Task execution method and device, electronic equipment, computer readable storage medium and computer program product

Abstract

The invention provides a task execution method and device, electronic equipment, a computer readable storage medium and a computer program product. The method comprises the steps of obtaining a task execution request; based on the first air interface user plane protocol stack, executing a target task corresponding to the task execution request; wherein the first air interface user plane protocol stack is obtained by performing at least one of the following processing on the second air interface user plane protocol stack: adding a calculation task management CTM layer; adding a first service quality mapping and updating QMU layer; adding a first calculation execution CE layer; adding a first computing connection collaboration CCC layer; updating a service data adaptation protocol SDAP layer in the second air interface user plane protocol stack; and updating a packet data convergence protocol PDCP layer in the second air interface user plane protocol stack. And the task execution effect is improved.

Description

Technical Field

[0001] This application relates to the field of data processing, and more particularly to a task execution method, apparatus, electronic device, computer-readable storage medium, and computer program product. Background Technology

[0002] Currently, the air interface user plane protocol stack in related technologies (e.g., the 5G air interface user plane protocol stack) mainly includes the following layers: Physical Layer, responsible for transforming uplink and downlink data and channel scheduling at the physical layer; MAC Layer (Medium Access Control Layer), responsible for assembling and segmenting data frames and coordinating with the Physical Layer; RLC Layer (Radio Link Control Layer), responsible for secure data transmission and data segment splicing, and coordinating with the MAC Layer; PDCP Layer (Packet Data Convergence Protocol Layer), mainly responsible for data compression, encryption, and decryption to ensure security and integrity, and also coordinating with the RLC Layer; and SDAP Layer (Service Data Adaptation Protocol Layer), responsible for mapping different service data to QoS streams and data radio bearers.

[0003] In edge-end collaborative tasks, the task can be divided into multiple parts. One part performs inference computation within the terminal device, while another part performs inference computation on the network-side device, transmitting intermediate computation results via air interface uplink and downlink channels. For such tasks, not only transmission but also computation is required, necessitating a shift from connection-oriented (corresponding to communication services) to task-oriented approaches. However, the air interface user plane protocol stacks in the aforementioned related technologies only consider the transmission performance of the computation results, ensuring transmission performance. But for these tasks, in addition to transmission, computation must also be considered. Executing such tasks using the air interface user plane protocol stacks in the aforementioned related technologies can easily lead to poor task execution results. Summary of the Invention

[0004] This application provides a task execution method, apparatus, and electronic device to solve the problem that existing computational partitioning graph adjustments can easily lead to the loss of task computation data.

[0005] To solve the above-mentioned technical problems, this application is implemented as follows:

[0006] In a first aspect, embodiments of this application provide a task execution method applied to a terminal device, the method comprising:

[0007] Obtain the task execution request;

[0008] Based on the first air interface user plane protocol stack, execute the target task corresponding to the task execution request;

[0009] The first air interface user plane protocol stack is obtained by performing at least one of the following processes on the second air interface user plane protocol stack: adding a Computation Task Management (CTM) layer; adding a first Quality of Service Mapping and Update (QMU) layer; adding a first Computation Execution (CE) layer; adding a first Computation Connection Coordination (CCC) layer; updating the Service Data Adaptation Protocol (SDAP) layer in the second air interface user plane protocol stack; and updating the Packet Data Convergence Protocol (PDCP) layer in the second air interface user plane protocol stack.

[0010] The CTM layer is used for computation task management; the first QMU layer is used for at least one of the following: task service quality (QoS) requirement division, QoS requirement update, and QoS evaluation; the first CE layer is used for task computation; the first CCC layer is used for compressing different task data packets with different weights; the updated SDAP layer is used for mapping the QoS flow of task data packets to the computation radio bearer (CRB); and the updated PDCP layer is used for encrypting and / or decrypting the data body in the task data packets.

[0011] Secondly, embodiments of this application provide a task execution method applied to a network device, the method comprising:

[0012] Execute the target task based on the third air interface user plane protocol stack;

[0013] The third air interface user plane protocol stack is obtained by performing at least one of the following processes on the second air interface user plane protocol stack: adding a second quality of service mapping and update (QMU) layer; adding a second computation execution (CE) layer; adding a second computation connection coordination (CCC) layer; updating the service data adaptation protocol (SDAP) layer in the second air interface user plane protocol stack; and updating the packet data aggregation protocol (PDCP) layer in the second air interface user plane protocol stack.

[0014] The second QMU layer is used for at least one of the following: QoS requirement division, QoS requirement update, and QoS assessment; the second CE layer is used to perform task calculation; the second CCC layer is used to compress different task data packets with different weights; the updated SDAP layer is used to calculate the reverse mapping from radio bearer CRB to QoS flow and / or the mapping from QoS flow to CRB for task data packets; and the updated PDCP layer is used to encrypt and / or decrypt the data body in the task data packets.

[0015] Thirdly, embodiments of this application provide a task execution apparatus, the apparatus comprising:

[0016] The first acquisition module is used to acquire task execution requests;

[0017] The first task execution module is used to execute the target task corresponding to the task execution request based on the first air interface user plane protocol stack;

[0018] The first air interface user plane protocol stack is obtained by performing at least one of the following processes on the second air interface user plane protocol stack: adding a Computation Task Management (CTM) layer; adding a first Quality of Service Mapping and Update (QMU) layer; adding a first Computation Execution (CE) layer; adding a first Computation Connection Coordination (CCC) layer; updating the Service Data Adaptation Protocol (SDAP) layer in the second air interface user plane protocol stack; and updating the Packet Data Convergence Protocol (PDCP) layer in the second air interface user plane protocol stack.

[0019] The CTM layer is used for computation task management; the first QMU layer is used for at least one of the following: task service quality (QoS) requirement division, QoS requirement update, and QoS evaluation; the first CE layer is used for task computation; the first CCC layer is used for compressing different task data packets with different weights; the updated SDAP layer is used for mapping the QoS flow of task data packets to the computation radio bearer (CRB); and the updated PDCP layer is used for encrypting and / or decrypting the data body in the task data packets.

[0020] Fourthly, embodiments of this application provide a task execution apparatus, the apparatus comprising:

[0021] The second task execution module is used to execute the target task based on the third air interface user plane protocol stack;

[0022] The third air interface user plane protocol stack is obtained by performing at least one of the following processes on the second air interface user plane protocol stack: adding a second quality of service mapping and update (QMU) layer; adding a second computation execution (CE) layer; adding a second computation connection coordination (CCC) layer; updating the service data adaptation protocol (SDAP) layer in the second air interface user plane protocol stack; and updating the packet data aggregation protocol (PDCP) layer in the second air interface user plane protocol stack.

[0023] The second QMU layer is used for at least one of the following: QoS requirement division, QoS requirement update, and QoS assessment; the second CE layer is used to perform task calculation; the second CCC layer is used to compress different task data packets with different weights; the updated SDAP layer is used to calculate the reverse mapping from radio bearer CRB to QoS flow and / or the mapping from QoS flow to CRB for task data packets; and the updated PDCP layer is used to encrypt and / or decrypt the data body in the task data packets.

[0024] Fifthly, embodiments of this application provide an electronic device, including a transceiver and a processor.

[0025] The processor is used for:

[0026] Obtain the task execution request;

[0027] Based on the first air interface user plane protocol stack, execute the target task corresponding to the task execution request;

[0028] The first air interface user plane protocol stack is obtained by performing at least one of the following processes on the second air interface user plane protocol stack: adding a Computation Task Management (CTM) layer; adding a first Quality of Service Mapping and Update (QMU) layer; adding a first Computation Execution (CE) layer; adding a first Computation Connection Coordination (CCC) layer; updating the Service Data Adaptation Protocol (SDAP) layer in the second air interface user plane protocol stack; and updating the Packet Data Convergence Protocol (PDCP) layer in the second air interface user plane protocol stack.

[0029] The CTM layer is used for computation task management; the first QMU layer is used for at least one of the following: task service quality (QoS) requirement division, QoS requirement update, and QoS evaluation; the first CE layer is used for task computation; the first CCC layer is used for compressing different task data packets with different weights; the updated SDAP layer is used for mapping the QoS flow of task data packets to the computation radio bearer (CRB); and the updated PDCP layer is used for encrypting and / or decrypting the data body in the task data packets.

[0030] Sixthly, embodiments of this application provide an electronic device, including a transceiver and a processor.

[0031] The processor is used for:

[0032] Execute the target task based on the third air interface user plane protocol stack;

[0033] The third air interface user plane protocol stack is obtained by performing at least one of the following processes on the second air interface user plane protocol stack: adding a second quality of service mapping and update (QMU) layer; adding a second computation execution (CE) layer; adding a second computation connection coordination (CCC) layer; updating the service data adaptation protocol (SDAP) layer in the second air interface user plane protocol stack; and updating the packet data aggregation protocol (PDCP) layer in the second air interface user plane protocol stack.

[0034] The second QMU layer is used for at least one of the following: QoS requirement division, QoS requirement update, and QoS assessment; the second CE layer is used to perform task calculation; the second CCC layer is used to compress different task data packets with different weights; the updated SDAP layer is used to calculate the reverse mapping from radio bearer CRB to QoS flow and / or the mapping from QoS flow to CRB for task data packets; and the updated PDCP layer is used to encrypt and / or decrypt the data body in the task data packets.

[0035] In a seventh aspect, embodiments of this application provide an electronic device, including: a processor, a memory, and a program stored in the memory and executable on the processor, wherein when the program is executed by the processor, it implements the steps of the task execution method described in the first aspect above.

[0036] Eighthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the task execution method described in the first aspect.

[0037] Ninthly, embodiments of this application provide a computer program product, including computer instructions, which, when executed by a processor, implement the steps of the method described above.

[0038] In this embodiment, the improved first air interface user plane protocol stack is determined by performing at least one of the above-mentioned processes on the basis of the second air interface user plane protocol stack. That is, the improved first air interface user plane protocol stack not only continues the functionality of the second air interface user plane protocol stack and takes transmission performance into consideration, but also adds at least one layer from the CTM layer, the first QMU layer, the first CE layer, and the first CCC layer. It can also update the SDAP and PDCP layers in the second air interface user plane protocol stack to support task computation. In this way, the improved first air interface user plane protocol stack is more adaptable to task execution, thereby improving task execution efficiency when executing the target task based on the improved first air interface user plane protocol stack. Attached Figure Description

[0039] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 This is one of the flowcharts of a task execution method provided in the embodiments of this application;

[0041] Figure 2 This is one of the flowcharts of a task execution method provided in the embodiments of this application;

[0042] Figure 3 This is a schematic diagram of an air interface user plane protocol stack for AI computing services oriented towards edge-end collaboration, provided in an embodiment of this application.

[0043] Figure 4 This is a schematic diagram of the protocol layers, bearers, and channels that the computation task data packet (CTP) passes through after reaching the C_SDAP layer, as provided in the embodiments of this application.

[0044] Figure 5 This is a schematic diagram of the protocol entities and configurations included in the CRB and DRB provided in the embodiments of this application;

[0045] Figure 6 This is a 5G MAC layer subheader format;

[0046] Figure 7 This application provides a 6G-added MAC layer subheader format;

[0047] Figure 8 This is one of the user plane processes for edge-end collaborative AI inference services provided in the embodiments of this application;

[0048] Figure 9 This is the second user plane process for edge-to-edge collaborative AI inference services provided in this application embodiment;

[0049] Figure 10 It is the CE layer PDU format provided in the embodiments of this application;

[0050] Figure 11 This is the downlink Data PDU format with C_SDAP header provided in the embodiments of this application;

[0051] Figure 12 It is a 5G PDCP layer user plane PDU format;

[0052] Figure 13 This is the 6G C_PDCP layer user plane PDU format provided in the embodiments of this application;

[0053] Figure 14 This is one of the structural schematic diagrams of a task execution device provided in the embodiments of this application;

[0054] Figure 15 This is one of the structural schematic diagrams of a task execution device provided in the embodiments of this application;

[0055] Figure 16 This is one of the structural schematic diagrams of an electronic device provided in the embodiments of this application;

[0056] Figure 17 This is a second schematic diagram of the structure of an electronic device provided in the embodiments of this application. Detailed Implementation

[0057] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0058] See Figure 1 , Figure 1 This is a flowchart of a task execution method provided in an embodiment of this application, which can be applied to terminal devices. For example... Figure 1 As shown, the task execution method provided in this embodiment includes the following steps:

[0059] Step 101: Obtain the task execution request.

[0060] If a target task needs to be executed, a task execution request can be generated on the terminal device. This task execution request can be used to request the execution of the target task. The target task can be any edge-to-edge collaborative task without specific limitations. For example, the target task can be, but is not limited to, an artificial intelligence (AI) inference task (also known as an AI model inference task).

[0061] Step 102: Based on the first air interface user plane protocol stack, execute the target task corresponding to the task execution request;

[0062] Specifically, the first air interface user plane protocol stack is obtained by performing at least one of the following processes on the second air interface user plane protocol stack: adding a Computing Task Management (CTM) layer; adding a first Quality of Service Mapping and Updating (QMU) layer; adding a first Computing Execution (CE) layer; adding a first Computing Connection Cooperating (CCC) layer; updating the Service Data Adaptation Protocol (SDAP) layer in the second air interface user plane protocol stack; and updating the Packet Data Convergence Protocol (C_PDCP) layer in the second air interface user plane protocol stack.

[0063] The CTM layer is used for computation task management; the first QMU layer is used for at least one of the following: task service quality (QoS) requirement division, QoS requirement update, and QoS evaluation; the first CE layer is used for task computation; the first CCC layer is used to compress different task data packets with different weights; the updated SDAP layer is used to map the QoS flow of task data packets to the computational radio bearer (CRB); and the updated PDCP layer is used to encrypt and / or decrypt the data body in the task data packets.

[0064] It should be understood that the second air interface user plane protocol stack can be an existing air interface user plane protocol stack, such as the existing fifth-generation mobile communication (5G) air interface user plane protocol stack. The task execution process involves task computation. The existing air interface user plane protocol stack only considers transmission performance and cannot be adapted to task execution. In this embodiment, the existing air interface user plane protocol stack can be improved. For example, new functions can be introduced on the basis of the second air interface user plane protocol stack. For example, at least one of the CTM layer, the first QMU layer, the first CE layer, and the first CCC layer can be added. Alternatively, the existing protocol layers in the second air interface user plane protocol stack can be updated. For example, at least one of the existing SDAP layer and PDCP layer in the second air interface user plane protocol stack can be updated. The updated SDAP layer can also be called the Computing Service Data Adaptation Protocol (C_SDAP) layer, and the updated PDCP layer can also be called the Computing Packet Data Convergence Protocol (C_PDCP) layer.

[0065] In this embodiment, the improved first air interface user plane protocol stack is determined by performing at least one of the above-mentioned processes on the basis of the second air interface user plane protocol stack. That is, the improved first air interface user plane protocol stack not only retains the functionality of the second air interface user plane protocol stack and takes transmission performance into consideration, but also adds at least one layer from the CTM layer, the first QMU layer, the first CE layer, and the first CCC layer. It can also update the SDAP and PDCP layers in the second air interface user plane protocol stack to support task computation. In this way, the improved first air interface user plane protocol stack is more adaptable to task execution, thereby improving task execution efficiency when executing the target task based on the improved first air interface user plane protocol stack.

[0066] In some embodiments, based on the first air interface user plane protocol stack, the target task corresponding to the task execution request is executed, including at least one of the following:

[0067] The QoS requirements for the task are obtained through the CTM layer, and the target task is started. The target task includes multiple subtasks.

[0068] The QoS requirements of the task are divided by the first QMU layer, and multiple QoS sub-requirements are determined. Each of the multiple QoS sub-requirements corresponds to a multiple sub-task.

[0069] If any subtask data packet is obtained through the first QMU layer, the corresponding QoS sub-requirement is added to the subtask data packet to update the subtask data packet, and the updated subtask data packet is transmitted to the first CE layer.

[0070] The first CE layer executes the corresponding subtask in the target task on the updated subtask data packet to obtain the first intermediate calculation result of the updated subtask data packet.

[0071] It is understood that the QoS requirement of the task can be the task requirement of the target task. The CTM layer can maintain the QoS requirement of the target task. In this embodiment, the task can be executed according to the task QoS requirement. Specifically, the target task includes multiple sub-tasks. The first QMU layer can divide the QoS requirement and determine multiple QoS sub-requirements. The multiple QoS sub-requirements correspond one-to-one with the multiple sub-tasks, that is, each sub-task in the multiple sub-tasks is assigned a corresponding QoS sub-requirement.

[0072] If a subtask data packet (a data packet related to a subtask in the target task, used to execute the corresponding subtask, which can be an intermediate result data packet or other data packets required to execute the subtask, etc.) can be obtained through the first QMU layer, the QoS sub-requirements corresponding to the subtask data packet (i.e., the QoS sub-requirements corresponding to the subtask corresponding to the subtask data packet) can be added to the subtask data packet to update it, and the updated subtask data packet can be transmitted to the first CE layer. It should be noted that in various embodiments of this application, the first intermediate calculation result and the first calculation result can both be in vector form. For example, the first intermediate calculation result can be a first intermediate calculation result vector, which may include multiple data elements.

[0073] In this embodiment, the target task may include multiple subtasks. The QoS requirements of the task can be divided through the first QMU layer to determine multiple QoS sub-requirements that correspond one-to-one with the multiple subtasks. In this way, when a subtask data packet required to execute a certain subtask is received, the corresponding QoS sub-requirements can be added to the subtask data packet to achieve updating. The corresponding subtask can be executed through the first CE layer on the updated subtask data packet to obtain the first intermediate calculation result and complete the execution of the subtask. In this way, the execution of the subtask can meet the corresponding QoS sub-requirements, thereby improving the task execution effect.

[0074] In some embodiments, after obtaining the first intermediate calculation result of the updated subtask data packet, at least one of the following is further included:

[0075] The first intermediate calculation result is transmitted from the first CE layer to the first QMU layer.

[0076] The first intermediate calculation result is transmitted to the network device through the first CCC layer, the updated SDAP, and the updated PDCP. The first intermediate calculation result is used by the network device to execute the corresponding sub-task in the target task to obtain the first calculation result.

[0077] That is, after the first CE layer performs task calculations and obtains the first intermediate calculation result, it can transmit it to the first QMU. The first QMU can then transmit it to the first CCC layer. Through the first CCC layer, the updated SDAP, and the updated PDCP (i.e., the order from the first CCC layer to the updated SDAP to the updated PDCP), the first intermediate calculation result is transmitted to the network device, so that the network device can perform subtask execution on the first intermediate calculation result to obtain the first calculation result. It should be noted that the first air interface user plane protocol stack may also include a first RLC layer, a first MAC layer, and a first PHY layer. These layers may not be updated relative to the second air interface user plane protocol stack, that is, they may be the same as the RLC layer, MAC layer, and PHY layer in the second air interface user plane protocol stack. For example, the first intermediate calculation result can be transmitted to the network device in sequence through the first CCC layer, the updated SDAP, the updated PDCP, the first RLC layer, the first MAC layer, and the first PHY layer (i.e., the order from the first CCC layer to the updated SDAP, to the updated PDCP, to the first RLC layer, to the first MAC layer, and then to the first PHY layer).

[0078] In some embodiments, the plurality of subtasks includes at least one computation subtask and at least one transmission subtask, and the plurality of QoS sub-requirements includes the QoS sub-requirement of each task in at least one computation subtask and the QoS sub-requirement of each task in at least one transmission subtask.

[0079] After transmitting the first intermediate calculation result from the first CE layer to the first QMU layer, it also includes:

[0080] The QoS sub-requirements of at least one computation subtask and at least one transmission subtask are updated through the first QMU layer.

[0081] That is, after the first QMU receives the first intermediate calculation result, it can update the QoS sub-requirements of at least one calculation subtask and the QoS sub-requirements of at least one transmission subtask to make the QoS sub-requirements more adaptable to the current task and improve the task execution effect.

[0082] In some embodiments, transmitting the first intermediate computation result to the network device via the first CCC layer, the updated SDAP, and the updated PDCP includes:

[0083] The weight of the first intermediate calculation result is determined by the first CCC layer, and the first intermediate calculation result is compressed based on the weight of the first intermediate calculation result to obtain the first data packet;

[0084] The updated SDAP uses the QoS sub-requirements corresponding to the first data packet to perform QoS flow mapping to CRB for the first data packet, and generates a second data packet. The data body of the second data packet includes the first data packet.

[0085] The data body of the second data packet is encrypted using the updated PDCP to obtain the third data packet, which is then sent to the network device. The third data packet is used by the network device to perform corresponding sub-tasks based on the third data packet to obtain the first calculation result.

[0086] In this embodiment, the first CCC layer can utilize the weight of the first intermediate calculation result and compress the first intermediate calculation result based on the weight of the first intermediate calculation result to improve the compression accuracy. Moreover, the first CCC layer can perform QoS flow mapping to CRB for the first data packet according to the QoS sub-requirements corresponding to the first data packet to improve the mapping effect. Then, the updated PDCP encrypts the data body of the second data packet to obtain the third data packet and sends the third data packet to the network device to protect privacy and improve data transmission security.

[0087] For example, the first intermediate calculation result is in vector form. The weight of the first intermediate calculation result may include the weight of each data element in the first intermediate calculation result. During compression, compression can be performed according to the weight of each data element in the first intermediate calculation result to improve the accuracy and quality of task execution while reducing the overhead of computing and communication resources. The first data packet is the result of compressing the first intermediate calculation result. The first data packet can also be in vector form, which includes multiple elements. During the mapping process, mapping can be performed at the data packet level, that is, the first data packet is treated as a whole and mapped from QoS flow to CRB. For example, a data packet with high QoS requirements can be mapped to a CRB with high QoS guarantee. Alternatively, mapping can be performed at the element level, mapping different elements in the first data packet from QoS flow to CRB separately. In the process of mapping at different element levels, the weight of the element can also be considered (for example, the weight used in the compression mentioned above). For example, elements with high weights can be mapped to a CRB with high QoS guarantee.

[0088] In some embodiments, after transmitting the first intermediate calculation result to the network device, the method further includes:

[0089] The system receives a first calculation result sent by a network device. The first calculation result is either the task result of the target task or a second intermediate calculation result. The second intermediate calculation result is used by the terminal device to execute the corresponding sub-task in the target task to obtain the task result.

[0090] It should be noted that in various embodiments of this application, the second intermediate calculation result can be in the form of a vector. For example, the second intermediate calculation result can be a vector of second intermediate calculation results.

[0091] In this embodiment, the terminal device can perform a portion of the inference computation task and then submit the data to the network device to execute the remaining inference computation task. In this case, the final task is completed by the network device, and the terminal device only receives the required inference result from the network device. The first computation result is the task result. Alternatively, the terminal device can also perform a portion of the inference computation task and then submit the data to the network device to execute that portion. Due to user privacy requirements, the network device will not complete the entire inference task but will transmit the intermediate computation results to the terminal device. The final computation task needs to be executed by the terminal device to obtain the task result. This allows task execution to be completed through either of these two methods, improving the flexibility of task execution.

[0092] In some embodiments, the method further includes at least one of the following:

[0093] If the first calculation result is the second intermediate calculation result, the corresponding subtask in the target task is executed on the second intermediate calculation result data packet through the first CE layer to obtain the task result;

[0094] If the first calculation result data is the task result, the task result is transmitted to the CTM layer through the first CE layer, and the target task execution is terminated through the CTM layer.

[0095] If the first calculation result is the second intermediate calculation result, it means that the network device has only executed part of the task and has not completed all the remaining tasks. In this case, after receiving the second intermediate calculation result, the terminal device can continue to execute the unfinished sub-tasks and obtain the task result. If the first calculation result is the task result, it means that the network device has completed all the remaining sub-tasks except for the tasks already executed by the terminal device, and obtains the task result, which is then transmitted to the terminal device. In this case, the terminal device receives the task result, indicating that the target task is completed, and can then end the task execution.

[0096] In some embodiments, when the first calculation result is a second intermediate calculation result, the method further includes:

[0097] The QoS sub-requirements of at least one computational subtask are updated through the first QMU layer.

[0098] That is, after the first QMU receives the first intermediate calculation result, it can update the QoS sub-requirements of at least one calculation subtask to make the QoS sub-requirements of the calculation task more suitable for the current task, so as to improve the task execution effect.

[0099] In some embodiments, the third data packet includes encryption algorithm indication information, which is used to indicate the target encryption algorithm used by the data body of the second data packet.

[0100] This allows the network device to receive the third data packet and, based on the target encryption algorithm indicated by the encryption algorithm instruction information, decide whether to decrypt the third data packet, thus facilitating subsequent processing of the third data packet.

[0101] See Figure 2 , Figure 2 This is a flowchart of a task execution method provided in an embodiment of this application, applied to a network device, such as... Figure 2 As shown, the task execution method provided in this embodiment includes the following steps:

[0102] Step 201: Execute the target task based on the third air interface user plane protocol stack;

[0103] The third air interface user plane protocol stack is obtained by performing at least one of the following processes on the second air interface user plane protocol stack: adding a second quality of service mapping and updating QMU layer; adding a second computation execution CE layer; adding a second computation connection coordination CCC layer; updating the service data adaptation protocol SDAP layer in the second air interface user plane protocol stack; and updating the packet data aggregation protocol PDCP layer in the second air interface user plane protocol stack.

[0104] The second QMU layer is used for at least one of the following: QoS requirement division, QoS requirement update, and QoS assessment; the second CE layer is used to perform task calculation; the second CCC layer is used to compress different task data packets with different weights; the updated SDAP layer is used to calculate the reverse mapping from radio bearer CRB to QoS flow and / or the mapping from QoS flow to CRB for task data packets; and the updated PDCP layer is used to encrypt and / or decrypt the data body in the task data packets.

[0105] In some embodiments, the target task includes multiple subtasks;

[0106] Based on the third air interface user plane protocol stack, execute the target tasks, including:

[0107] Receive a third data packet sent by a terminal device; and at least one of the following:

[0108] The first calculation result is obtained by performing the corresponding sub-task in the target task on the third data packet through the second CE layer;

[0109] The first calculation result is sent to the terminal device through the second CCC layer, the updated SDAP, and the updated PDCP.

[0110] In some embodiments, multiple subtasks correspond one-to-one with multiple QoS sub-requirements. The multiple subtasks include at least one computation subtask and at least one transmission subtask. The multiple QoS sub-requirements include the QoS sub-requirements of each task in at least one computation subtask and the QoS sub-requirements of each task in at least one transmission subtask.

[0111] After receiving the first intermediate calculation result sent by the terminal device, the method further includes:

[0112] The QoS sub-requirements of at least one computation subtask and at least one transport subtask are updated through the second QMU layer.

[0113] In some embodiments, sending the first calculation result to the terminal device via the second CCC layer, the updated SDAP, and the updated PDCP includes:

[0114] The weight of the first calculation result is determined by the second CCC layer, and the first calculation result is compressed based on the weight of the first calculation result to obtain the fourth data packet;

[0115] By using the updated SDAP to perform a reverse mapping from CRB to QoS flow for the fourth data packet corresponding to the QoS sub-requirement, a fifth data packet is generated. The data body of the fifth data packet includes the fourth data packet.

[0116] The data body of the fifth data packet is encrypted using the updated PDCP to obtain the sixth data packet, which is then sent to the terminal device. The sixth data packet is used by the terminal device to obtain the first calculation result.

[0117] In some embodiments, the first calculation result is the task result of the target task or the second intermediate calculation result. The second intermediate calculation result is used by the terminal device to execute the corresponding sub-task in the target task to obtain the task result.

[0118] In some embodiments, the third data packet includes encryption algorithm indication information, which is used to indicate the target encryption algorithm.

[0119] In some embodiments, the first calculation result is obtained by performing a corresponding subtask in the target task on the third data packet through the second CE layer, including at least one of the following:

[0120] When the target encryption algorithm indicated by the encryption algorithm indication information is a homomorphic encryption algorithm, the third data packet is reverse-mapped from CRB to QOS stream through the updated SDAP to obtain the seventh data packet. The corresponding subtask in the target task is executed on the seventh data packet through the second CE layer to obtain the first calculation result.

[0121] When the target encryption algorithm indicated by the encryption algorithm indication information is a non-homomorphic encryption algorithm, the third data packet is decrypted through the updated PDCP to obtain the eighth data packet; the eighth data packet is reverse-mapped from CRB to QOS stream through the updated SDAP to obtain the ninth data packet; and the corresponding subtask in the target task is executed through the second CE layer on the ninth data packet to obtain the first calculation result.

[0122] The process of the above method will be specifically described below with some specific embodiments.

[0123] Introduction to related technologies:

[0124] The OSI seven-layer model, or Open Systems Interconnection Reference Model, provides a functional structural framework for open, interconnected information systems. From lowest to highest, it consists of: Physical Layer, Data Link Layer, Network Layer, Transport Layer, Session Layer, Presentation Layer, and Application Layer. Network products from different manufacturers can achieve interconnection, interoperability, and portability as long as they conform to this reference model. In other words, any system that follows the OSI standard can communicate with each other as long as they are physically connected.

[0125] TCP / IP Transport Protocol: Also known as Transmission Control / Network Protocol, or network communication protocol, it is the most fundamental communication protocol used in network operations. TCP / IP specifies the standards and methods for communication between different parts of the Internet. Furthermore, TCP / IP is one of the two crucial protocols for ensuring the timely and complete transmission of network data. Strictly speaking, TCP / IP has a four-layer architecture, encompassing the application layer, transport layer, network layer, and data link layer.

[0126] 5G User Plane Protocol Stack: The 5G user plane protocol stack structure can be roughly divided into the following layers: Physical Layer, responsible for uplink and downlink data transformation and physical channel scheduling; MAC Layer (Medium Access Control Layer), responsible for data frame assembly and segmented transmission, as well as coordination with the Physical Layer; RLC Layer (Radio Link Control Layer), responsible for secure data transmission and data segment concatenation, as well as coordination with the MAC Layer; PDCP Layer (Packet Data Convergence Protocol Layer), mainly responsible for data compression, encryption, and decryption, ensuring security and integrity, and also coordinating with the RLC Layer; SDAP Layer (Service Data Adaptation Protocol Layer), responsible for mapping different service data to QoS streams and data radio bearers.

[0127] The OSI seven-layer model has several drawbacks: Firstly, it is purely a theoretical model, neglecting resource and technology availability, thus limiting its practical implementation. Secondly, the OSI model and its related service definitions and protocols are extremely complex. The data link layer and network layer have numerous functions, divided into multiple sublayers, making the overall structure bloated. Thirdly, functions such as addressing, flow control, and error control are repeated across different layers, reducing system efficiency. The repetition of each function across every layer leads to an overall decrease in efficiency.

[0128] Disadvantages of the TCP / IP protocol suite: The TCP / IP protocol suite itself lacks built-in security mechanisms, making it vulnerable to external attacks such as man-in-the-middle attacks, IP spoofing, and port scanning. TCP / IP does not provide a guarantee for Quality of Service (QoS), which may cause priority services (such as voice and video) to be interfered with or delayed by other services. Furthermore, it does not clearly distinguish between protocols, interfaces, and services, making it unsuitable for mobile communication networks.

[0129] Regarding the 5G user plane protocol stack in related technologies:

[0130] The transition from 5G to 6G requires a shift from connection-oriented to task-oriented approaches. For the user plane of air interface transmission, the endpoints of communication services and AI inference services differ. Communication services connect end-to-end, terminating at the server hosting the service, while AI inference tasks are completed on the wireless side, with the user plane terminating within the wireless network. Protocol stack processing related to IP packet forwarding is no longer needed and can be simplified. Furthermore, AI computing services may require the network side to use private data for computation, a function that existing 5G encryption methods cannot achieve; therefore, improvements to privacy protection and encryption / decryption schemes at the PDCP layer are needed.

[0131] From a QoS assurance perspective, multiple aspects need to be designed, including the control plane, user plane, and end-to-end processes. Specifically, the QoS of communication services and computing services differ in terms of metric design, mapping mechanisms, and assurance mechanisms. In terms of metric design, communication services use metrics such as GBR, MBR, and latency. The air interface user plane needs to guarantee bit-level success rate of communication data transmission, with optimization targets of BER or BLER. In contrast, the user plane for AI tasks needs to consider both air interface transmission and computing performance, with optimization targets of AI task performance metrics (such as inference latency or accuracy). Regarding mapping mechanisms, AI tasks need to be decomposed into two QoS components: transmission and computation. As for assurance mechanisms, communication services guarantee 5QI through parameter configuration at each layer of the protocol stack, the specific effect depending on implementation. AI tasks, because they involve both transmission and computation, and have high real-time and high-performance QoS requirements, require QoS evaluation and updates during the process. The method proposed in this application can solve the mapping and assurance mechanism problems of the user plane, adding a QoS mapping and update layer to achieve QoS assurance for end-edge collaborative AI inference tasks.

[0132] From a resource coordination perspective, since communication and AI computing resources will be coordinated and scheduled over the air interface, the execution and connection of AI computing require joint design, namely, a new design for both the control plane and the user plane. The control plane needs to be extended based on RRC to achieve coordinated control of communication and computing resources; the user plane needs to coordinate the computing execution process and the transmission of computing results to improve storage and transmission efficiency; therefore, a new computing connection coordination layer needs to be designed.

[0133] From a data importance perspective, different data points within a single communication task in 5G protocols have equal importance, with consistent accuracy and integrity requirements. However, for AI inference tasks, different data points within the same task may have varying degrees of importance, impacting task execution and completion to varying degrees. Using the existing protocol stack design would lead to inefficiency and resource waste. Therefore, it is necessary to improve the existing SDAP layer to map computational data of different granularities to computational wireless bearers, providing differentiated QoS guarantees for different data points within the same AI inference task.

[0134] In general, the existing 5G RAN needs to be extended in the user plane and a new computing plane needs to be added. This patent adds a QoS mapping and update layer, a computing service data adaptation protocol layer, etc. to extend the user plane, and incorporates computing execution functions as the content of the computing plane. For future wireless network scenarios (e.g., 6G), it proposes an air interface user plane protocol stack scheme that can support the establishment, connection, and execution of AI inference tasks for end-to-end collaborative AI inference services.

[0135] This solution addresses 6G wireless network scenarios and proposes an air interface user plane function design for end-edge collaborative AI inference services. This user plane, as shown... Figure 1 As shown, the protocol stack is divided into the UE side and the network side. The UE side comprises nine layers: Computing Task Management (CTM), Quality of Service Mapping and Updating (QMU), Computing Execution (CE), Computing Connection Cooperating (CCC), Computing Service Data Adaptation Protocol (C_SDAP), Computing Packet Data Convergence Protocol (C_PDCP), Radio Link Control (RLC), Media Access Control (MAC), and Physical Layer (PHY). The network side protocol stack consists of eight layers excluding CTM.

[0136] like Figure 3 As shown, the proposed embodiment of this application introduces new functions on the basis of the existing 5G air interface user plane protocol stack, such as CTM, QMU, CE, and CCC. The CTM layer controls and manages the computing tasks; the QMU layer is responsible for the mapping and updating of the QoS of specific computing tasks; the CE layer is responsible for the execution of computing tasks (which may be related to implementation), including the deployment and migration of specific tasks; and the CCC layer is responsible for the coordination and buffering between computing execution and computing transmission, maintaining data queues, evaluating the importance of different elements in the intermediate computing result vector and performing quantization and compression.

[0137] Furthermore, this application embodiment can also improve and optimize the existing SDAP layer and PDCP layer protocols according to the transmission characteristics of intermediate computing results of AI inference tasks, and update them to C_SDAP layer and C_PDCP layer, so as to better adapt to the needs of future computing services and the characteristics of AI data.

[0138] Considering the reuse of air interface wireless resources for AI computing transmission and communication data transmission, the RLC layer, MAC layer, and PHY layer protocols can remain unchanged to provide information transmission services for higher layers.

[0139] From the perspective of channels and radio bearers, this solution proposes a new logical channel—the Computing-Dedicated Traffic Channel (C-DTCH)—as the logical channel for AI computing tasks; and a new bearer—the Computing Radio Bearer (CRB)—as the radio bearer for AI computing task data. On the entity side, two new entities, C_SDAP Entity and C_PDCP Entity, are added to functionally adapt to the proposed protocol stack. Overall, this protocol stack can effectively support end-to-end collaborative AI computing services in 6G networks. Users and base stations will transmit air interface data under this protocol stack to jointly complete AI computing tasks.

[0140] Regarding the air interface user plane protocol stack scheme in this application embodiment:

[0141] In edge-end collaborative AI computing services, the AI ​​model is divided into multiple parts: one part performs inference computation within the terminal, and another part performs inference computation on the network side, transmitting intermediate computation results through air interface uplink and downlink channels. The air interface user plane protocol stack for edge-end collaborative AI computing services is also divided into two parts: the UE side and the network side.

[0142] UE side:

[0143] CTM Layer: The computing task management layer, a newly added layer, is innovative in its management of a new task type in 6G—computing tasks. Computing tasks differ significantly from communication tasks in many aspects, such as optimization objectives and resource scheduling. Therefore, a new CTM layer is needed to achieve independent management of computing tasks. The main functions of the CTM layer include:

[0144] Perform UE-side computing task management, control, and scheduling. Computing task control includes task start, end, and restart; computing task management includes maintaining the QoS requirements of computing tasks, evaluating whether the task QoS meets the requirements, prioritizing and scheduling tasks according to the task QoS requirements, handling tasks that do not meet the requirements, and handling the failure of the entire computing task.

[0145] The CTM layer is responsible to the upper layer, interacts with the application layer, receives data packets (and corresponding QoS requirements) submitted by the upper layer as input for the calculation task, and submits the calculation results and corresponding QoS evaluation values ​​to the upper layer.

[0146] QMU layer: Quality of Service Mapping and Update layer, is a newly added layer, and its functions are mainly related to QoS.

[0147] CE Layer: Computation Execution Layer. This is a newly added layer (potentially related to implementation). Computation tasks differ from communication tasks; they require deployment on specific devices and execution using computing resources. The main function of the CE layer is to handle the specific deployment and execution of computation tasks on the UE side. Uplink CE tasks use UE-side data for computation, while downlink CE tasks use intermediate computation results from the network side. Therefore, when AI computation is split between the UE and the network, the uplink CE must exist; when the termination layer of AI computation and the computation result are located on the network side, the UE directly receives the results from the network side, and the downlink CE is not necessary. The specific functions of the CE layer likely depend on the implementation, including:

[0148] Receive the computation input data packet and calculate the QoS requirements from the upper layer;

[0149] The computation process of each AI computing task being divided into subtasks for that layer;

[0150] Submit the calculation results to the lower-level functions;

[0151] The requirements for protocol design include:

[0152] The task to which the computation result belongs, the data input, and whether it is an intermediate computation result or an inference result are added to the data packet as a header.

[0153] CCC layer: Computational connection collaboration layer. Its key functions are to evaluate and set the weights of different element values ​​in the intermediate computation result vector, and to quantize and compress them.

[0154] C_SDAP layer: Computation Service Data Adaptation Protocol layer, which is partially modified from the 5G SDAP layer. C_SDAP can realize the mapping of computation data of different granularities (different elements in CTP / intermediate computation result vector) to CRB.

[0155] C_PDCP layer: Computational Packet Data Convergence Protocol layer, which is partially modified from the 5G PDCP layer, mainly optimized in terms of encryption and privacy protection.

[0156] RLC layer: Services and functions are the same as the MAC layer of the 5G protocol stack.

[0157] MAC layer: Services and functions are the same as the 5G protocol stack MAC layer.

[0158] PHY layer: Services and functions are the same as the 5G protocol stack PHY layer.

[0159] Network device side:

[0160] QMU layer: Quality of Service Mapping and Update Layer.

[0161] CE Layer: Computation Execution Layer, its main function is to execute network-side computation tasks. The network-side CE layer does not distinguish between uplink and downlink. The specific functions of the CE layer likely depend on the implementation and may include:

[0162] Receive the computation input data packet and calculate the QoS requirements from the upper layer;

[0163] The computation process of each AI computing task being divided into subtasks for that layer;

[0164] The calculation results are then passed to the lower-level functions.

[0165] The requirements for protocol design include:

[0166] The task to which the computation result belongs, the data input, and whether it is an intermediate computation result or an inference result are added to the data packet as a header.

[0167] CCC layer: Its key function is to evaluate and set the weights of different element values ​​in the intermediate calculation result vector, and to quantize and compress them.

[0168] C_SDAP layer: Computing service data adaptation protocol layer.

[0169] C_PDCP layer: Computational Packet Data Convergence Protocol layer. The main differences between it and the PDCP layer in 5G include optimizations in encryption / decryption and privacy protection functions.

[0170] RLC layer: Services and functions are the same as the MAC layer of the 5G protocol stack.

[0171] MAC layer: Services and functions are the same as the 5G protocol stack MAC layer.

[0172] PHY layer: Services and functions are the same as the 5G protocol stack PHY layer.

[0173] The proposed solution improves the PDCP layer in terms of privacy protection and simplifies some functions, optimizing it into a C_PDCP layer:

[0174] UE side:

[0175] C_PDCP: Computational Packet Data Convergence Protocol Layer, partially modified from the 5G PDCP layer, mainly optimized in terms of encryption and privacy protection. In 5G, header compression and integrity protection are selectively configured by RRC. AI computing tasks do not require header compression, and the data integrity requirements are weaker than those of communication tasks. Therefore, header compression and integrity protection are abandoned to reduce packet overhead and processing time.

[0176] Key features of the uplink C_PDCP layer include:

[0177] Maintenance of C_PDCP SN;

[0178] Encryption and privacy protection are implemented for computation task data packets. The encryption and decryption functions of the NR PDCP layer typically use symmetric encryption algorithms, based on Snow 3G, AES, and ZUC. In 6G, the encryption of computation task data should consider homomorphic encryption algorithms. Homomorphic encryption can ensure that the result of computation on ciphertext is the same as the result of computation on plaintext after decryption, and it is generally an asymmetric encryption algorithm. Common homomorphic encryption algorithms include RSA and Paillier algorithm.

[0179] Encrypted and privacy-protected data packets are then delivered to the RLC layer.

[0180] The requirements for protocol design include:

[0181] Since the network side does not need to decrypt the data packet after receiving it when using homomorphic encryption algorithm, it can directly perform the calculation. Furthermore, the result of the calculation does not need to be encrypted again when transmitted to the UE. Therefore, a new field should be added to indicate whether the data packet uses homomorphic encryption, so that the network side process can match it.

[0182] Key features of downlink include:

[0183] Maintenance of C_PDCP SN;

[0184] Decrypt the computation task data packet;

[0185] The decrypted data packet is then handed over to the C_SDAP layer.

[0186] Network side:

[0187] C_PDCP: Computational Packet Data Convergence Protocol Layer. Its main difference from the PDCP layer in 5G lies in the optimization of encryption / decryption and privacy protection functions. Key uplink functions include:

[0188] Maintenance of C_PDCP SN;

[0189] Decrypt the computation task data packets that need to be decrypted (decryption is not required when the UE side uses homomorphic encryption algorithm);

[0190] The processed data packet is then handed over to the C_SDAP layer.

[0191] Key features of downlink include:

[0192] Maintenance of C_PDCP SN;

[0193] Privacy protection and encryption are performed on the data packets for computing tasks (encryption is not required when the UE side uses homomorphic encryption algorithm);

[0194] Encrypted and privacy-protected data packets are then delivered to the RLC layer.

[0195] The requirements for protocol design include:

[0196] The new field indicates whether homomorphic encryption was used for the data packet.

[0197] For the newly added QMU layer in this application's design, QoS requirements are mapped and updated:

[0198] UE side:

[0199] QMU Layer: This is a newly added layer, primarily concerned with QoS (Quality of Service) mapping and updating. While 5G only has QoS related to communication transmission, computational tasks require consideration of computational QoS. The innovative concept of this layer is to break down the overall QoS requirements of computational tasks into transmission and computational aspects, and to perform QoS updates and evaluations after the computational task is executed / data packets are transmitted over the air interface. This solution proposes establishing a 6QI (6G QoS Identifier) ​​standard for 6G based on the existing 5QI standard's development principles. This standard includes 6QI-C (6G QoS Identifier-Transmission) and 6QI-T (6G QoS Identifier-Computing). 6QI-C indicates the QoS requirements for computation, and 6QI-T indicates the QoS requirements for computation; therefore, 6QI-T can directly inherit from the existing 5QI standard. The main functions of the QMU layer include:

[0200] It has the function of QoS mapping and decomposition, which can map the QoS requirements of the entire computing task (task-level QoS) to the QoS of each computing / transmission subtask, that is, decompose it into specific 6QI-C and 6QI-T (resource-level QoS);

[0201] After the computation task is executed / data packets are transmitted over the air interface, the computation QoS and transmission QoS are updated based on the specific computation execution and transmission conditions. The update methods may include: calculating and updating the QoS evaluation values ​​of executed subtasks (updating one or more), and calculating and updating the 6QI-C and 6QI-T values ​​of subsequent subtasks to be executed (updating one or more).

[0202] The requirements for protocol design include:

[0203] Add QoS as a header to the data packet. The added QoS can be the QoS assessment value of the executed subtask, or the 6QI-C and 6QI-T requirements of the subtask to be executed.

[0204] Network side:

[0205] QMU layer: Service Quality Mapping and Update Layer. Its main functions include:

[0206] It has the function of QoS mapping and decomposition, which can map the QoS requirements of the entire computing task (task-level QoS) to the QoS of each computing / transmission subtask, that is, decompose it into specific 6QI-C and 6QI-T (resource-level QoS);

[0207] After the computation task is executed / data packets are transmitted over the air interface, the computation QoS and transmission QoS are updated based on the specific computation execution and transmission conditions. The update methods may include: calculating and updating the QoS evaluation values ​​of executed subtasks (updating one or more), and calculating and updating the 6QI-C and 6QI-T values ​​of subsequent subtasks to be executed (updating one or more).

[0208] The requirements for protocol design include:

[0209] Add QoS as a header to the data packet. The added QoS can be the QoS assessment value of the executed subtask, or the 6QI-C and 6QI-T requirements of the subtask to be executed.

[0210] The proposed embodiment of this application adds a CCC layer to coordinate the execution of computational tasks and the transmission of computational results.

[0211] UE side:

[0212] CCC Layer: A newly added computational connectivity and collaboration layer. In AI computation tasks, different data may have varying degrees of importance, impacting task execution and completion to varying degrees. To address this, this solution proposes adding a CCC layer to evaluate and set the weights of different element values ​​in the intermediate computation result vector, performing quantization and compression to ensure inference accuracy and quality while reducing computational and communication resource overhead. The main uplink functions include:

[0213] The data packet queue for intermediate calculation results is managed by maintaining a pair of I / O queues. The I queue receives the inference calculation results from the UE side, and the O queue receives the inference calculation results from the network side.

[0214] It can read the contents of data packets, evaluate and set the weights of different element values ​​in the intermediate calculation result vector, and perform quantization and compression;

[0215] The quantized and compressed CTP is then delivered to the C_SDAP layer.

[0216] The main functions of downlink include:

[0217] The data packet queue for intermediate calculation results is managed by maintaining a pair of I / O queues. The I queue receives the inference calculation results from the UE side, and the O queue receives the inference calculation results from the network side.

[0218] It can read the contents of data packets, interpret the weights of different intermediate calculation results, and perform dequantization and decompression.

[0219] The processed data packet is then passed to the upper layer: if the data packet is an intermediate calculation result, it is passed to the QMU layer; if it is an inference result, it is passed to the CTM layer.

[0220] Network side:

[0221] CCC Layer: Computational Connectivity Collaboration Layer. Its main uplink functions include:

[0222] The data packet queue for intermediate calculation results is managed by maintaining a pair of I / O queues. The I queue receives the inference calculation results from the UE side, and the O queue receives the inference calculation results from the network side.

[0223] It can read the contents of data packets, interpret the weights of different intermediate calculation results, and perform dequantization and decompression;

[0224] The intermediate calculation results after dequantization and decompression are submitted to the QMU layer.

[0225] The main functions of downlink include:

[0226] The data packet queue for intermediate calculation results is managed by maintaining a pair of I / O queues. The I queue receives the inference calculation results from the UE side, and the O queue receives the inference calculation results from the network side.

[0227] It can read the contents of data packets, evaluate and set the weights of different element values ​​in the intermediate calculation result vector, and perform quantization and compression;

[0228] The quantized and compressed CTP is then delivered to the C_SDAP layer.

[0229] In this application's embodiment, the SDAP layer is improved to a C_SDAP layer, which maps computational data of different granularities to the CRB:

[0230] UE side:

[0231] C_SDAP: Computation Service Data Adaptation Protocol Layer, partially modified from the 5G SDAP layer. Its innovation lies in implementing the mapping from CTP to the newly added CRB. CRB is a dedicated bearer for AI computing task data. C_SDAP can map computing data of different granularities (different elements in CTP / intermediate computing result vectors) to CRB. Its main uplink functions include:

[0232] The data CTP for computation tasks is mapped to the CRB based on 6QI-T for QoS flow. The mapping method can be based on the CTP as the granularity, mapping high QoS requirements CTP to high QoS guaranteed CRB; or based on different elements in the intermediate computation result vector as the granularity, mapping high-weight elements to high QoS guaranteed CRB.

[0233] The processed data packet is then handed over to the C_PDCP layer.

[0234] The requirements for protocol design include:

[0235] Mark the QoS Flow ID (QFI) in the uplink data packet.

[0236] The main functions of downlink include:

[0237] The processed SDU (i.e., CTP) is then submitted to the CCC layer.

[0238] The requirements for protocol design include:

[0239] If the CRB is configured with a C_SDAP header, reflective QoS flow mapping to the CRB is performed (the UE monitors the mapping of QoS flow to the CRB in the downlink and applies it to the uplink), and RQI processing is performed (the RQI and QFI values ​​are submitted to the NAS layer). The PDU header is removed to obtain the SDU. If the CRB is not configured with a C_SDAP header, then the PDU is the SDU.

[0240] Network side:

[0241] C_SDAP: Computation Service Data Adaptation Protocol Layer, its main uplink functions include:

[0242] The processed SDU (i.e., CTP) is then submitted to the CCC layer.

[0243] The requirements for protocol design include:

[0244] If the CRB is configured with a C_SDAP header, the SDU is obtained by removing the header of the PDU; if the CRB is not configured with a C_SDAP header, then the PDU is the SDU.

[0245] The main functions of downlink include:

[0246] The data CTP for computation tasks is mapped to the CRB based on 6QI-T for QoS flow. The mapping method can be based on the CTP as the granularity, mapping high QoS requirements CTP to high QoS guaranteed CRB; or based on different elements in the intermediate computation result vector as the granularity, mapping high-weight elements to high QoS guaranteed CRB.

[0247] The processed data packet is then handed over to the C_PDCP layer.

[0248] The requirements for protocol design include:

[0249] Mark the QoS Flow ID (QFI) in the downlink data packet.

[0250] The impact of new bearers and channels on the protocol:

[0251] After processing the data packet, the CCC layer passes it to the C_SDAP layer. Subsequent processes and the channels and bearers involved are as follows. Figure 4 As shown. Figure 4 The middle section contains concepts specific to computing tasks. In terms of bearers, a new Computational Radio Bearer (CRB) has been added; in terms of channels, a new logical channel—Computation-Dedicated Service Channel (C-DTCH)—has been added as the logical channel for AI computing tasks; and in terms of entities, two new entities have been added: the C_SDAP layer entity C_SDAP Entity and the C_PDCP layer entity C_PDCP Entity. The reasons for each addition and its specific functions will be explained in detail below.

[0252] The Radio Bearer (RB) is the collective term for the different protocol entities and configurations allocated by the base station to the UE, including PDCP protocol entities, RLC protocol entities, MAC protocol entities, and a series of resources allocated to the PHY. The RB is the channel (including PHY, MAC, RLC, and PDCP) connecting the gNodeB and the UE via the radio interface; all data transmitted on the radio interface must pass through the RB. There are two types of radio bearers in 5G networks: signaling bearers called SRBs and data bearers called DRBs. DRBs represent the data radio bearers for packet processing in the radio interface (Uu), and the gNB maps DRBs to QoS in the wireless network.

[0253] However, because the embodiments of this application have improved and optimized the functions of the SDAP and PDCP protocol layers, the DRB used in traditional communication services is no longer suitable for the protocol stack proposed in this solution, that is, it cannot be used as a transmission bearer for computational data. Therefore, a new bearer CRB is proposed. Figure 5As shown, the CRB is a dedicated bearer for computational data packet processing in the 6G radio interface. It is a collective term for different layer protocol entities and configurations allocated by the base station for the UE's computational tasks, including the C_PDCP protocol entity, RLC protocol entity, MAC protocol entity, and a series of resources allocated by the PHY. Different CRBs are responsible for providing different data packet air interface transmission processing for the user's computational data packets. The C_SDAP layer can map it to different CRBs according to the CTP's QoS requirements. The CRB serves as the channel for transmitting computational task data packets between the C_SDAP layer and the C_PDCP layer, adapting to the functions of the protocol layers proposed in this scheme. Therefore, there are three types of radio bearers in the 6G network: SRB, DRB, and CRB, which serve as signaling bearers, dedicated bearers for communication data packet processing, and dedicated bearers for computational data packet processing, respectively.

[0254] In 5G, DRB is configured through the RadioBearerConfig of the RRC layer, and the parameter drb-Identity is the identifier of the DRB. Therefore, a new parameter crb-Identity can be added as the identifier of the CRB in the 6G protocol.

[0255] In the 5G protocol stack, each RB (except SRB0 for the Uu interface) is associated with one PDCP entity. Depending on the RB characteristics (e.g., unidirectional / bidirectional or decoupled / non-decoupled) or RLC mode, each PDCP entity is associated with one, two, three, four, six, or eight RLC entities. Similarly, each CRB is associated with one C_PDCP entity, and depending on the CRB characteristics or C_RLC mode, each C_PDCP entity may be associated with one or more C_RLC entities.

[0256] In 5G, the logical channel DTCH is a point-to-point channel dedicated to a single UE, used solely for transmitting user plane information. DTCH can exist in both the uplink and downlink. In the downlink, it is mapped to the transport channel DL-SCH, and in the uplink, it is mapped to the UL-SCH.

[0257] To support 6G computing services, a new logical channel, C-DTCH, is proposed to adapt to the improved protocol stack. C-DTCH is a logical channel dedicated to transmitting user plane information for computing services. It exists in both the uplink and downlink. Its mapping to the transport channel can be the same as DTCH—mapped to the transport channel DL-SCH in the downlink and UL-SCH in the uplink—or it can be mapped to transport channels specifically designed for computing services, such as DL-C-SCH and UL-C-SCH. The design of these two new transport channels is beyond the scope of this patent. In 6G networks, DTCH is only used to transmit user plane information for communication services.

[0258] In 5G, the logical channel instance of a MAC SDU is identified by the LCID and eLCID fields in the MAC layer subheader. C-DTCH can continue to use the eLCID field for identification, but considering the increased number of logical channels in 6G, the length of the eLCID field needs to be expanded. In the 5G protocol, the eLCID field is 1 or 2 8-bit bytes long, and the LCID field value of 35-46 is retained. Therefore, it can be specified that when the LCID field value is 35, the eLCID field length is 3 8-bit bytes. Taking a MAC layer subheader containing an 8-bit long L field as an example, the 5G protocol specifies three main formats, such as... Figure 6 As shown. Additionally, as... Figure 7 As shown, to accommodate the increase in the number of logical channels in 6G, additional channels can be added. Figure 7 The format shown is the MAC layer subheader.

[0259] Taking edge-edge collaborative AI inference as an example, the instantiation of the above user plane functions can be divided into the following two cases:

[0260] Scenario 1: After performing a portion of the inference computation task on the terminal device (UE) side, the data is submitted to the network side to execute the remaining inference computation task. In this case, the final task is completed on the network side, and the UE ultimately only receives the required inference results from the network side.

[0261] Scenario 2: After the UE performs a portion of the inference computation, it submits the data to the network side to perform a portion of the inference computation. Due to user privacy requirements and other reasons, the network side will not complete all the inference tasks, but will wirelessly transmit the intermediate computation results to the UE side. The final computation task needs to be executed by the UE.

[0262] For scenario 1, the overall process of a single base station collaborating with the UE to perform inference services is as follows: Figure 8 As shown, the specific process is as follows:

[0263] Step 1: CTM interacts with the application layer to obtain QoS requirements and starts the computing task.

[0264] Step 2: The QMU layer maps the QoS of the entire computation task to the QoS of each sub-computation / transmission task, breaking it down into specific 6QI-C and 6QI-T. Then, the QoS requirements are added to the data packet as a header, and the packet is passed to the CE layer.

[0265] Taking latency requirements as an example, the overall QoS requirements for the task and the 6QI-C and 6QI-T for each sub-computation / transmission task can be shown in Table 1 below.

[0266] Table 1. Overall Task QoS, Sub-task 6QI-C and 6QI-T Examples

[0267]

[0268]

[0269] Step 3. After receiving the computation task, the CE layer deploys and executes the task, using UE-side computing resources for computation. Intermediate computation results are obtained and transmitted back to the QMU layer.

[0270] Step 4. After receiving the intermediate calculation results, the QMU layer updates 6QI-C and 6QI-T according to the specific calculation execution situation, and passes the data packet to the CCC layer.

[0271] Step 5.

[0272] 1) The CCC layer manages the data packet queue of intermediate calculation results, reads the contents of the data packets, evaluates and sets the weights of different intermediate calculation results, quantizes and compresses the intermediate results according to the weights, and submits the processed CTP to the C_SDAP layer.

[0273] 2) The C_SDAP layer performs QoS flow mapping to CRB for the data of the computing task.

[0274] 3) The C_PDCP layer encrypts and protects the privacy of data packets for computing tasks.

[0275] 4) The services and functions of the RLC layer, MAC layer and PHY layer are the same as those of the original 5G protocol stack.

[0276] Step 6. Calculate the task data packet and wirelessly transmit it to the network device (e.g., base station).

[0277] 1) The services and functions of the RLC layer, MAC layer and PHY layer are the same as those of the original 5G protocol stack.

[0278] 2) The C_PDCP layer mainly performs data decryption and other tasks.

[0279] 3) The C_SDAP layer performs CRB to QoS flow reflection for computation task data packets.

[0280] 4) The CCC layer manages the data packet queue, performs dequantization and decompression, and submits the intermediate calculation results to the QMU layer.

[0281] Step 7. The QMU layer updates 6QI-C and 6QI-T based on the previous specific calculation execution and transmission information, and then passes the data packet to the CE layer.

[0282] Step 8. The CE layer deploys and executes the computation task, using base station resources for computation. The inference results are then passed back to the QMU layer.

[0283] Step 9. The QMU layer updates 6QI-T based on the previous specific calculation execution results.

[0284] Step 10.

[0285] 1) The CCC layer manages the packet queue and delivers the CTP to the C_SDAP layer;

[0286] 2) The C_SDAP layer maps the data of the computing task to the CRB via QoS flow;

[0287] 3) The C_PDCP layer encrypts and protects the privacy of the inference result data packets;

[0288] 4) The services and functions of the RLC layer, MAC layer and PHY layer are the same as those of the original 5G protocol stack.

[0289] Step 11. Data is transmitted to the UE side via the channel.

[0290] 1) The services and functions of the RLC layer, MAC layer, and PHY layer are the same as those of the original 5G protocol stack;

[0291] 2) The C_PDCP layer mainly performs data decryption and other tasks;

[0292] 3) The C_SDAP layer performs CRB to QoS flow reflection for computation task data packets;

[0293] 4) The CCC layer manages the data packet queue, performing dequantization and decompression.

[0294] Step 12. The CCC layer submits the inference results to the CTM layer, and the CTM layer ends the entire computational inference task.

[0295] For scenario 2, the key difference is that the final computation task needs to be performed on the UE side. In this case, the overall process of a single base station collaborating with the UE to perform inference services is as follows: Figure 9 As shown, the specific process is as follows:

[0296] Steps 1-8 are the same as in Case 1. Since the remaining part of the computation task needs to be performed on the UE side, the subsequent steps are modified, and steps 13-14 are added. The subsequent steps are described in detail below:

[0297] Step 9. The QMU layer updates the 6QI-T and 6QI-C requirements based on the previous specific calculation execution.

[0298] Step 10.

[0299] 1) The CCC layer manages the data packet queue of intermediate calculation results, reads the contents of the data packets, evaluates and sets the weights of different intermediate calculation results, quantizes and compresses the intermediate results according to the weights, and submits the processed CTP to the C_SDAP layer.

[0300] 2) The C_SDAP layer maps the data of the computing task to the CRB via QoS flow;

[0301] 3) The C_PDCP layer encrypts and protects the privacy of data packets for computation tasks;

[0302] 4) The services and functions of the RLC layer, MAC layer, and PHY layer are the same as those of the original 5G protocol stack;

[0303] Step 11. Data is transmitted to the UE side through the channel. The specific steps are the same as step 11 in case 1.

[0304] Step 12. The QMU layer updates 6QI-C based on the previous specific calculation execution and transmission information.

[0305] Step 13. The CE layer deploys and executes the computing tasks, using UE-side resources for computation.

[0306] Step 14. The CE layer submits the inference results to the CTM layer, and the CTM layer ends the entire computational inference task.

[0307] The CE layer needs to identify the task to which the computation result belongs, the data input, and whether it is an intermediate computation result or an inference result. The CE layer PDU format can be designed as follows: Figure 10 As shown, where:

[0308] I / F: Indicates whether the data is an intermediate calculation result or a final inference result;

[0309] R: Reserved bit;

[0310] TID: Indicates the computation task to which the calculation result belongs (this field can have various lengths);

[0311] DIID: Indicates the data input corresponding to the calculation result (this field can have various lengths).

[0312] The data packet format of the C_SDAP layer is the same as that of the SDAP layer in 5G, but the RDI field of the downlink Data PDU in the C_SDAP header has been updated, such as... Figure 11 As shown, where:

[0313] RDI: Indicates whether the mapping rules from QoS to DRB need to be updated (5G) — Updated to indicate whether the mapping rules from QoS to CRB need to be updated;

[0314] RQI: Indicates whether changes to the SDF-to-QoS flow mapping rules need to be notified to the NAS layer;

[0315] QFI: Declare QoS Stream ID.

[0316] The difference between the data packet format of the C_PDCP layer and the PDCP layer in 5G is that the MAC-I field generated by integrity protection has been removed, and a new HE field has been added to identify whether the data packet is homomorphically encrypted. The two user plane PDUs have a 12-bit SN length, for example... Figure 12 and Figure 13 As shown, where:

[0317] D / C: Declare whether it is a data PDU or a control PDU;

[0318] HE: Declares whether the PDU has been homomorphically encrypted;

[0319] R: Reserved bit;

[0320] C_PDCP SN: The SN number of the C_PDCP PDU.

[0321] In related technologies, the 5G user plane protocol stack is only designed for communication services and their data, and cannot be adapted to computing tasks in 6G networks. The solution proposed in this application is for AI computing services with end-to-end collaboration. By adding new protocol layers and improving existing protocol layers, it effectively supports the entire process of inference tasks, including control, management, QoS formulation and guarantee, execution, and collaboration, providing a reference for the design of 6G air interface user plane protocol stacks.

[0322] Furthermore, this solution considers the differences between AI tasks and communication tasks, and merges the execution of computation and the connection design. It also proposes a new bearer for computational connections, CRB, and a new logical channel, C-DTCH, as transmission channels between different protocol layers of computational tasks, enabling the overall protocol stack to achieve interoperability.

[0323] In addition, the CCC layer proposed in this proposal can quantize and compress different intermediate calculation results according to their importance, which effectively reduces the overhead of computing and communication resources while ensuring inference accuracy and quality, and reduces the load on the UE side and the network side to perform computing tasks.

[0324] like Figure 14 As shown, Figure 14 This is a schematic diagram of the structure of a task execution device provided in an embodiment of this application, as shown below. Figure 14 As shown, the task execution device 1400 includes:

[0325] The first acquisition module 1401 is used to acquire task execution requests;

[0326] The first task execution module 1402 is used to execute the target task corresponding to the task execution request based on the first air interface user plane protocol stack.

[0327] The first air interface user plane protocol stack is obtained by performing at least one of the following processes on the second air interface user plane protocol stack: adding a Computation Task Management (CTM) layer; adding a first Quality of Service Mapping and Update (QMU) layer; adding a first Computation Execution (CE) layer; adding a first Computation Connection Coordination (CCC) layer; updating the Service Data Adaptation Protocol (SDAP) layer in the second air interface user plane protocol stack; and updating the Packet Data Convergence Protocol (PDCP) layer in the second air interface user plane protocol stack.

[0328] Among them, the CTM layer is used for computation task management, the first QMU layer is used for at least one of the following: task service quality (QoS) requirement division, QoS requirement update and QoS evaluation, the first CE layer is used for task computation, the first CCC layer is used for compressing different task data packets with different weights, the updated SDAP layer is used for mapping the QoS flow of task data packets to the computation radio bearer (CRB), and the updated PDCP layer is used for encrypting and / or decrypting the data body in the task data packets.

[0329] In some embodiments, the first task execution module includes at least one of the following:

[0330] The initiation unit is used to obtain the QoS requirements of the task through the CTM layer and start the target task, which includes multiple subtasks.

[0331] The partitioning unit is used to partition the QoS requirements of the task through the first QMU layer, determine multiple QoS sub-requirements, and the multiple QoS sub-requirements correspond one-to-one with multiple sub-tasks;

[0332] The first processing unit is used to add the corresponding QoS sub-requirements to the subtask data packet when any subtask data packet is obtained through the first QMU layer, so as to update the subtask data packet and transmit the updated subtask data packet to the first CE layer.

[0333] The first execution unit is used to execute the corresponding subtask in the target task through the first CE layer on the updated subtask data packet, and obtain the first intermediate calculation result of the updated subtask data packet.

[0334] In some embodiments, the first task execution module further includes at least one of the following:

[0335] The first transmission unit is used to transmit the first intermediate calculation result to the first QMU layer through the first CE layer;

[0336] The second transmission unit is used to transmit a first intermediate calculation result to the network device through the first CCC layer, the updated SDAP, and the updated PDCP. The first intermediate calculation result is used by the network device to obtain the first calculation result by executing the corresponding sub-task in the target task.

[0337] In some embodiments, the plurality of subtasks includes at least one computation subtask and at least one transmission subtask, and the plurality of QoS sub-requirements includes the QoS sub-requirement of each task in at least one computation subtask and the QoS sub-requirement of each task in at least one transmission subtask.

[0338] The device also includes:

[0339] The first update module is used to update the QoS sub-requirements of at least one computation subtask and the QoS sub-requirements of at least one transmission subtask through the first QMU layer.

[0340] In some embodiments, the second transmission unit includes:

[0341] The first data packet determination unit is used to determine the weight of the first intermediate calculation result through the first CCC layer, and compress the first intermediate calculation result based on the weight of the first intermediate calculation result to obtain the first data packet;

[0342] The second data packet determination unit is used to perform QoS flow mapping to CRB for the first data packet using the updated SDAP and the corresponding QoS sub-requirements of the first data packet, and generate a second data packet. The data body of the second data packet includes the first data packet.

[0343] The third data packet determination unit is used to encrypt the data body of the second data packet using the updated PDCP to obtain the third data packet, and send the third data packet to the network device. The third data packet is used by the network device to perform corresponding sub-tasks based on the third data packet to obtain the first calculation result.

[0344] In some embodiments, the first task execution module further includes:

[0345] The first receiving unit is used to receive a first calculation result sent by the network device. The first calculation result is either the task result of the target task or a second intermediate calculation result. The second intermediate calculation result is used by the terminal device to execute the corresponding sub-task in the target task to obtain the task result.

[0346] In some embodiments, the apparatus further includes at least one of the following:

[0347] The task result acquisition module is used to execute the corresponding subtask in the target task through the first CE layer on the second intermediate calculation result data packet when the first calculation result is the second intermediate calculation result, so as to obtain the task result.

[0348] The transmission module is used to transmit the task result from the first CE layer to the CTM layer when the first calculation result data is the task result, and to terminate the execution of the target task through the CTM layer.

[0349] In some embodiments, when the first calculation result is a second intermediate calculation result, the apparatus further includes:

[0350] The second update module is used to update the QoS sub-requirements of at least one computational subtask through the first QMU layer.

[0351] In some embodiments, the third data packet includes encryption algorithm indication information, which is used to indicate the target encryption algorithm used by the data body of the second data packet.

[0352] The task execution device 1400 provided in this embodiment can realize the various processes of the above-described task execution method applied to terminal devices. The technical features are one-to-one and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0353] See Figure 15 , Figure 15 This is a schematic diagram of the structure of a task execution device provided in an embodiment of this application, as shown below. Figure 15 As shown, the task execution device 1500 includes:

[0354] The second task execution module 1501 is used to execute the target task based on the third air interface user plane protocol stack;

[0355] The third air interface user plane protocol stack is obtained by performing at least one of the following processes on the second air interface user plane protocol stack: adding a second quality of service mapping and updating QMU layer; adding a second computation execution CE layer; adding a second computation connection coordination CCC layer; updating the service data adaptation protocol SDAP layer in the second air interface user plane protocol stack; and updating the packet data aggregation protocol PDCP layer in the second air interface user plane protocol stack.

[0356] The second QMU layer is used for at least one of the following: QoS requirement division, QoS requirement update, and QoS assessment; the second CE layer is used to perform task calculation; the second CCC layer is used to compress different task data packets with different weights; the updated SDAP layer is used to calculate the reverse mapping from radio bearer CRB to QoS flow and / or the mapping from QoS flow to CRB for task data packets; and the updated PDCP layer is used to encrypt and / or decrypt the data body in the task data packets.

[0357] In some embodiments, the target task includes multiple subtasks;

[0358] The second task execution module includes:

[0359] The second receiving unit is configured to receive a third data packet sent by the terminal device; and at least one of the following:

[0360] The second execution unit is used to execute the corresponding sub-task in the target task of the third data packet through the second CE layer to obtain the first calculation result;

[0361] The sending unit is used to send the first calculation result to the terminal device through the second CCC layer, the updated SDAP, and the updated PDCP.

[0362] In some embodiments, multiple subtasks correspond one-to-one with multiple QoS sub-requirements. The multiple subtasks include at least one computation subtask and at least one transmission subtask. The multiple QoS sub-requirements include the QoS sub-requirements of each task in at least one computation subtask and the QoS sub-requirements of each task in at least one transmission subtask.

[0363] The device also includes:

[0364] The third update module is used to update the QoS sub-requirements of at least one computation subtask and the QoS sub-requirements of at least one transmission subtask through the second QMU layer.

[0365] In some embodiments, the transmitting unit includes:

[0366] The fourth data packet determination unit is used to determine the weight of the first calculation result through the second CCC layer, and compress the first calculation result based on the weight of the first calculation result to obtain the fourth data packet;

[0367] The fifth data packet determination unit is used to perform a reverse mapping from CRB to QoS flow for the fourth data packet using the updated SDAP and the corresponding QoS sub-requirement of the fourth data packet, and generate the fifth data packet. The data body of the fifth data packet includes the fourth data packet.

[0368] The sixth data packet determination unit is used to encrypt the data body of the fifth data packet using the updated PDCP to obtain the sixth data packet, and send the sixth data packet to the terminal device. The sixth data packet is used by the terminal device to obtain the first calculation result.

[0369] In some embodiments, the first calculation result is the task result of the target task or the second intermediate calculation result. The second intermediate calculation result is used by the terminal device to execute the corresponding sub-task in the target task to obtain the task result.

[0370] In some embodiments, the third data packet includes encryption algorithm indication information, which is used to indicate the target encryption algorithm.

[0371] In some embodiments, the second execution unit includes at least one of the following:

[0372] The first anti-mapping unit is used to perform CRB-to-QoS flow anti-mapping on the third data packet through the updated SDAP when the target encryption algorithm indicated by the encryption algorithm indication information is a homomorphic encryption algorithm, to obtain the seventh data packet; the first execution subunit is used to execute the corresponding sub-task in the target task on the seventh data packet through the second CE layer to obtain the first calculation result.

[0373] The decryption unit is used to decrypt the third data packet to obtain the eighth data packet when the target encryption algorithm indicated by the encryption algorithm indication information is a non-homomorphic encryption algorithm; the second anti-mapping unit is used to perform CRB to QoS flow anti-mapping on the eighth data packet through the updated SDAP to obtain the ninth data packet; the second execution subunit is used to execute the corresponding sub-task in the target task on the ninth data packet through the second CE layer to obtain the first calculation result.

[0374] The task execution device 1500 provided in this embodiment can implement the various processes of the above-described task execution method applied to network devices. The technical features are one-to-one and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0375] This application also provides an electronic device, including: a processor, a memory, and a program stored in the memory and executable on the processor. When the program is executed by the processor, it implements the various processes of the above-described task execution method embodiments applied to terminal devices and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0376] For details, see Figure 16 This application also provides an electronic device, including a bus 1601, a transceiver 1602, an antenna 1603, a bus interface 1604, a processor 1605, and a memory 1606.

[0377] The processor is used for:

[0378] Obtain the task execution request;

[0379] Based on the first air interface user plane protocol stack, execute the target task corresponding to the task execution request;

[0380] The first air interface user plane protocol stack is obtained by performing at least one of the following processes on the second air interface user plane protocol stack: adding a Computation Task Management (CTM) layer; adding a first Quality of Service Mapping and Update (QMU) layer; adding a first Computation Execution (CE) layer; adding a first Computation Connection Coordination (CCC) layer; updating the Service Data Adaptation Protocol (SDAP) layer in the second air interface user plane protocol stack; and updating the Packet Data Convergence Protocol (PDCP) layer in the second air interface user plane protocol stack.

[0381] Among them, the CTM layer is used for computation task management, the first QMU layer is used for at least one of the following: task service quality (QoS) requirement division, QoS requirement update and QoS evaluation, the first CE layer is used for task computation, the first CCC layer is used for compressing different task data packets with different weights, the updated SDAP layer is used for mapping the QoS flow of task data packets to the computation radio bearer (CRB), and the updated PDCP layer is used for encrypting and / or decrypting the data body in the task data packets.

[0382] exist Figure 16 In this document, a bus architecture (represented by bus 1601) is used. Bus 1601 can include any number of interconnected buses and bridges, linking various circuits including one or more processors represented by processor 1605 and memory represented by memory 1606. Bus 1601 can also link various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. Bus interface 1604 provides an interface between bus 1601 and transceiver 1602. Transceiver 1602 can be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by processor 1605 is transmitted over a wireless medium via antenna 1603, which further receives data and transmits it to processor 1605.

[0383] Processor 1605 manages bus 1601 and general processing, and also provides various functions, including timing, peripheral interface, voltage regulation, power management, and other control functions. Memory 1606 can be used to store data used by processor 1605 during operation.

[0384] Optionally, the processor 1605 can be a CPU, ASIC, FPGA, or CPLD.

[0385] The processing of the electronic device provided in this embodiment can realize the various processes of the above-described task execution method applied to network devices. The technical features are one-to-one and can achieve the same technical effect. To avoid repetition, they will not be described again here.

[0386] This application also provides a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the various processes of the task execution method embodiments described above for terminal devices, achieving the same technical effects. To avoid repetition, these will not be described again here. The computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0387] This application also provides an electronic device, including: a processor, a memory, and a program stored in the memory and executable on the processor. When the program is executed by the processor, it implements the various processes of the above-described task execution method embodiments applied to network devices and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0388] For details, see Figure 17 As shown in the figure, this application embodiment also provides an electronic device, including a bus 1701, a transceiver 1702, an antenna 1703, a bus interface 1704, a processor 1705, and a memory 1706.

[0389] The processor is used for:

[0390] Execute the target task based on the third air interface user plane protocol stack;

[0391] The third air interface user plane protocol stack is obtained by performing at least one of the following processes on the second air interface user plane protocol stack: adding a second quality of service mapping and updating QMU layer; adding a second computation execution CE layer; adding a second computation connection coordination CCC layer; updating the service data adaptation protocol SDAP layer in the second air interface user plane protocol stack; and updating the packet data aggregation protocol PDCP layer in the second air interface user plane protocol stack.

[0392] The second QMU layer is used for at least one of the following: QoS requirement division, QoS requirement update, and QoS assessment; the second CE layer is used to perform task calculation; the second CCC layer is used to compress different task data packets with different weights; the updated SDAP layer is used to calculate the reverse mapping from radio bearer CRB to QoS flow and / or the mapping from QoS flow to CRB for task data packets; and the updated PDCP layer is used to encrypt and / or decrypt the data body in the task data packets.

[0393] exist Figure 17In this document, a bus architecture (represented by bus 1701) is used. Bus 1701 can include any number of interconnected buses and bridges, linking various circuits including one or more processors represented by processor 1705 and memory represented by memory 1706. Bus 1701 can also link various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. Bus interface 1704 provides an interface between bus 1701 and transceiver 1702. Transceiver 1702 can be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by processor 1705 is transmitted over a wireless medium via antenna 1703, which further receives data and transmits it to processor 1705.

[0394] Processor 1705 manages bus 1701 and general processing, and also provides various functions, including timing, peripheral interface, voltage regulation, power management, and other control functions. Memory 1706 can be used to store data used by processor 1705 during operation.

[0395] Optionally, the processor 1705 can be a CPU, ASIC, FPGA, or CPLD.

[0396] The processing of the electronic device provided in this embodiment can realize the various processes of the above-described task execution method applied to network devices. The technical features are one-to-one and can achieve the same technical effect. To avoid repetition, they will not be described again here.

[0397] This application also provides a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the various processes of the above-described task execution method embodiments applied to network devices, achieving the same technical effects. To avoid repetition, further details are omitted here. The computer-readable storage medium may be, for example, ROM, RAM, magnetic disk, or optical disk.

[0398] This application provides a computer program product, including computer instructions. When the computer instructions are executed by a processor, they implement the various processes of the method described in the embodiment. The technical features are one-to-one and can achieve the same technical effect. To avoid repetition, they will not be described again here.

[0399] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0400] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, air conditioner, or first network device, etc.) to execute the methods of the various embodiments of this application.

[0401] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

Claims

1. A task execution method, characterized in that, Applied to a terminal device, the method includes: Obtain the task execution request; Based on the first air interface user plane protocol stack, execute the target task corresponding to the task execution request; The first air interface user plane protocol stack is obtained by performing at least one of the following processes on the second air interface user plane protocol stack: adding a Computation Task Management (CTM) layer; adding a first Quality of Service Mapping and Update (QMU) layer; adding a first Computation Execution (CE) layer; adding a first Computation Connection Coordination (CCC) layer; updating the Service Data Adaptation Protocol (SDAP) layer in the second air interface user plane protocol stack; and updating the Packet Data Convergence Protocol (PDCP) layer in the second air interface user plane protocol stack. The CTM layer is used for computation task management; the first QMU layer is used for at least one of the following: task service quality (QoS) requirement division, QoS requirement update, and QoS evaluation; the first CE layer is used for task computation; the first CCC layer is used for compressing different task data packets with different weights; the updated SDAP layer is used for mapping the QoS flow of task data packets to the computation radio bearer (CRB); and the updated PDCP layer is used for encrypting and / or decrypting the data body in the task data packets.

2. The method according to claim 1, characterized in that, The execution of the target task corresponding to the task execution request based on the first air interface user plane protocol stack includes at least one of the following: The QoS requirements for the task are obtained through the CTM layer, and the target task is started. The target task includes multiple subtasks. The task QoS requirements are divided by the first QMU layer to determine multiple QoS sub-requirements, and the multiple QoS sub-requirements correspond one-to-one with the multiple sub-tasks; If any subtask data packet is obtained through the first QMU layer, the corresponding QoS sub-requirement is added to the subtask data packet to update the subtask data packet, and the updated subtask data packet is transmitted to the first CE layer. The first CE layer executes the corresponding subtask in the target task on the updated subtask data packet to obtain the first intermediate calculation result of the updated subtask data packet.

3. The method according to claim 2, characterized in that, After obtaining the first intermediate calculation result of the updated subtask data packet, the method further includes at least one of the following: The first intermediate calculation result is transmitted from the first CE layer to the first QMU layer. The first intermediate calculation result is transmitted to the network device through the first CCC layer, the updated SDAP, and the updated PDCP. The first intermediate calculation result is used by the network device to execute the corresponding sub-task in the target task to obtain the first calculation result.

4. The method according to claim 3, characterized in that, The plurality of subtasks includes at least one computation subtask and at least one transmission subtask, and the plurality of QoS sub-requirements includes the QoS sub-requirements of each task in the at least one computation subtask and the QoS sub-requirements of each task in the at least one transmission subtask. After transmitting the first intermediate calculation result to the first QMU layer through the first CE layer, the method further includes: The QoS sub-requirements of the at least one computation subtask and the at least one transmission subtask are updated through the first QMU layer.

5. The method according to claim 3, characterized in that, The step of transmitting the first intermediate calculation result to the network device through the first CCC layer, the updated SDAP, and the updated PDCP includes: The weight of the first intermediate calculation result is determined by the first CCC layer, and the first intermediate calculation result is compressed based on the weight of the first intermediate calculation result to obtain the first data packet; The updated SDAP uses the QoS sub-requirements corresponding to the first data packet to perform QoS flow mapping to CRB for the first data packet, generating a second data packet whose data body includes the first data packet; The data body of the second data packet is encrypted using the updated PDCP to obtain a third data packet, which is then sent to the network device. The third data packet is used by the network device to perform corresponding sub-tasks based on the third data packet to obtain the first calculation result.

6. The method according to claim 3, characterized in that, After transmitting the first intermediate calculation result to the network device, the method further includes: The terminal device receives the first calculation result sent by the network device. The first calculation result is either the task result of the target task or a second intermediate calculation result. The second intermediate calculation result is used by the terminal device to execute the corresponding sub-task in the target task to obtain the task result.

7. The method according to claim 6, characterized in that, The method further includes at least one of the following: If the first calculation result is the second intermediate calculation result, the corresponding subtask in the target task is executed on the second intermediate calculation result data packet through the first CE layer to obtain the task result; If the first calculation result data is the task result, the task result is transmitted from the first CE layer to the CTM layer, and the execution of the target task is terminated through the CTM layer.

8. The method according to claim 7, characterized in that, If the first calculation result is the second intermediate calculation result, the method further includes: The QoS sub-requirements of the at least one computational subtask are updated through the first QMU layer.

9. The method according to claim 5, characterized in that, The third data packet includes encryption algorithm indication information, which is used to indicate the target encryption algorithm used by the data body of the second data packet.

10. A task execution method, characterized in that, Applied to network devices, the method includes: Execute the target task based on the third air interface user plane protocol stack; The third air interface user plane protocol stack is obtained by performing at least one of the following processes on the second air interface user plane protocol stack: adding a second quality of service mapping and update (QMU) layer; adding a second computation execution (CE) layer; adding a second computation connection coordination (CCC) layer; updating the service data adaptation protocol (SDAP) layer in the second air interface user plane protocol stack; and updating the packet data aggregation protocol (PDCP) layer in the second air interface user plane protocol stack. The second QMU layer is used for at least one of the following: QoS requirement division, QoS requirement update, and QoS assessment; the second CE layer is used to perform task calculation; the second CCC layer is used to compress different task data packets with different weights; the updated SDAP layer is used to calculate the reverse mapping from radio bearer CRB to QoS flow and / or the mapping from QoS flow to CRB for task data packets; and the updated PDCP layer is used to encrypt and / or decrypt the data body in the task data packets.

11. The method according to claim 10, characterized in that, The target task includes multiple sub-tasks; Based on the third air interface user plane protocol stack, execute the target tasks, including: Receive a third data packet sent by a terminal device; and at least one of the following: The first calculation result is obtained by performing the corresponding sub-task in the target task on the third data packet through the second CE layer; The first calculation result is sent to the terminal device through the second CCC layer, the updated SDAP, and the updated PDCP.

12. The method according to claim 11, characterized in that, The multiple subtasks correspond one-to-one with multiple QoS sub-requirements. The multiple subtasks include at least one computation subtask and at least one transmission subtask. The multiple QoS sub-requirements include the QoS sub-requirements of each task in the at least one computation subtask and the QoS sub-requirements of each task in the at least one transmission subtask. After receiving the third data packet sent by the receiving terminal device, the method further includes: The QoS sub-requirements of the at least one computation subtask and the at least one transmission subtask are updated through the second QMU layer.

13. The method according to claim 11, characterized in that, Sending the first calculation result to the terminal device through the second CCC layer, the updated SDAP, and the updated PDCP includes: The weight of the first calculation result is determined by the second CCC layer, and the first calculation result is compressed based on the weight of the first calculation result to obtain the fourth data packet; The updated SDAP utilizes the QoS sub-requirements corresponding to the fourth data packet to perform a reverse mapping from CRB to QoS flow for the fourth data packet, generating a fifth data packet whose data body includes the fourth data packet; The data body of the fifth data packet is encrypted using the updated PDCP to obtain the sixth data packet, which is then sent to the terminal device. The sixth data packet is used by the terminal device to obtain the first calculation result.

14. The method according to claim 13, characterized in that, The first calculation result is the task result of the target task or the second intermediate calculation result. The second intermediate calculation result is used by the terminal device to execute the corresponding sub-task in the target task to obtain the task result.

15. The method according to claim 11, characterized in that, The third data packet includes encryption algorithm indication information, which is used to indicate the target encryption algorithm.

16. The method according to claim 15, characterized in that, The step of performing the corresponding subtask in the target task on the third data packet through the second CE layer to obtain the first calculation result includes at least one of the following: When the target encryption algorithm indicated by the encryption algorithm indication information is a homomorphic encryption algorithm, the third data packet is reverse-mapped from CRB to QoS stream through the updated SDAP to obtain the seventh data packet. The corresponding subtask in the target task is executed on the seventh data packet through the second CE layer to obtain the first calculation result. When the target encryption algorithm indicated by the encryption algorithm indication information is a non-homomorphic encryption algorithm, the third data packet is decrypted using the updated PDCP to obtain the eighth data packet; the eighth data packet is reverse-mapped from CRB to QoS stream using the updated SDAP to obtain the ninth data packet; and the corresponding subtask in the target task is executed on the ninth data packet using the second CE layer to obtain the first calculation result.

17. A task execution device, characterized in that, include: The first acquisition module is used to acquire task execution requests; The first task execution module is used to execute the target task corresponding to the task execution request based on the first air interface user plane protocol stack; The first air interface user plane protocol stack is obtained by performing at least one of the following processes on the second air interface user plane protocol stack: adding a Computation Task Management (CTM) layer; adding a first Quality of Service Mapping and Update (QMU) layer; adding a first Computation Execution (CE) layer; adding a first Computation Connection Coordination (CCC) layer; updating the Service Data Adaptation Protocol (SDAP) layer in the second air interface user plane protocol stack; and updating the Packet Data Convergence Protocol (PDCP) layer in the second air interface user plane protocol stack. The CTM layer is used for computation task management; the first QMU layer is used for at least one of the following: task service quality (QoS) requirement division, QoS requirement update, and QoS evaluation; the first CE layer is used for task computation; the first CCC layer is used for compressing different task data packets with different weights; the updated SDAP layer is used for mapping the QoS flow of task data packets to the computation radio bearer (CRB); and the updated PDCP layer is used for encrypting and / or decrypting the data body in the task data packets.

18. A task execution device, characterized in that, include: The second task execution module is used to execute the target task based on the third air interface user plane protocol stack; The third air interface user plane protocol stack is obtained by performing at least one of the following processes on the second air interface user plane protocol stack: adding a second quality of service mapping and update (QMU) layer; adding a second computation execution (CE) layer; adding a second computation connection coordination (CCC) layer; updating the service data adaptation protocol (SDAP) layer in the second air interface user plane protocol stack; and updating the packet data aggregation protocol (PDCP) layer in the second air interface user plane protocol stack. The second QMU layer is used for at least one of the following: QoS requirement division, QoS requirement update, and QoS assessment; the second CE layer is used to perform task calculation; the second CCC layer is used to compress different task data packets with different weights; the updated SDAP layer is used to calculate the reverse mapping from radio bearer CRB to QoS flow and / or the mapping from QoS flow to CRB for task data packets; and the updated PDCP layer is used to encrypt and / or decrypt the data body in the task data packets.

19. An electronic device, characterized in that, Including transceivers and processors, The processor is used for: Obtain the task execution request; Based on the first air interface user plane protocol stack, execute the target task corresponding to the task execution request; The first air interface user plane protocol stack is obtained by performing at least one of the following processes on the second air interface user plane protocol stack: adding a Computation Task Management (CTM) layer; adding a first Quality of Service Mapping and Update (QMU) layer; adding a first Computation Execution (CE) layer; adding a first Computation Connection Coordination (CCC) layer; updating the Service Data Adaptation Protocol (SDAP) layer in the second air interface user plane protocol stack; and updating the Packet Data Convergence Protocol (PDCP) layer in the second air interface user plane protocol stack. The CTM layer is used for computation task management; the first QMU layer is used for at least one of the following: task service quality (QoS) requirement division, QoS requirement update, and QoS evaluation; the first CE layer is used for task computation; the first CCC layer is used for compressing different task data packets with different weights; the updated SDAP layer is used for mapping the QoS flow of task data packets to the computation radio bearer (CRB); and the updated PDCP layer is used for encrypting and / or decrypting the data body in the task data packets.

20. An electronic device, characterized in that, Including transceivers and processors, The processor is used for: Execute the target task based on the third air interface user plane protocol stack; The third air interface user plane protocol stack is obtained by performing at least one of the following processes on the second air interface user plane protocol stack: adding a second quality of service mapping and update (QMU) layer; adding a second computation execution (CE) layer; adding a second computation connection coordination (CCC) layer; updating the service data adaptation protocol (SDAP) layer in the second air interface user plane protocol stack; and updating the packet data aggregation protocol (PDCP) layer in the second air interface user plane protocol stack. The second QMU layer is used for at least one of the following: QoS requirement division, QoS requirement update, and QoS assessment; the second CE layer is used to perform task calculation; the second CCC layer is used to compress different task data packets with different weights; the updated SDAP layer is used to calculate the reverse mapping from radio bearer CRB to QoS flow and / or the mapping from QoS flow to CRB for task data packets; and the updated PDCP layer is used to encrypt and / or decrypt the data body in the task data packets.

21. An electronic device, characterized in that, include: A processor, a memory, and a program stored in the memory and executable on the processor, wherein the program, when executed by the processor, implements the steps of the method as claimed in any one of claims 1 to 9, or implements the steps of the method as claimed in any one of claims 10 to 16.

22. A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the method according to any one of claims 1-9, or implements the steps of the method according to any one of claims 10-16.

23. A computer program product, characterized in that, Includes computer instructions that, when executed by a processor, implement the steps of the method as described in any one of claims 1-16.

Images