任务的期望完成时间范围、算力任务的期望完成成功率。
可选地,作为一个实施例,所述第一发送模块501,还用于:发送所述算力请求给所述终端;
其中,在所述目标应用发送的算力请求包括所述算力任务描述信息,所述终端发送的算力请求包括所述算力需求信息的情况下,基于所述算力任务描述信息确定所述算力需求信息;
将所述算力需求信息发送给所述网络侧设备。
可选地,作为一个实施例,所述设备信息包括以下至少一项:设备地址信息、设备标识符;其中,所述第二发送模块503,还用于:
基于所述设备地址信息和所述设备标识符中的至少一项,与所述目标算力设备建立连接;
在成功建立连接的情况下,将所述算力任务发送给所述目标算力设备。
可选地,作为一个实施例,所述目标算力设备的个数为一个或多个;
其中,在所述目标算力设备的个数为一个的情况下,所述目标算力设备用于独立执行所述算力任务,或,所述目标算力设备用于将所述算力任务拆分后分配给其他算力设备,以共同执行所述算力任务,所述其他算力设备由所述网络侧设备确定;
在所述目标算力设备的个数为多个的情况下,所述第二发送模块503,还用于:
将所述算力任务进行拆分,得到多个算力子任务;
将所述多个算力子任务发送给多个目标算力设备,不同的目标算力设备用于执行不同的算力子任务。
可选地,作为一个实施例,所述接收模块502,还用于接收所述目标算力设备返回的算力任务执行结果;
所述第一发送模块501,还用于在确定所述算力任务执行结果满足预设条件或基于所述算力任务执行结果确定所述算力任务执行成功的情况下,向所述网络侧设备发送通知信息,所述通知信息用于所述网络侧设备释放为所述算力任务预留的算力。
根据本申请实施例的装置500可以参照对应本申请实施例的方法200的流程,并且,该装置500中的各个单元/模块和上述其他操作和/或功能分别为了实现方法200中的相应流程,并且能够达到相同或等同的技术效果,为了简洁,在此不再赘述。
图6是根据本申请实施例的算力分配装置的结构示意图,该装置可以对应于其他实施例中的网络侧设备。如图6所示,装置600包括如下模块。
接收模块601,用于接收来自终端的算力请求,所述算力请求用于为所述终端中的算力任务请求算力;
确定模块602,用于基于所述算力请求确定目标算力设备,所述目标算力设备包括具有算力能力的设备或设备集群;
发送模块603,用于将所述目标算力设备的设备信息发送给所述终端,由所述终端基于所述设备信息将所述算力任务发送给所述目标算力设备,并由所述目标算力设备执行所述算力任务。
可选地,作为一个实施例,所述接收模块601,还用于以下至少一项:
接收来自所述终端的所述算力任务的算力需求信息,所述算力需求信息 用于确定匹配算力需求的目标算力设备,所述算力需求信息包括以下至少一项:算力任务所需的算力总量、算力任务所需的算力类型、算力任务的期望完成时间、算力任务所需的算法模型信息;
接收来自所述终端的算力任务描述信息,所述算力任务描述信息用于确定所述算力需求信息,所述算力任务描述信息包括以下至少一项:算力任务类型、算力任务所需的算法模型的类型、算力任务的任务量描述、算力任务的期望完成时间范围、算力任务的期望完成成功率。
可选地,作为一个实施例,所述确定模块602,还用于以下至少一项:
在所述算力请求包括所述算力需求信息的情况下,基于所述算力需求信息确定所述目标算力设备;
在所述算力请求包括所述算力任务描述信息的情况下,基于所述算力任务描述信息确定所述算力需求信息;基于所述算力需求信息确定所述目标算力设备。
可选地,作为一个实施例,所述确定模块602,还用于:
预先获得多个算力设备的算力状态信息,所述算力状态信息包括以下至少一项:算力总量、算力使用量、算力类型、可用算力;
将所述算力需求信息与所述多个算力设备的算力状态信息进行匹配,确定相匹配的至少一个候选算力设备;
基于所述至少一个候选算力设备确定所述目标算力设备。
可选地,作为一个实施例,所述确定模块602,还用于:
向所述至少一个候选算力设备发送算力预留请求;针对任一候选算力设备,若接收到所述候选算力设备返回的预留算力成功信息,则将所述候选算力设备确定为所述目标算力设备;或,
基于算力设备的优先级从高到低的顺序,向所述至少一个候选算力设备发送算力预留请求;将所述至少一个候选算力设备中的第一候选算力设备确定为所述目标算力设备,所述第一候选算力设备为接收到所述算力预留请求 并返回预留算力成功信息的算力设备;所述算力设备的优先级基于以下至少一项确定:算力设备的算力总量、算力总量与算力任务的匹配程度、算力设备与所述终端之间的路由距离。
可选地,作为一个实施例,所述目标算力设备的个数为一个或多个;
其中,在所述目标算力设备的个数为多个的情况下,所述发送模块603,还用于:
将多个目标算力设备的设备信息均发送给所述终端;或,
确定所述多个目标算力设备中的主设备和辅设备;将所述主设备的设备信息发送给所述终端,以及将所述辅设备的设备信息发送给所述主设备,由所述终端基于所述主设备的设备信息将所述算力任务发送给所述主设备,并由所述主设备将所述算力任务拆分后基于所述辅设备的设备信息分配给所述辅设备,以共同执行所述算力任务。
可选地,作为一个实施例,所述接收模块601,还用于接收来自所述终端的通知信息,所述通知信息由所述终端接收到所述目标算力设备返回的算力任务执行结果后,在确定所述算力任务执行结果满足预设条件或基于所述算力任务执行结果确定所述算力任务执行成功的情况下发送;
所述发送模块603,还用于向所述目标算力设备发送算力释放请求,以释放所述目标算力设备为所述算力任务预留的算力。
根据本申请实施例的装置600可以参照对应本申请实施例的方法300的流程,并且,该装置600中的各个单元/模块和上述其他操作和/或功能分别为了实现方法300中的相应流程,并且能够达到相同或等同的技术效果,为了简洁,在此不再赘述。
图7是根据本申请实施例的算力执行装置的结构示意图,该装置可以对应于其他实施例中的目标算力设备。如图7所示,装置700包括如下模块。
接收模块701,用于接收来自终端的算力任务,所述算力任务由所述终端基于所述目标算力设备的设备信息发送,所述设备信息由网络侧设备在接 收到所述终端的算力请求并确定所述目标算力设备后返回给所述终端,所述算力请求用于为所述终端中的算力任务请求算力,所述目标算力设备包括具有算力能力的设备或设备集群;
任务执行模块702,用于执行所述算力任务。
可选地,作为一个实施例,所述目标算力设备的个数为一个或多个,其中:
在所述目标算力设备的个数为一个的情况下,所述任务执行模块702,还用于:
独立执行所述算力任务;或,
将所述算力任务进行拆分后分配给其他算力设备,以共同执行所述算力任务,所述其他算力设备由所述网络侧设备确定,所述其他算力设备完成算力任务后,将算力任务执行结果发送至所述目标算力设备;
在所述目标算力设备的个数为多个的情况下,所述任务执行模块702,还用于:
执行来自所述终端的算力子任务,所述算力子任务由所述终端对所述算力任务进行拆分后得到。
可选地,作为一个实施例,所述任务执行模块702,还用于将算力任务执行结果发送给所述终端;
所述接收模块701,还用于接收来自所述网络侧设备的算力释放请求,所述算力释放请求由所述网络侧设备接收到所述终端的通知信息后发送,所述通知信息由所述终端接收到所述算力任务执行结果后,在确定所述算力任务执行结果满足预设条件或基于所述算力任务执行结果确定所述算力任务执行成功的情况下发送给所述网络侧设备;基于所述算力释放请求,释放为所述算力任务预留的算力。
根据本申请实施例的装置700可以参照对应本申请实施例的方法400的流程,并且,该装置700中的各个单元/模块和上述其他操作和/或功能分别为 了实现方法400中的相应流程,并且能够达到相同或等同的技术效果,为了简洁,在此不再赘述。
本申请实施例中的算力请求装置、算力分配装置和算力执行装置可以是装置,具有操作系统的装置或电子设备,也可以是终端中的部件、集成电路、或芯片。该装置或电子设备可以是移动终端,也可以为非移动终端。示例性的,移动终端可以包括但不限于上述所列举的终端11的类型,非移动终端可以为服务器、网络附属存储器(Network Attached Storage,NAS)、个人计算机(personal computer,PC)、电视机(television,TV)、柜员机或者自助机等,本申请实施例不作具体限定。
本申请实施例提供的算力请求装置、算力分配装置和算力执行装置能够分别实现图2至图4的方法实施例实现的各个过程,并达到相同的技术效果,为避免重复,这里不再赘述。
可选的,如图8所示,本申请实施例还提供一种通信设备800,包括处理器801,存储器802,存储在存储器802上并可在所述处理器801上运行的程序或指令,例如,该通信设备800为终端时,该程序或指令被处理器801执行时实现上述算力请求方法实施例的各个过程,且能达到相同的技术效果。该通信设备800为网络侧设备时,该程序或指令被处理器801执行时实现上述算力分配方法实施例的各个过程,且能达到相同的技术效果,为避免重复,这里不再赘述。该通信设备800为目标算力设备时,该程序或指令被处理器801执行时实现上述算力执行方法实施例的各个过程,且能达到相同的技术效果,为避免重复,这里不再赘述。
本申请实施例还提供一种终端,包括处理器和通信接口,通信接口用于向网络侧设备发送算力请求,所述算力请求用于为所述终端中的算力任务请求算力;接收所述网络侧设备返回的目标算力设备的设备信息,所述目标算力设备包括具有算力能力的设备或设备集群;基于所述设备信息将所述算力任务发送给所述目标算力设备,由所述目标算力设备执行所述算力任务。该 终端实施例是与上述终端侧方法实施例对应的,上述方法实施例的各个实施过程和实现方式均可适用于该终端实施例中,且能达到相同的技术效果。具体地,图9为实现本申请实施例的一种终端的硬件结构示意图。
该终端900包括但不限于:射频单元901、网络模块902、音频输出单元903、输入单元904、传感器905、显示单元906、用户输入单元907、接口单元908、存储器909、以及处理器910等中的至少部分部件。
本领域技术人员可以理解,终端900还可以包括给各个部件供电的电源(比如电池),电源可以通过电源管理系统与处理器910逻辑相连,从而通过电源管理系统实现管理充电、放电、以及功耗管理等功能。图9中示出的终端结构并不构成对终端的限定,终端可以包括比图示更多或更少的部件,或者组合某些部件,或者不同的部件布置,在此不再赘述。
应理解的是,本申请实施例中,输入单元904可以包括图形处理器(Graphics Processing Unit,GPU)9041和麦克风9042,图形处理器9041对在视频捕获模式或图像捕获模式中由图像捕获装置(如摄像头)获得的静态图片或视频的图像数据进行处理。显示单元906可包括显示面板9061,可以采用液晶显示器、有机发光二极管等形式来配置显示面板9061。用户输入单元907包括触控面板9071以及其他输入设备9072。触控面板9071,也称为触摸屏。触控面板9071可包括触摸检测装置和触摸控制器两个部分。其他输入设备9072可以包括但不限于物理键盘、功能键(比如音量控制按键、开关按键等)、轨迹球、鼠标、操作杆,在此不再赘述。
本申请实施例中,射频单元901将来自网络侧设备的下行数据接收后,给处理器910处理;另外,将上行的数据发送给网络侧设备。通常,射频单元901包括但不限于天线、至少一个放大器、收发信机、耦合器、低噪声放大器、双工器等。
存储器909可用于存储软件程序或指令以及各种数据。存储器909可主要包括存储程序或指令区和存储数据区,其中,存储程序或指令区可存储操 作系统、至少一个功能所需的应用程序或指令(比如声音播放功能、图像播放功能等)等。此外,存储器909可以包括高速随机存取存储器,还可以包括非易失性存储器,其中,非易失性存储器可以是只读存储器(Read-OnlyMemory,ROM)、可编程只读存储器(ProgrammableROM,PROM)、可擦除可编程只读存储器(ErasablePROM,EPROM)、电可擦除可编程只读存储器(ElectricallyEPROM,EEPROM)或闪存。例如至少一个磁盘存储器件、闪存器件、或其他非易失性固态存储器件。
处理器910可包括一个或多个处理单元;可选的,处理器910可集成应用处理器和调制解调处理器,其中,应用处理器主要处理操作系统、用户界面和应用程序或指令等,调制解调处理器主要处理无线通信,如基带处理器。可以理解的是,上述调制解调处理器也可以不集成到处理器910中。
其中,射频单元910用于向网络侧设备发送算力请求,所述算力请求用于为所述终端中的算力任务请求算力;接收所述网络侧设备返回的目标算力设备的设备信息,所述目标算力设备包括具有算力能力的设备或设备集群;基于所述设备信息将所述算力任务发送给所述目标算力设备,由所述目标算力设备执行所述算力任务。
由于终端在有算力任务时,可以向网络侧设备请求算力,由网络侧设备确定具有算力能力的目标算力设备,并将目标算力设备的算力分配给终端,使得终端可以借助于目标算力设备的算力执行算力任务,因此,在终端算力不足的情况下,可以保证算力任务的正常执行。
本申请实施例提供的终端900还可以实现上述算力请求方法实施例的各个过程,且能达到相同的技术效果,为避免重复,这里不再赘述。
本申请实施例还提供一种网络侧设备,包括处理器和通信接口,通信接口用于接收来自终端的算力请求,所述算力请求用于为所述终端中的算力任务请求算力;处理器用于基于所述算力请求确定目标算力设备,所述目标算力设备包括具有算力能力的设备或设备集群;通信接口还用于将所述目标算 力设备的设备信息发送给所述终端,由所述终端基于所述设备信息将所述算力任务发送给所述目标算力设备,并由所述目标算力设备执行所述算力任务。该网络侧设备实施例是与上述网络侧设备方法实施例对应的,上述方法实施例的各个实施过程和实现方式均可适用于该网络侧设备实施例中,且能达到相同的技术效果。
具体地,本申请实施例还提供了一种网络侧设备。如图10所示,该网络设备1000包括:天线101、射频装置102、基带装置103。天线101与射频装置102连接。在上行方向上,射频装置102通过天线101接收信息,将接收的信息发送给基带装置103进行处理。在下行方向上,基带装置103对要发送的信息进行处理,并发送给射频装置102,射频装置102对收到的信息进行处理后经过天线101发送出去。
上述频带处理装置可以位于基带装置103中,以上实施例中网络侧设备执行的方法可以在基带装置103中实现,该基带装置103包括处理器104和存储器105。
基带装置103例如可以包括至少一个基带板,该基带板上设置有多个芯片,如图10所示,其中一个芯片例如为处理器104,与存储器105连接,以调用存储器105中的程序,执行以上方法实施例中所示的网络设备操作。
该基带装置103还可以包括网络接口106,用于与射频装置102交互信息,该接口例如为通用公共无线接口(common public radio interface,简称CPRI)。
具体地,本发明实施例的网络侧设备还包括:存储在存储器105上并可在处理器104上运行的指令或程序,处理器104调用存储器105中的指令或程序执行图6所示各模块执行的方法,并达到相同的技术效果,为避免重复,故不在此赘述。
本申请实施例还提供一种可读存储介质,所述可读存储介质上存储有程序或指令,该程序或指令被处理器执行时实现上述算力请求方法实施例的各 个过程,或实现上述算力分配方法实施例的各个过程,或实现上述算力执行方法实施例的各个过程,且能达到相同的技术效果,为避免重复,这里不再赘述。
其中,所述处理器为上述实施例中所述的终端中的处理器。所述可读存储介质,包括计算机可读存储介质,如计算机只读存储器(Read-Only Memory,ROM)、随机存取存储器(Random Access Memory,RAM)、磁碟或者光盘等。
本申请实施例另提供了一种芯片,所述芯片包括处理器和通信接口,所述通信接口和所述处理器耦合,所述处理器用于运行程序或指令,实现上述算力请求方法实施例的各个过程,或实现上述算力分配方法实施例的各个过程,或实现上述算力执行方法实施例的各个过程,且能达到相同的技术效果,为避免重复,这里不再赘述。
应理解,本申请实施例提到的芯片还可以称为系统级芯片,系统芯片,芯片系统或片上系统芯片等。
本申请实施例另提供了一种计算机程序/程序产品,所述计算机程序/程序产品被存储在非易失的存储介质中,所述程序/程序产品被至少一个处理器执行以实现上述算力请求方法实施例的各个过程,或实现上述算力分配方法实施例的各个过程,或实现上述算力执行方法实施例的各个过程,且能达到相同的技术效果,为避免重复,这里不再赘述。
需要说明的是,在本文中,术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含,从而使得包括一系列要素的过程、方法、物品或者装置不仅包括那些要素,而且还包括没有明确列出的其他要素,或者是还包括为这种过程、方法、物品或者装置所固有的要素。在没有更多限制的情况下,由语句“包括一个……”限定的要素,并不排除在包括该要素的过程、方法、物品或者装置中还存在另外的相同要素。此外,需要指出的是,本申请实施方式中的方法和装置的范围不限按示出或讨论的顺序来执行功能,还 可包括根据所涉及的功能按基本同时的方式或按相反的顺序来执行功能,例如,可以按不同于所描述的次序来执行所描述的方法,并且还可以添加、省去、或组合各种步骤。另外,参照某些示例所描述的特征可在其他示例中被组合。
通过以上的实施方式的描述,本领域的技术人员可以清楚地了解到上述实施例方法可借助软件加必需的通用硬件平台的方式来实现,当然也可以通过硬件,但很多情况下前者是更佳的实施方式。基于这样的理解,本申请的技术方案本质上或者说对现有技术做出贡献的部分可以以计算机软件产品的形式体现出来,该计算机软件产品存储在一个存储介质(如ROM/RAM、磁碟、光盘)中,包括若干指令用以使得一台终端(可以是手机,计算机,服务器,空调器,或者网络设备等)执行本申请各个实施例所述的方法。
上面结合附图对本申请的实施例进行了描述,但是本申请并不局限于上述的具体实施方式,上述的具体实施方式仅仅是示意性的,而不是限制性的,本领域的普通技术人员在本申请的启示下,在不脱离本申请宗旨和权利要求所保护的范围情况下,还可做出很多形式,均属于本申请的保护之内。

Claims (36)

  1. 一种算力请求方法,其中,包括:
    终端向网络侧设备发送算力请求,所述算力请求用于为所述终端中的算力任务请求算力;
    接收所述网络侧设备返回的目标算力设备的设备信息,所述目标算力设备包括具有算力能力的设备或设备集群;
    基于所述设备信息将所述算力任务发送给所述目标算力设备,由所述目标算力设备执行所述算力任务。
  2. 根据权利要求1所述的方法,其中,向网络侧设备发送针对算力任务的算力请求,包括以下至少一项:
    向网络侧设备发送所述算力任务的算力需求信息,所述算力需求信息用于确定匹配算力需求的目标算力设备,所述算力需求信息包括以下至少一项:算力任务所需的算力总量、算力任务所需的算力类型、算力任务的期望完成时间、算力任务所需的算法模型信息;
    向网络侧设备发送算力任务描述信息,所述算力任务描述信息用于确定所述算力需求信息,所述算力任务描述信息包括以下至少一项:算力任务类型、算力任务所需的算法模型的类型、算力任务的任务量描述、算力任务的期望完成时间范围、算力任务的期望完成成功率。
  3. 根据权利要求2所述的方法,其中,终端向网络侧设备发送算力请求之前,所述方法还包括:
    所述终端中的目标应用发送所述算力请求给所述终端;
    其中,在所述目标应用发送的算力请求包括所述算力任务描述信息,所述终端发送的算力请求包括所述算力需求信息的情况下,所述方法还包括:
    基于所述算力任务描述信息确定所述算力需求信息;
    将所述算力需求信息发送给所述网络侧设备。
  4. 根据权利要求1所述的方法,其中,所述设备信息包括以下至少一项: 设备地址信息、设备标识符;其中,基于所述设备信息将所述算力任务发送给所述目标算力设备,包括:
    基于所述设备地址信息和所述设备标识符中的至少一项,与所述目标算力设备建立连接;
    在成功建立连接的情况下,将所述算力任务发送给所述目标算力设备。
  5. 根据权利要求1所述的方法,其中,
    所述目标算力设备的个数为一个或多个;
    其中,在所述目标算力设备的个数为一个的情况下,所述目标算力设备用于独立执行所述算力任务,或,所述目标算力设备用于将所述算力任务拆分后分配给其他算力设备,以共同执行所述算力任务,所述其他算力设备由所述网络侧设备确定;
    在所述目标算力设备的个数为多个的情况下,基于所述设备信息将所述算力任务发送给所述目标算力设备,包括:
    将所述算力任务进行拆分,得到多个算力子任务;
    将所述多个算力子任务发送给多个目标算力设备,不同的目标算力设备用于执行不同的算力子任务。
  6. 根据权利要求1所述的方法,其中,在将所述算力任务发送给所述目标算力设备后,所述方法还包括:
    接收所述目标算力设备返回的算力任务执行结果;
    在确定所述算力任务执行结果满足预设条件或基于所述算力任务执行结果确定所述算力任务执行成功的情况下,向所述网络侧设备发送通知信息,所述通知信息用于所述网络侧设备释放为所述算力任务预留的算力。
  7. 一种算力分配方法,其中,包括:
    网络侧设备接收来自终端的算力请求,所述算力请求用于为所述终端中的算力任务请求算力;
    基于所述算力请求确定目标算力设备,所述目标算力设备包括具有算力 能力的设备或设备集群;
    将所述目标算力设备的设备信息发送给所述终端,由所述终端基于所述设备信息将所述算力任务发送给所述目标算力设备,并由所述目标算力设备执行所述算力任务。
  8. 根据权利要求7所述的方法,其中,接收来自终端的算力请求,包括以下至少一项:
    接收来自所述终端的所述算力任务的算力需求信息,所述算力需求信息用于确定匹配算力需求的目标算力设备,所述算力需求信息包括以下至少一项:算力任务所需的算力总量、算力任务所需的算力类型、算力任务的期望完成时间、算力任务所需的算法模型信息;
    接收来自所述终端的算力任务描述信息,所述算力任务描述信息用于确定所述算力需求信息,所述算力任务描述信息包括以下至少一项:算力任务类型、算力任务所需的算法模型的类型、算力任务的任务量描述、算力任务的期望完成时间范围、算力任务的期望完成成功率。
  9. 根据权利要求8所述的方法,其中,基于所述算力请求确定目标算力设备,包括以下至少一项:
    在所述算力请求包括所述算力需求信息的情况下,基于所述算力需求信息确定所述目标算力设备;
    在所述算力请求包括所述算力任务描述信息的情况下,基于所述算力任务描述信息确定所述算力需求信息;基于所述算力需求信息确定所述目标算力设备。
  10. 根据权利要求9所述的方法,其中,基于所述算力需求信息确定所述目标算力设备,包括:
    预先获得多个算力设备的算力状态信息,所述算力状态信息包括以下至少一项:算力总量、算力使用量、算力类型、可用算力;
    将所述算力需求信息与所述多个算力设备的算力状态信息进行匹配,确 定相匹配的至少一个候选算力设备;
    基于所述至少一个候选算力设备确定所述目标算力设备。
  11. 根据权利要求10所述的方法,其中,基于所述至少一个候选算力设备确定所述目标算力设备,包括:
    向所述至少一个候选算力设备发送算力预留请求;针对任一候选算力设备,若接收到所述候选算力设备返回的预留算力成功信息,则将所述候选算力设备确定为所述目标算力设备;或,
    基于算力设备的优先级从高到低的顺序,向所述至少一个候选算力设备发送算力预留请求;将所述至少一个候选算力设备中的第一候选算力设备确定为所述目标算力设备,所述第一候选算力设备为接收到所述算力预留请求并返回预留算力成功信息的算力设备;所述算力设备的优先级基于以下至少一项确定:算力设备的算力总量、算力总量与算力任务的匹配程度、算力设备与所述终端之间的路由距离。
  12. 根据权利要求7所述的方法,其中,所述目标算力设备的个数为一个或多个;
    其中,在所述目标算力设备的个数为多个的情况下,将所述目标算力设备的设备信息发送给所述终端,包括:
    将多个目标算力设备的设备信息均发送给所述终端;或,
    确定所述多个目标算力设备中的主设备和辅设备;将所述主设备的设备信息发送给所述终端,以及将所述辅设备的设备信息发送给所述主设备,由所述终端基于所述主设备的设备信息将所述算力任务发送给所述主设备,并由所述主设备将所述算力任务拆分后基于所述辅设备的设备信息分配给所述辅设备,以共同执行所述算力任务。
  13. 根据权利要求7所述的方法,其中,在将所述目标算力设备的设备信息发送给所述终端后,所述方法还包括:
    接收来自所述终端的通知信息,所述通知信息由所述终端接收到所述目 标算力设备返回的算力任务执行结果后,在确定所述算力任务执行结果满足预设条件或基于所述算力任务执行结果确定所述算力任务执行成功的情况下发送;
    向所述目标算力设备发送算力释放请求,以释放所述目标算力设备为所述算力任务预留的算力。
  14. 一种算力执行方法,其中,包括:
    目标算力设备接收来自终端的算力任务,所述算力任务由所述终端基于所述目标算力设备的设备信息发送,所述设备信息由网络侧设备在接收到所述终端的算力请求并确定所述目标算力设备后返回给所述终端,所述算力请求用于为所述终端中的算力任务请求算力,所述目标算力设备包括具有算力能力的设备或设备集群;
    执行所述算力任务。
  15. 根据权利要求14所述的方法,其中,所述目标算力设备的个数为一个或多个,其中:
    在所述目标算力设备的个数为一个的情况下,执行所述算力任务,包括:
    独立执行所述算力任务;或,
    将所述算力任务进行拆分后分配给其他算力设备,以共同执行所述算力任务,所述其他算力设备由所述网络侧设备确定,所述其他算力设备完成算力任务后,将算力任务执行结果发送至所述目标算力设备;
    在所述目标算力设备的个数为多个的情况下,执行所述算力任务,包括:
    执行来自所述终端的算力子任务,所述算力子任务由所述终端对所述算力任务进行拆分后得到。
  16. 根据权利要求14所述的方法,其中,执行所述算力任务后,所述方法还包括:
    将算力任务执行结果发送给所述终端;
    接收来自所述网络侧设备的算力释放请求,所述算力释放请求由所述网 络侧设备接收到所述终端的通知信息后发送,所述通知信息由所述终端接收到所述算力任务执行结果后,在确定所述算力任务执行结果满足预设条件或基于所述算力任务执行结果确定所述算力任务执行成功的情况下发送给所述网络侧设备;
    基于所述算力释放请求,释放为所述算力任务预留的算力。
  17. 一种算力请求装置,其中,包括:
    第一发送模块,用于向网络侧设备发送算力请求,所述算力请求用于为所述终端中的算力任务请求算力;
    接收模块,用于接收所述网络侧设备返回的目标算力设备的设备信息,所述目标算力设备包括具有算力能力的设备或设备集群;
    第二发送模块,用于基于所述设备信息将所述算力任务发送给所述目标算力设备,由所述目标算力设备执行所述算力任务。
  18. 根据权利要求17所述的装置,其中,所述第一发送模块,还用于以下至少一项:
    向网络侧设备发送所述算力任务的算力需求信息,所述算力需求信息用于确定匹配算力需求的目标算力设备,所述算力需求信息包括以下至少一项:算力任务所需的算力总量、算力任务所需的算力类型、算力任务的期望完成时间、算力任务所需的算法模型信息;
    向网络侧设备发送算力任务描述信息,所述算力任务描述信息用于确定所述算力需求信息,所述算力任务描述信息包括以下至少一项:算力任务类型、算力任务所需的算法模型的类型、算力任务的任务量描述、算力任务的期望完成时间范围、算力任务的期望完成成功率。
  19. 根据权利要求18所述的装置,其中,所述第一发送模块,还用于:发送所述算力请求给所述终端;
    其中,在所述目标应用发送的算力请求包括所述算力任务描述信息,所述终端发送的算力请求包括所述算力需求信息的情况下,基于所述算力任务 描述信息确定所述算力需求信息;
    将所述算力需求信息发送给所述网络侧设备。
  20. 根据权利要求17所述的装置,其中,所述设备信息包括以下至少一项:设备地址信息、设备标识符;其中,所述第二发送模块,还用于:
    基于所述设备地址信息和所述设备标识符中的至少一项,与所述目标算力设备建立连接;
    在成功建立连接的情况下,将所述算力任务发送给所述目标算力设备。
  21. 根据权利要求17所述的装置,其中,
    所述目标算力设备的个数为一个或多个;
    其中,在所述目标算力设备的个数为一个的情况下,所述目标算力设备用于独立执行所述算力任务,或,所述目标算力设备用于将所述算力任务拆分后分配给其他算力设备,以共同执行所述算力任务,所述其他算力设备由所述网络侧设备确定;
    在所述目标算力设备的个数为多个的情况下,所述第二发送模块,还用于:
    将所述算力任务进行拆分,得到多个算力子任务;
    将所述多个算力子任务发送给多个目标算力设备,不同的目标算力设备用于执行不同的算力子任务。
  22. 根据权利要求17所述的装置,其中,
    所述接收模块,还用于接收所述目标算力设备返回的算力任务执行结果;
    所述第一发送模块,还用于在确定所述算力任务执行结果满足预设条件或基于所述算力任务执行结果确定所述算力任务执行成功的情况下,向所述网络侧设备发送通知信息,所述通知信息用于所述网络侧设备释放为所述算力任务预留的算力。
  23. 一种算力分配装置,其中,包括:
    接收模块,用于接收来自终端的算力请求,所述算力请求用于为所述终 端中的算力任务请求算力;
    确定模块,用于基于所述算力请求确定目标算力设备,所述目标算力设备包括具有算力能力的设备或设备集群;
    发送模块,用于将所述目标算力设备的设备信息发送给所述终端,由所述终端基于所述设备信息将所述算力任务发送给所述目标算力设备,并由所述目标算力设备执行所述算力任务。
  24. 根据权利要求23所述的装置,其中,所述接收模块,还用于以下至少一项:
    接收来自所述终端的所述算力任务的算力需求信息,所述算力需求信息用于确定匹配算力需求的目标算力设备,所述算力需求信息包括以下至少一项:算力任务所需的算力总量、算力任务所需的算力类型、算力任务的期望完成时间、算力任务所需的算法模型信息;
    接收来自所述终端的算力任务描述信息,所述算力任务描述信息用于确定所述算力需求信息,所述算力任务描述信息包括以下至少一项:算力任务类型、算力任务所需的算法模型的类型、算力任务的任务量描述、算力任务的期望完成时间范围、算力任务的期望完成成功率。
  25. 根据权利要求24所述的装置,其中,所述确定模块,还用于以下至少一项:
    在所述算力请求包括所述算力需求信息的情况下,基于所述算力需求信息确定所述目标算力设备;
    在所述算力请求包括所述算力任务描述信息的情况下,基于所述算力任务描述信息确定所述算力需求信息;基于所述算力需求信息确定所述目标算力设备。
  26. 根据权利要求25所述的装置,其中,所述确定模块,还用于:
    预先获得多个算力设备的算力状态信息,所述算力状态信息包括以下至少一项:算力总量、算力使用量、算力类型、可用算力;
    将所述算力需求信息与所述多个算力设备的算力状态信息进行匹配,确定相匹配的至少一个候选算力设备;
    基于所述至少一个候选算力设备确定所述目标算力设备。
  27. 根据权利要求26所述的装置,其中,所述确定模块,还用于:
    向所述至少一个候选算力设备发送算力预留请求;针对任一候选算力设备,若接收到所述候选算力设备返回的预留算力成功信息,则将所述候选算力设备确定为所述目标算力设备;或,
    基于算力设备的优先级从高到低的顺序,向所述至少一个候选算力设备发送算力预留请求;将所述至少一个候选算力设备中的第一候选算力设备确定为所述目标算力设备,所述第一候选算力设备为接收到所述算力预留请求并返回预留算力成功信息的算力设备;所述算力设备的优先级基于以下至少一项确定:算力设备的算力总量、算力总量与算力任务的匹配程度、算力设备与所述终端之间的路由距离。
  28. 根据权利要求23所述的装置,其中,所述目标算力设备的个数为一个或多个;
    其中,在所述目标算力设备的个数为多个的情况下,所述发送模块,还用于:
    将多个目标算力设备的设备信息均发送给所述终端;或,
    确定所述多个目标算力设备中的主设备和辅设备;将所述主设备的设备信息发送给所述终端,以及将所述辅设备的设备信息发送给所述主设备,由所述终端基于所述主设备的设备信息将所述算力任务发送给所述主设备,并由所述主设备将所述算力任务拆分后基于所述辅设备的设备信息分配给所述辅设备,以共同执行所述算力任务。
  29. 根据权利要求23所述的装置,其中,
    所述接收模块,还用于接收来自所述终端的通知信息,所述通知信息由所述终端接收到所述目标算力设备返回的算力任务执行结果后,在确定所述 算力任务执行结果满足预设条件或基于所述算力任务执行结果确定所述算力任务执行成功的情况下发送;
    所述发送模块,还用于向所述目标算力设备发送算力释放请求,以释放所述目标算力设备为所述算力任务预留的算力。
  30. 一种算力执行装置,其中,包括:
    接收模块,用于接收来自终端的算力任务,所述算力任务由所述终端基于所述目标算力设备的设备信息发送,所述设备信息由网络侧设备在接收到所述终端的算力请求并确定所述目标算力设备后返回给所述终端,所述算力请求用于为所述终端中的算力任务请求算力,所述目标算力设备包括具有算力能力的设备或设备集群;
    任务执行模块,用于执行所述算力任务。
  31. 根据权利要求30所述的装置,其中,所述目标算力设备的个数为一个或多个,其中:
    在所述目标算力设备的个数为一个的情况下,所述任务执行模块,还用于:
    独立执行所述算力任务;或,
    将所述算力任务进行拆分后分配给其他算力设备,以共同执行所述算力任务,所述其他算力设备由所述网络侧设备确定,所述其他算力设备完成算力任务后,将算力任务执行结果发送至所述目标算力设备;
    在所述目标算力设备的个数为多个的情况下,所述任务执行模块,还用于:
    执行来自所述终端的算力子任务,所述算力子任务由所述终端对所述算力任务进行拆分后得到。
  32. 根据权利要求30所述的装置,其中,
    所述任务执行模块,还用于将算力任务执行结果发送给所述终端;
    所述接收模块,还用于接收来自所述网络侧设备的算力释放请求,所述 算力释放请求由所述网络侧设备接收到所述终端的通知信息后发送,所述通知信息由所述终端接收到所述算力任务执行结果后,在确定所述算力任务执行结果满足预设条件或基于所述算力任务执行结果确定所述算力任务执行成功的情况下发送给所述网络侧设备;基于所述算力释放请求,释放为所述算力任务预留的算力。
  33. 一种终端,其中,包括处理器,存储器及存储在所述存储器上并可在所述处理器上运行的程序或指令,所述程序或指令被所述处理器执行时实现如权利要求1至6任一项所述的算力请求方法的步骤。
  34. 一种网络侧设备,其中,包括处理器,存储器及存储在所述存储器上并可在所述处理器上运行的程序或指令,所述程序或指令被所述处理器执行时实现如权利要求7至13任一项所述的算力分配方法的步骤。
  35. 一种目标算力设备,其中,包括处理器,存储器及存储在所述存储器上并可在所述处理器上运行的程序或指令,所述程序或指令被所述处理器执行时实现如权利要求14至16任一项所述的算力执行方法的步骤。
  36. 一种可读存储介质,其中,所述可读存储介质上存储程序或指令,所述程序或指令被处理器执行时实现如权利要求1至6任一项所述的算力请求方法,或者实现如权利要求7至13任一项所述的算力分配方法的步骤,或者实现如权利要求14至16任一项所述的算力执行方法的步骤。
PCT/CN2022/112497 2021-08-18 2022-08-15 算力请求、算力分配、算力执行方法、终端及网络侧设备 WO2023020432A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202110950241.2A CN115714774A (zh) 2021-08-18 2021-08-18 算力请求、算力分配、算力执行方法、终端及网络侧设备
CN202110950241.2 2021-08-18

Publications (1)

Publication Number Publication Date
WO2023020432A1 true WO2023020432A1 (zh) 2023-02-23

Family

ID=85229988

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/112497 WO2023020432A1 (zh) 2021-08-18 2022-08-15 算力请求、算力分配、算力执行方法、终端及网络侧设备

Country Status (2)

Country Link
CN (1) CN115714774A (zh)
WO (1) WO2023020432A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116909758B (zh) * 2023-09-13 2024-01-26 中移(苏州)软件技术有限公司 算力任务的处理方法、装置及电子设备

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150301814A1 (en) * 2012-12-31 2015-10-22 Huawei Technologies Co., Ltd. Application deployment method and scheduler
CN110532092A (zh) * 2019-08-21 2019-12-03 云湾科技(嘉兴)有限公司 资源匹配方法、装置及设备
CN110851529A (zh) * 2019-11-01 2020-02-28 腾讯科技(深圳)有限公司 算力调度方法及相关设备
US20200195524A1 (en) * 2018-08-24 2020-06-18 Embotics Corporation Distributed multicloud service placement engine and method therefor
CN111866162A (zh) * 2020-07-28 2020-10-30 中国联合网络通信集团有限公司 一种业务分配方法及装置
CN112799789A (zh) * 2021-03-22 2021-05-14 腾讯科技(深圳)有限公司 一种节点集群管理方法、装置、设备及存储介质

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150301814A1 (en) * 2012-12-31 2015-10-22 Huawei Technologies Co., Ltd. Application deployment method and scheduler
US20200195524A1 (en) * 2018-08-24 2020-06-18 Embotics Corporation Distributed multicloud service placement engine and method therefor
CN110532092A (zh) * 2019-08-21 2019-12-03 云湾科技(嘉兴)有限公司 资源匹配方法、装置及设备
CN110851529A (zh) * 2019-11-01 2020-02-28 腾讯科技(深圳)有限公司 算力调度方法及相关设备
CN111866162A (zh) * 2020-07-28 2020-10-30 中国联合网络通信集团有限公司 一种业务分配方法及装置
CN112799789A (zh) * 2021-03-22 2021-05-14 腾讯科技(深圳)有限公司 一种节点集群管理方法、装置、设备及存储介质

Also Published As

Publication number Publication date
CN115714774A (zh) 2023-02-24

Similar Documents

Publication Publication Date Title
WO2018082709A1 (zh) 移动边缘计算应用负荷分担的方法和控制器
WO2020015643A1 (zh) 用于监听pdcch的方法、终端及网络设备
TWI640214B (zh) 一種網路接入方法、相關設備和系統
WO2022156752A1 (zh) 数据传输方法、终端及网络侧设备
US20230354237A1 (en) Method and apparatus for determining transmission delay, device, and storage medium
CN109155939A (zh) 一种负载迁移方法、装置及系统
WO2023020432A1 (zh) 算力请求、算力分配、算力执行方法、终端及网络侧设备
Pratap et al. Bandwidth-constrained task throughput maximization in IoT-enabled 5G networks
WO2023246757A1 (zh) 算力服务方法、装置及终端
CN108270839B (zh) 访问频率控制系统及方法
WO2015062400A1 (zh) 视频通道分配管理方法和相关设备及通信系统
WO2023020431A1 (zh) 算力资源的确定方法和设备
WO2023016460A1 (zh) 计算任务的策略确定或资源分配方法、装置、网元及介质
CN112565390A (zh) 服务分配方法、装置、电子设备及存储介质
CN116596202A (zh) 工单处理方法、相关装置及存储介质
CN111245878A (zh) 一种基于混合云计算和雾计算的计算卸载通信网络的方法
CN114679750A (zh) 切换小区的方法、终端设备和网络设备
CN114691351A (zh) 信息处理方法、装置、设备及存储介质
WO2023016396A1 (zh) 计算会话更新方法、装置及通信设备
WO2023016461A1 (zh) 计算任务的资源分配方法、装置、网元及介质
WO2023280067A1 (zh) 一种算力信息获取方法及相关设备
WO2023246584A1 (zh) 算力处理方法、装置及通信设备
WO2023143442A1 (zh) 路径偏好确定方法、终端及网络侧设备
KR102059372B1 (ko) 무선 네트워크에서 가상 도킹을 가능하게 하기 위한 방법 및 장치
WO2023143421A1 (zh) 中继切换处理方法、装置、远端终端、中继终端及网络侧设备

Legal Events

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

Ref document number: 22857744

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 2022857744

Country of ref document: EP

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2022857744

Country of ref document: EP

Effective date: 20240318