Computing power service request processing system, method and device, computer equipment and medium

Abstract

The invention relates to a computing power service request processing system, method and device, computer equipment and a medium. The method comprises the following steps: a terminal obtains a computing power indicator carrying a user intention of a target user and uploads the computing power indicator to a cloud end under the condition that the target user initiates a computing power service request, the cloud end forwards the computing power indicator to an edge side, and sends the computing power indicator to the edge side according to the computing power indicator, geographic position information of the terminal and an SLA guarantee requirement of the target user; the terminal and the edge side are indicated to process the computing power service request interactively, and the edge side obtains a KQI index matched with the computing power service request according to the current network condition of the network to which the edge side belongs and the computing power indicator; under the condition that the KQI does not meet the SLA guarantee requirement, a request indicator of a target user is generated and sent to the cloud, and finally the cloud processes the computing power service request in cooperation with the terminal and provides computing power support for the terminal according to the user intention, the computing power indicator and the request indicator. By adopting the method, the processing efficiency of the computing power service request is improved.

Description

Technical Field

[0001] This application relates to the fields of wireless communication technology and AI terminal technology, and in particular to a computing power service request processing system, method, apparatus, computer equipment, computer-readable storage medium and computer program product. Background Technology

[0002] AI terminals support a hybrid AI architecture with edge and cloud and local deployment of small models on the edge, enabling multimodal content generation and full-scene context awareness with security and reliability. However, the model's computing power is limited, making it difficult to support hundreds of billions of operations and balancing energy efficiency. Large models running on cloud server clusters have unlimited computing power and support training on clusters of hundreds of cards, but they are subject to network transmission, computing power fluctuations, and interference with link quality, which will lead to high latency, making it difficult to achieve SLA agreements.

[0003] The pursuit of low power consumption and high real-time performance on the edge is inherently conflicted with the large scale and latency of cloud models, resulting in low processing efficiency in the target computing power service request processing method. Summary of the Invention

[0004] Therefore, it is necessary to provide a computing power service request processing system, method, apparatus, computer equipment, computer-readable storage medium, and computer program product that can improve processing efficiency in response to the above-mentioned technical problems.

[0005] In the first aspect, this application provides a computing power service request processing system, which includes a terminal, an edge device, and a cloud.

[0006] The terminal is used to obtain the terminal computing power indicator and upload it to the cloud when the target user initiates a computing power service request; the terminal computing power indicator carries the user intent of the target user;

[0007] In the cloud, it is used to forward terminal computing power indicators to the edge side, and instruct the terminal to interact with the edge side to process computing power service requests based on the terminal computing power indicators, the terminal's geographical location information, and the target user's SLA guarantee requirements.

[0008] On the edge side, it is used to obtain KQI indicators that match the computing power service request based on the current network status of the network to which the edge side belongs and the terminal computing power indicator; if the KQI indicators do not meet the SLA guarantee requirements, it generates the computing power service request indicator for the target user and sends it to the cloud.

[0009] In the cloud, it is used to process computing power service requests in collaboration with the terminal based on user intent, terminal computing power indicator, and computing power service request indicator, and to provide computing power support for the cloud-based large model to the terminal.

[0010] In conjunction with the first aspect, in one embodiment, the cloud is further configured to determine the anchor UPF corresponding to the terminal based on the terminal's geographical location information and SLA guarantee requirements, under the instruction of the terminal computing power symbol, so as to instruct the terminal to interact with the edge side to process computing power service requests based on the anchor UPF.

[0011] In conjunction with the first aspect, in one embodiment, the cloud is further configured to obtain the KQI guarantee required by the target user based on the computing power service request indicator. If the KQI guarantee representation requires anchor UPF replacement, a new anchor UPF is determined from the candidate UPFs based on the real-time cell location information of the candidate UPFs. Based on the new anchor UPF and the user's intent, the cloud collaborates with the terminal to process the computing power service request and provides the terminal with computing power support from the cloud's large model. If the KQI guarantee representation does not require anchor UPF replacement, the anchor UPF is used as the new anchor UPF, and SLA guarantee policy recommendations are subscribed to from the NWDAF based on the terminal's computing power indicator and computing power service request indicator. The cloud is also configured to send user session update recommendations to the PCF to establish an association between the PCF and the new anchor UPF.

[0012] In conjunction with the first aspect, in an exemplary embodiment, the cloud is further configured to report experience data of the subscription terminal to the new anchor UPF via NWDAF. The subscription information carries a user identifier, so that the new anchor UPF uploads the experience data of the target user corresponding to the user identifier to NWDAF. NWDAF predicts the network support service status based on the experience data, and generates an initial SLA guarantee strategy suggestion based on the predicted network support service status and the terminal's needs, and returns it to the cloud.

[0013] In the cloud, it is further used to receive initial SLA guarantee strategy recommendations, and to plan solutions based on the initial SLA guarantee strategy recommendations and user intentions to obtain corresponding execution plans; according to the execution plans, it processes computing power service requests.

[0014] In conjunction with the first aspect, in one embodiment, the cloud is further configured to perform uplink / downlink CSI channel switching at the physical layer, perform bandwidth adaptation at the network layer, and establish QoS dedicated load policies at the application layer; perform service requirement decomposition, estimation, and coordination that matches the service characteristics of computing power service requests; generate SLA guarantee policy recommendations; send SLA guarantee policy recommendations and policy conflict priorities to the PCF; the PCF is configured to generate corresponding SLA guarantee rules based on the SLA guarantee policy recommendations, policy conflict priorities, and terminal location information, and send them to the SMF, so that the SMF can establish dedicated loads according to the SLA guarantee rules; and if the dedicated load recommendation is successful, send a dedicated load establishment success indication to the NWDAF through the cloud.

[0015] In conjunction with the first aspect, in one embodiment, the cloud is further configured to receive an updated initial SLA guarantee strategy recommendation sent by the NWDAF when the SLA guarantee strategy recommendation cannot meet the SLA guarantee requirements; plan a solution based on the updated initial SLA guarantee strategy recommendation and user intent to obtain a corresponding new execution plan; process computing power service requests according to the new execution plan; the updated initial SLA guarantee strategy recommendation is obtained by the NWDAF receiving new experience data of the target user reported by the new anchor UPF; predicting the network support service status based on the new experience data to obtain a new predicted network support service status; and updating the initial SLA guarantee strategy recommendation based on the new predicted network support service status and the terminal's needs.

[0016] In conjunction with the first aspect, in an exemplary embodiment, the edge side is further configured to interact with the terminal to process computing power service requests, provided that the KQI metrics meet the SLA guarantee requirements; the KQI metrics include at least one of the following: number of stutters, stutter duration, bit rate, resolution, response latency, setup latency, jitter, and bit error rate.

[0017] In conjunction with the first aspect, in one embodiment, the terminal is further configured to acquire the terminal's communication data, user behavior data, personalized preference data, terminal computing power information, load information, and link quality information, as well as the session scenario information in which the terminal is located; generate a terminal computing power indicator based on the session scenario information, terminal computing power information, load information, link quality information, and computing power service requests; and perform local processing on the communication data, user behavior data, and personalized preference data to obtain user intent; the local processing includes personal privacy data protection encryption, semantic query, and intent recognition.

[0018] In conjunction with the first aspect, in one embodiment, the terminal is further configured to filter malicious and sensitive content in communication data, user behavior data, and personalized preference data to obtain cleaned communication data, user behavior data, and personalized preference data; and to perform local data processing on the cleaned communication data, user behavior data, and personalized preference data to obtain user intent.

[0019] Secondly, this application also provides a method for processing computing power service requests, applied in the cloud, including:

[0020] The terminal computing power indicator uploaded by the receiving terminal is obtained by the terminal after the target user initiates a computing power service request; the terminal computing power indicator carries the user intent of the target user.

[0021] The terminal computing power indicator is forwarded to the edge side, and based on the terminal computing power indicator, the terminal's geographical location information, and the target user's SLA guarantee requirements, the terminal is instructed to interact with the edge side to process computing power service requests.

[0022] The edge side receives the target user's computing power service request indicator sent by the edge side; the computing power service request indicator is generated by the edge side based on the current network status of the network to which the edge side belongs and the terminal computing power indicator, obtaining the KQI indicator that matches the computing power service request, and generating it if the KQI indicator does not meet the SLA guarantee requirements.

[0023] Based on user intent, terminal computing power indicator, and computing power service request indicator, it works with the terminal to process computing power service requests and provides computing power support for cloud-based large models to the terminal.

[0024] In conjunction with the second aspect, in one embodiment, based on the terminal computing power indicator, the terminal's geographical location information, and the target user's SLA guarantee requirements, the terminal is instructed to interact with the edge side to process computing power service requests, including:

[0025] Under the guidance of the terminal computing power symbol, the corresponding anchor UPF is determined based on the terminal's geographical location information and SLA guarantee requirements, so as to instruct the terminal to interact with the edge side to process computing power service requests based on the anchor UPF.

[0026] In conjunction with the second aspect, in one embodiment, based on the user intent, the terminal computing power indicator, and the computing power service request indicator, the system collaboratively processes computing power service requests with the terminal and provides computing power support for the cloud-based large model to the terminal, including:

[0027] Obtain the KQI guarantee required by the target user based on the computing power service request indicator;

[0028] When the KQI guarantee representation requires the anchor UPF to be replaced, a new anchor UPF is determined from the candidate UPFs based on the real-time cell location information of the candidate UPFs. Based on the new anchor UPF and the user intent, the computing power service request is processed in collaboration with the terminal and the computing power support of the cloud big model is provided to the terminal.

[0029] If the KQI guarantee representation does not require the anchor UPF to be replaced, the anchor UPF will be used as the new anchor UPF, and the SLA guarantee strategy recommendation will be subscribed to from the NWDAF based on the terminal computing power indicator and the computing power service request indicator.

[0030] Send a user session update suggestion to the PCF to establish an association between the PCF and the new anchored UPF.

[0031] In conjunction with the second aspect, in an exemplary embodiment, based on the new anchored UPF and user intent, the system collaboratively processes computing power service requests with the terminal and provides computing power support for large cloud models to the terminal, including:

[0032] The NWDAF reports the experience data of the subscription terminal to the new anchor UPF. The subscription information carries the user identifier, so that the new anchor UPF will upload the experience data of the target user corresponding to the user identifier to the NWDAF. The NWDAF predicts the network support service status based on the experience data, and generates an initial SLA guarantee strategy suggestion based on the predicted network support service status and the terminal's needs, and returns it to the cloud.

[0033] Receive initial SLA guarantee strategy recommendations, and plan the solution based on the initial SLA guarantee strategy recommendations and user intent to obtain the corresponding execution plan;

[0034] The computing power service requests are processed according to the execution plan.

[0035] In conjunction with the second aspect, in one embodiment, the computing power service request is processed according to the execution plan, including:

[0036] At the physical layer, uplink / downlink CSI channel switching is performed; at the network layer, bandwidth adaptation is performed; and at the application layer, QoS dedicated load policies are established. Service requirements are decomposed, calculated, and coordinated to match the service characteristics of computing power service requests, and SLA guarantee policy recommendations are generated.

[0037] The PCF sends SLA guarantee policy recommendations and policy conflict priorities to the PCF. The PCF generates corresponding SLA guarantee rules based on the SLA guarantee policy recommendations, policy conflict priorities, and terminal location information, and sends them to the SMF so that the SMF can establish a dedicated load based on the SLA guarantee rules. If the dedicated load recommendation is successful, the PCF sends a successful dedicated load establishment indication to the NWDAF via the cloud.

[0038] In conjunction with the second aspect, in one embodiment, the method further includes:

[0039] If the SLA guarantee strategy recommendation cannot meet the SLA guarantee requirements, the system receives the updated initial SLA guarantee strategy recommendation sent by the NWDAF. The updated initial SLA guarantee strategy recommendation is obtained by the NWDAF receiving new experience data of the target user reported by the new anchor UPF, predicting the network support service status based on the new experience data, obtaining the new predicted network support service status, and updating the initial SLA guarantee strategy recommendation based on the new predicted network support service status and the terminal's needs.

[0040] Based on the updated initial SLA assurance strategy recommendations and user intent, a new implementation plan is developed.

[0041] The computing power service requests will be processed according to the new implementation plan.

[0042] Thirdly, this application also provides a computing power service request processing method, applied to the edge side, including:

[0043] Receive terminal computing power indicator forwarded by the cloud; the terminal computing power indicator is obtained by the terminal when the target user initiates a computing power service request and is sent to the cloud. The terminal computing power indicator carries the user intent of the target user.

[0044] Based on the terminal's computing power indicator, the terminal's geographical location information, and the target user's SLA guarantee requirements, the cloud instructs the terminal to interact with the edge side to process computing power service requests. Then, based on the current network status of the network to which the edge side belongs and the terminal's computing power indicator, it obtains the KQI indicator that matches the computing power service request.

[0045] If the KQI metric does not meet the SLA guarantee requirements, a computing power service request indicator for the target user is generated and sent to the cloud. The cloud is used to process the computing power service request in collaboration with the terminal based on the user's intent and the terminal's computing power indicator and computing power service request indicator, and to provide the terminal with computing power support for the cloud's large model.

[0046] In conjunction with the third aspect, in one embodiment, the method further includes:

[0047] Provided that the KQI metric meets the SLA guarantee requirements, it interacts with the terminal to process computing power service requests.

[0048] In conjunction with the third aspect, in one embodiment, the KQI metric includes at least one of the following: number of stutters, stutter duration, bit rate, resolution, response latency, setup latency, jitter, and bit error rate.

[0049] Fourthly, this application also provides a method for processing computing power service requests, applied to a terminal, including:

[0050] When a target user initiates a computing power service request, the terminal computing power indicator is obtained and uploaded to the cloud. The terminal computing power indicator carries the user intent of the target user. The cloud is used to forward the terminal computing power indicator to the edge side and, based on the terminal computing power indicator, the terminal's geographical location information, and the target user's SLA guarantee requirements, instruct the terminal to interact with the edge side to process the computing power service request.

[0051] The edge side is used to obtain KQI indicators that match the computing power service request based on the current network status of the network to which the edge side belongs and the terminal computing power indicator; if the KQI indicators do not meet the SLA guarantee requirements, it generates the computing power service request indicator for the target user and sends it to the cloud.

[0052] The cloud is used to process computing power service requests in collaboration with the terminal based on user intent, terminal computing power indicator, and computing power service request indicator, and to provide computing power support for the cloud's large model to the terminal.

[0053] In conjunction with the fourth aspect, in one embodiment, obtaining the terminal computing power indicator includes:

[0054] Acquire terminal communication data, user behavior data, personalized preference data, terminal computing power information, load information, and link quality information, as well as the session scenario information of the terminal;

[0055] Based on session scenario information, terminal computing power information, load information, link quality information, and computing power service requests, a terminal computing power indicator is generated.

[0056] User intent is obtained through the following steps:

[0057] Local processing is performed on communication data, user behavior data, and personalized preference data to obtain user intent; local processing includes encryption for personal privacy data protection, semantic query, and intent recognition.

[0058] In conjunction with the fourth aspect, in one embodiment, communication data, user behavior data, and personalized preference data are processed locally to obtain user intent, including:

[0059] Malicious and sensitive content in communication data, user behavior data, and personalized preference data is filtered to obtain cleaned communication data, user behavior data, and personalized preference data.

[0060] Local data processing is performed on the cleaned communication data, user behavior data, and personalized preference data to obtain user intent.

[0061] Fifthly, this application also provides a computing power service request processing device, applied in the cloud, comprising:

[0062] The first receiving module is used to receive the terminal computing power indicator uploaded by the terminal; the terminal computing power indicator is obtained by the terminal after the target user initiates a computing power service request; the terminal computing power indicator carries the user intent of the target user;

[0063] The first processing module is used to forward the terminal computing power indicator to the edge side, and instruct the terminal to interact with the edge side to process computing power service requests based on the terminal computing power indicator, the terminal's geographical location information and the target user's SLA guarantee requirements.

[0064] The second receiving module is used to receive the computing power service request indicator of the target user sent by the edge side. The computing power service request indicator is obtained by the edge side based on the current network status of the network to which the edge side belongs and the terminal computing power indicator, and is generated when the KQI indicator does not meet the SLA guarantee requirements.

[0065] The second processing module is used to process computing power service requests in collaboration with the terminal based on user intent, terminal computing power indicator, and computing power service request indicator, and to provide computing power support for cloud-based large models to the terminal.

[0066] Sixthly, this application also provides a computing power service request processing device, applied at the edge, comprising:

[0067] The third receiving module is used to receive the terminal computing power indicator forwarded by the cloud. The terminal computing power indicator is obtained by the terminal when the target user initiates a computing power service request and is sent to the cloud. The terminal computing power indicator carries the user intent of the target user.

[0068] The first acquisition module is used to instruct the terminal to interact with the edge side to process computing power service requests based on the terminal computing power indicator, the terminal's geographical location information, and the target user's SLA guarantee requirements in the cloud. Then, based on the current network status of the network to which the edge side belongs and the terminal computing power indicator, it acquires the KQI indicator that matches the computing power service request.

[0069] The generation module is used to generate a computing power service request indicator for the target user and send it to the cloud when the KQI indicator does not meet the SLA guarantee requirements. The cloud is used to process the computing power service request in collaboration with the terminal and provide computing power support for the cloud's large model to the terminal based on the user's intent and the terminal's computing power indicator and computing power service request indicator.

[0070] Seventhly, this application also provides a computing power service request processing device, applied to a terminal, comprising:

[0071] The second acquisition module is used to acquire the terminal computing power indicator and upload it to the cloud when the target user initiates a computing power service request. The terminal computing power indicator carries the user intent of the target user. The cloud is used to forward the terminal computing power indicator to the edge side and instruct the terminal to interact with the edge side to process the computing power service request based on the terminal computing power indicator, the terminal's geographical location information and the target user's SLA guarantee requirements.

[0072] The edge side is used to obtain KQI indicators that match the computing power service request based on the current network status of the network to which the edge side belongs and the terminal computing power indicator; if the KQI indicators do not meet the SLA guarantee requirements, it generates the computing power service request indicator for the target user and sends it to the cloud.

[0073] The cloud is used to process computing power service requests in collaboration with the terminal based on user intent, terminal computing power indicator, and computing power service request indicator, and to provide computing power support for the cloud's large model to the terminal.

[0074] Eighthly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement various computing power service request processing methods.

[0075] Ninthly, this application also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements various aspects of computing power service request processing methods.

[0076] In a tenth aspect, this application also provides a computer program product, including a computer program that, when executed by a processor, implements various aspects of computing power service request processing methods.

[0077] The aforementioned computing power service request processing system, method, apparatus, computer equipment, computer-readable storage medium, and computer program product, when a target user initiates a computing power service request, allows the terminal to obtain a terminal computing power indicator carrying the target user's intent and upload it to the cloud. The cloud forwards the terminal computing power indicator to the edge side and, based on the terminal computing power indicator, the terminal's geographical location information, and the target user's SLA guarantee requirements, instructs the terminal to interact with the edge side to process the computing power service request. The edge side, based on the current network status of its network and the terminal computing power indicator, obtains a KQI indicator matching the computing power service request. If the KQI indicator does not meet the SLA guarantee requirements, a computing power service request indicator for the target user is generated and sent to the cloud. Finally, the cloud, based on the user intent, the terminal computing power indicator, and the computing power service request indicator, collaborates with the terminal to process the computing power service request and provides the terminal with computing power support for large cloud models. By efficiently coordinating end-edge-cloud computing power resources, the terminal can achieve low-power, high-real-time processing of computing power tasks, ensuring SLA compliance and improving the user experience of AI terminal applications. Attached Figure Description

[0078] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0079] Figure 1 This is an application environment diagram of a computing power service processing method in one embodiment;

[0080] Figure 2 This is a system framework diagram of a computing power service request processing system in one embodiment;

[0081] Figure 3 This is a flowchart illustrating a computing power service request processing method in one embodiment;

[0082] Figure 4 This is a flowchart illustrating the computing power service request processing method in another embodiment;

[0083] Figure 5 This is an interactive diagram of a computing power service request processing method in one embodiment;

[0084] Figure 6 This is a schematic diagram of the computing power process coordination and dynamic task unloading process in another embodiment;

[0085] Figure 7 This is a structural block diagram of a computing power service request processing device in one embodiment;

[0086] Figure 8 This is a structural block diagram of a computing power service request processing device in another embodiment;

[0087] Figure 9 This is a structural block diagram of the computing power service request processing device in another embodiment;

[0088] Figure 10 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0089] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0090] The computing power service request processing method provided in this application embodiment can be applied to, for example, Figure 1In the application environment shown, terminal 106 communicates with edge device 104 and cloud device 102 via a network. The data storage system can store data that needs to be processed by edge device 104 and cloud device 102. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, projection devices, etc. Portable wearable devices can include smartwatches, smart bracelets, head-mounted devices, etc. Head-mounted devices can be virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, etc. Edge device 104 and cloud device 102 can be independent physical servers, server clusters or distributed systems composed of multiple physical servers, or cloud servers providing cloud computing services.

[0091] In one exemplary embodiment, such as Figure 2 As shown, a computing power service request processing system is provided, including a terminal, an edge device, and a cloud.

[0092] The terminal is used to obtain the terminal computing power indicator and upload it to the cloud when the target user initiates a computing power service request; the terminal computing power indicator carries the user intent of the target user;

[0093] In the cloud, it is used to forward terminal computing power indicators to the edge side, and instruct the terminal to interact with the edge side to process computing power service requests based on the terminal computing power indicators, the terminal's geographical location information, and the target user's SLA guarantee requirements.

[0094] On the edge side, it is used to obtain KQI indicators that match the computing power service request based on the current network status of the network to which the edge side belongs and the terminal computing power indicator; if the KQI indicators do not meet the SLA guarantee requirements, it generates the computing power service request indicator for the target user and sends it to the cloud.

[0095] In the cloud, it is used to process computing power service requests in collaboration with the terminal based on user intent, terminal computing power indicator, and computing power service request indicator, and to provide computing power support for the cloud-based large model to the terminal.

[0096] In this context, the target user can be understood as a user who has signed an SLA (Service Level Agreement), the computing power service request can be understood as a request to obtain or allocate computing resources and computing power-related service capabilities, the terminal computing power indicator can be understood as an identifier used in computing power service or resource scheduling scenarios to express "the computing power status / capabilities that the end user can see or obtain", and the user intent can be understood as the user's true intention in initiating the computing power service request.

[0097] The edge side can be understood as the edge computing gateway server deployed with a large-scale communication industry model, and the SLA guarantee requirements can be understood as the basic requirements that users and service providers should meet regarding service quality, availability, performance, etc.

[0098] Among them, the current network status can be understood as the status information of the network to which the edge side belongs, which can include availability and connectivity, performance indicators, routing and network paths, service quality and constraints, etc. KQI indicators can be understood as core indicators used to measure the actual user experience and service quality, which can include the number of stutters, stutter duration, bit rate, resolution, response latency, setup latency, jitter, bit error rate, etc. Computing power service request indicator can be understood as a quantitative description of the user or system's request signal for computing resources, which usually includes the quantity and timeliness requirements of resources such as CPU, GPU, memory, storage, and network bandwidth.

[0099] Optionally, when a target user who has signed an SLA initiates a computing power service request, the terminal obtains the target user's user intent and the terminal's computing power indicator. The terminal's computing power indicator carries the user intent and is uploaded to the cloud. The cloud forwards the terminal's computing power indicator to the edge side and determines the corresponding anchor UPF based on the terminal's computing power indicator, the terminal's geographical location information, and the target user's SLA guarantee requirements. The cloud instructs the terminal to interact with the edge side to process the computing power service request based on the anchor UPF. The edge side obtains the KQI indicator that matches the computing power service request based on the current network status of the network to which the edge side belongs and the terminal's computing power indicator. If the KQI indicator does not meet the SLA guarantee requirements, the edge side generates a computing power service request indicator for the target user and sends it to the cloud. The cloud, based on the user intent, the terminal's computing power indicator, and the computing power service request indicator, collaborates with the terminal to process the computing power service request and provides the terminal with computing power support from the cloud's large model.

[0100] In the aforementioned computing power service request processing system, when a target user initiates a computing power service request, the terminal obtains a terminal computing power indicator carrying the target user's intent and uploads it to the cloud. The cloud forwards the terminal computing power indicator to the edge side and, based on the terminal computing power indicator, the terminal's geographical location information, and the target user's SLA guarantee requirements, instructs the terminal to interact with the edge side to process the computing power service request. The edge side obtains a KQI indicator matching the computing power service request based on the current network status of its network and the terminal computing power indicator. If the KQI indicator does not meet the SLA guarantee requirements, a computing power service request indicator for the target user is generated and sent to the cloud. Finally, the cloud, based on the user intent, the terminal computing power indicator, and the computing power service request indicator, collaborates with the terminal to process the computing power service request and provides the terminal with computing power support for large cloud models. By efficiently coordinating end-edge-cloud computing power resources, the terminal can achieve low-power, high-real-time processing of computing power tasks, ensuring SLA compliance and improving the user experience of AI terminal applications.

[0101] In one embodiment, the cloud is further configured to determine the anchor UPF corresponding to the terminal based on the terminal's geographical location information and SLA guarantee requirements, under the instruction of the terminal computing power symbol, so as to instruct the terminal to interact with the edge side to process computing power service requests based on the anchor UPF.

[0102] Among them, the anchored UPF can be understood as the network element component that configures the protection application (SubAppID) and the protection application category to which it belongs.

[0103] For example, under the instruction of the terminal's computing power indicator, the cloud obtains the terminal's geographical location information and selects the corresponding anchored UPF from the candidate UPF pool based on the terminal's geographical location information and SLA guarantee requirements. This instructs the terminal to interact with the edge side to process computing power service requests based on the anchored UPF. By using geographically proximate anchored UPFs and edge computing resources, end-to-end latency is significantly reduced, real-time performance and interactive experience are improved, thereby ensuring the timely processing of computing power service requests.

[0104] In one embodiment, the cloud is further configured to obtain the KQI guarantee required by the target user based on the computing power service request indicator. If the KQI guarantee representation requires anchor UPF replacement, a new anchor UPF is determined from the candidate UPFs based on the real-time cell location information of the candidate UPFs. Based on the new anchor UPF and the user's intent, the cloud collaborates with the terminal to process the computing power service request and provides the terminal with computing power support from the cloud's large model. If the KQI guarantee representation does not require anchor UPF replacement, the anchor UPF is used as the new anchor UPF, and SLA guarantee policy recommendations are subscribed to from the NWDAF based on the terminal's computing power indicator and computing power service request indicator. The cloud is also configured to send user session update recommendations to the PCF to establish an association between the PCF and the new anchor UPF.

[0105] Among them, KQI assurance can be understood as the specific KQI indicators that need to be met, NWDAF can be understood as configuring the application to be assured (SubAppID) and the application to be assured (AppID); the network element component that configures the GFBR / MFBR information of the application to be assured, PCF can be understood as configuring the corresponding quality poor assurance scheme based on the application category (AppID); and the network element component that discovers NWDAF based on TA and event type service.

[0106] Optionally, after obtaining the specific KQI indicators that the target user is not meeting based on the computing power service request indicator, the cloud determines whether to perform an anchor UPF replacement based on the specific non-compliant KQI indicators. If it is determined that an anchor UPF replacement is required, a new anchor UPF is selected from the candidate UPFs based on the real-time cell location information of the candidate UPFs. Based on the new anchor UPF and the user's intent, the cloud collaborates with the terminal to process the computing power service request and provides computing power support from the cloud's large-scale model to the terminal. If it is determined that an anchor UPF replacement is not required, the anchor UPF is used as the new anchor UPF. Based on the terminal's computing power indicator and computing power service request indicator, the cloud subscribes to the NWDAF for SLA guarantee policy suggestions. The cloud also uses this to send user session update suggestions to the PCF to establish an association between the PCF and the new anchor UPF. By combining the anchor UPF, which is closer to the terminal, with edge computing power resources, transmission latency and queue waiting time are reduced. Dynamic scheduling and anchor switching respond quickly when indicators are not met, accelerating the achievement of key SLA indicators.

[0107] In an exemplary embodiment, the cloud is further configured to report experience data of the new anchored UPF subscription terminal via NWDAF. The subscription information carries a user identifier, enabling the new anchored UPF to upload the experience data of the target user corresponding to the user identifier to NWDAF. NWDAF predicts the network support service status based on the experience data, and generates an initial SLA guarantee strategy suggestion based on the predicted network support service status and the terminal's needs, and returns it to the cloud. The cloud is further configured to receive the initial SLA guarantee strategy suggestion, and plan a solution based on the initial SLA guarantee strategy suggestion and user intent to obtain a corresponding execution plan. The computing power service request is then processed according to the execution plan.

[0108] Among them, experience data can be understood as the service experience on the terminal side.

[0109] For example, the cloud subscribes to the NWDAF for initial SLA guarantee policy recommendations. The NWDAF, based on the subscription information, reports the experience data of the subscribed terminal to the new anchor UPF. The subscription information carries a user identifier, enabling the new anchor UPF to upload the experience data of the target user corresponding to the user identifier to the NWDAF. The NWDAF predicts the network support service status based on the experience data, wireless cell load, and previous network quality. Based on the predicted network support service requests and the terminal's requirements, it determines whether to generate initial SLA guarantee policy recommendations. If generation is determined, initial SLA guarantee policy recommendations are generated and returned to the cloud. The cloud receives the initial SLA guarantee policy recommendations and, based on the initial SLA guarantee policy recommendations and user intent, plans a solution to obtain the corresponding execution plan, and processes the computing power service requests according to the execution plan. By predictively adjusting resources, the impact of sudden congestion on the experience is reduced, improving the SLA achievement rate. Experience data-driven strategies are elevated to session-level dynamic adjustments, improving the availability and reliability of critical tasks (low-latency / high-throughput tasks). Data aggregation and analysis by geographical region and cell dimension facilitates compliance with local regulations and privacy requirements.

[0110] In one embodiment, the cloud is further configured to perform uplink / downlink CSI channel switching at the physical layer, perform bandwidth adaptation at the network layer, and establish QoS dedicated load policies at the application layer; perform service requirement decomposition, estimation, and coordination that matches the service characteristics of computing power service requests; generate SLA guarantee policy recommendations; send SLA guarantee policy recommendations and policy conflict priorities to the PCF; the PCF is configured to generate corresponding SLA guarantee rules based on the SLA guarantee policy recommendations, policy conflict priorities, and terminal location information, and send them to the SMF, so that the SMF can establish a dedicated load according to the SLA guarantee rules; and if the dedicated load recommendation is successful, send a dedicated load establishment success indication to the NWDAF through the cloud.

[0111] Among them, CSI (Channel State Information) can be understood as a collection of current state information of the wireless channel; QoS (Quality of Service) dedicated policies can be understood as a set of rules that clearly define "who goes first / who has priority, how many resources to allocate, when to allocate, and how to degrade" at the system's scheduling / resource allocation layer; SMF can be understood as configuring SLA guarantee functions and selecting network element components of the intelligent UPF according to the instructions issued by PCF.

[0112] Optionally, the cloud performs uplink / downlink CSI channel switching at the physical layer, performs bandwidth sub-adaptation at the network layer, and suggests QoS dedicated load policies at the application layer. It performs service requirement decomposition, estimation, and coordination that matches the service characteristics of computing power service requests, generates SLA guarantee policy suggestions, and sends SLA guarantee policy suggestions and policy conflict priorities to the PCF. The PCF makes further decisions based on the SLA guarantee policy suggestions, policy conflict priorities, and terminal location information, generates corresponding SLA guarantee rules, and sends them to the SMF, so that the SMF can suggest dedicated load according to the SLA guarantee rules. If the dedicated load suggestion is successful, the cloud sends a feedback indication of successful dedicated load suggestion to the NWDAF.

[0113] By executing corresponding policies at the physical and network layers, the cloud updates and adjusts the user's intent and the initial SLA policy recommendations received, making the SLA policy recommendations sent to the PCF more closely aligned with user needs, thereby ensuring the user's service experience.

[0114] In one embodiment, the cloud is further configured to receive an updated initial SLA guarantee strategy recommendation sent by the NWDAF when the SLA guarantee strategy recommendation cannot meet the SLA guarantee requirements; plan a solution based on the updated initial SLA guarantee strategy recommendation and user intent to obtain a corresponding new execution plan; process computing power service requests according to the new execution plan; the updated initial SLA guarantee strategy recommendation is obtained by the NWDAF receiving new experience data of the target user reported by the new anchor UPF; predicting the network support service status based on the new experience data to obtain a new predicted network support service status; and updating the initial SLA guarantee strategy recommendation based on the new predicted network support service status and the terminal's needs.

[0115] For example, the NWDAF receives new experience data of the target user reported by the new anchored UPF, predicts the network support service status based on the new experience data, obtains a new predicted network support service status, updates the initial SLA guarantee policy recommendation based on the new predicted network support service status and the terminal's needs, and sends the updated initial SLA guarantee policy recommendation to the cloud. If the SLA guarantee policy recommendation cannot meet the SLA guarantee requirements, the cloud receives the updated initial SLA guarantee policy recommendation sent by the NWDAF, and plans a solution based on the updated initial SLA guarantee policy recommendation and the user's intent to obtain a corresponding new execution plan; and processes the computing power service request according to the new execution plan. A clear division of labor and collaboration has been formed between the cloud and NWDAF. When the SLA guarantee policy recommendation cannot meet the SLA guarantee requirements, NWDAF will update the initial SLA guarantee policy recommendation and upload the updated initial SLA guarantee policy recommendation to the cloud. The cloud will then determine whether to update the updated initial SLA guarantee policy recommendation based on the user's intent. This makes the SLA guarantee policy recommendation issued to PCF more suitable for the current application environment, thereby ensuring the processing effect and timeliness of the target user's computing power service requests.

[0116] In an exemplary embodiment, the edge side is further configured to interact with the terminal to process computing power service requests, provided that the KQI metrics meet the SLA guarantee requirements; the KQI metrics include at least one of the following: number of stutters, stutter duration, bit rate, resolution, response latency, setup latency, jitter, and bit error rate.

[0117] Optionally, if at least one of the KQI metrics, including the number of stutters, stutter duration, bitrate, resolution, response latency, setup latency, jitter, and bit error rate, meets the pre-set threshold requirements in the SLA guarantee, the edge side interacts with the terminal to process computing power service requests. When the KQI metrics meet the SLA guarantee requirements, the edge side directly interacts with the terminal to process computing power service requests, ensuring processing efficiency and effectiveness for the requests.

[0118] In one embodiment, the terminal is further configured to acquire terminal communication data, user behavior data, personalized preference data, terminal computing power information, load information, and link quality information, as well as the session scenario information in which the terminal is located; generate a terminal computing power indicator based on the session scenario information, terminal computing power information, load information, link quality information, and computing power service requests; and perform local processing on the communication data, user behavior data, and personalized preference data to obtain user intent; the local processing includes personal privacy data protection encryption, semantic query, and intent recognition.

[0119] Among these, communication data can be understood as underlying data directly related to network transmission, including channel information, traffic statistics, forwarding paths, packet reception / transmission timestamps, bit rate, bit error rate, latency, packet loss, etc.; user behavior data can be understood as records of actual user behavior in the application, such as clicks, swipes, session duration, session frequency, session interruption points, retries, browsing paths, location-triggered interactions, etc.; personalized preference data can be understood as explicit user preferences (such as language, theme, application settings) and implicit preferences (such as frequently used functions, time period preferences, content type preferences, privacy / security preferences); terminal computing power information can be understood as the status of available computing resources on the terminal device, including the utilization rate of CPU / GPU / NPU / dedicated acceleration units. Available memory, available computing bandwidth, power consumption status, thermal characteristics, current task load, etc.; load information can be understood as the workload data of the system / edge node, UPF, gateway, computing node, etc. at a certain moment, such as CPU / memory / network bandwidth usage, queue length, queuing waiting time, request arrival rate, etc.; link quality information can be understood as the performance indicators related to the transmission link, such as round-trip time, end-to-end latency, jitter, packet loss rate, throughput, bit error rate, channel state information, MOS / experience score, etc.; session scenario information can be understood as the specific session scenario description of the current terminal at the application layer, such as entertainment / office / real-time game / AR / VR, whether it is in the edge computing collaboration area, whether collaborative inference is performed, whether a low-latency path is required, etc.

[0120] For example, the terminal acquires call data, user behavior data, personalized preference data, terminal computing power information, load information, and link quality information, as well as the session scenario information in which the terminal is located. Based on the session scenario information, terminal computing power information, load information, link quality information, and computing power service requests, it generates a terminal computing power indicator. It also performs personal privacy data protection encryption, semantic querying, and intent recognition on the communication data, user behavior data, and personalized preference data to identify the target user's intent. By performing data processing and data integration, the terminal computing power indicator and user intent are generated, laying the data foundation for subsequent processing of computing power service requests.

[0121] In one embodiment, the terminal is further configured to filter malicious and sensitive content in communication data, user behavior data, and personalized preference data to obtain cleaned communication data, user behavior data, and personalized preference data; and to perform local data processing on the cleaned communication data, user behavior data, and personalized preference data to obtain user intent.

[0122] Among them, malicious sensitive content can be understood as both malicious content and sensitive content. Malicious content can be understood as content used to harm, deceive, interfere with, disrupt, illegally obtain information, or infringe on the legitimate rights and interests of others, including but not limited to: malicious software / phishing links, extortion, spreading hate speech, inciting violence, harassment and threats, account theft, and falsification of information. Sensitive content can be understood as content that may cause privacy, ethical, legal, and social risks. Even if it is not intentionally harmful, it still needs to be strictly controlled. Common dimensions include: personal identity information, medical / financial information, child-related content, sexual content, violent descriptions, extreme political / hate speech, harassment and bullying, etc.

[0123] Optionally, the terminal filters malicious and sensitive content from communication data, user behavior data, and personalized preference data to obtain cleaned communication data, user behavior data, and personalized preference data. This cleaned data is then subjected to privacy protection encryption, semantic querying, and intent recognition to identify the target user's intent. By performing corresponding data cleaning operations on each data item, the accuracy of the identified user intent is improved.

[0124] In one exemplary embodiment, such as Figure 3 As shown, a method for processing computing power service requests is provided, applicable to applications such as... Figure 1 The cloud 102 includes steps S301 to S304, wherein:

[0125] Step S301: Receive the terminal computing power indicator uploaded by the terminal; the terminal computing power indicator is obtained by the terminal after the target user initiates a computing power service request; the terminal computing power indicator carries the user intent of the target user.

[0126] For example, when a target user who has signed an SLA initiates a computing power service request, terminal 106 obtains the target user's user intent and the terminal's computing power indicator, and uploads the terminal computing power indicator carrying the user intent to cloud 102, and cloud 102 receives the terminal computing power indicator carrying the user intent.

[0127] Step S302: Forward the terminal computing power indicator to the edge side, and instruct the terminal to interact with the edge side to process computing power service requests based on the terminal computing power indicator, the terminal's geographical location information, and the target user's SLA guarantee requirements.

[0128] Optionally, the cloud 102 forwards the terminal computing power indicator to the edge side 104, and determines the anchor UPF corresponding to the terminal 106 based on the terminal computing power indicator, the geographical location information of the terminal 106 and the SLA guarantee requirements of the target user, and instructs the terminal 106 to interact with the edge side 104 to process computing power service requests based on the anchor UPF.

[0129] Step S303: Receive the computing power service request indicator of the target user sent by the edge side; the computing power service request indicator is generated by the edge side based on the current network status of the network to which the edge side belongs and the terminal computing power indicator, obtaining the KQI indicator that matches the computing power service request, and generating it if the KQI indicator does not meet the SLA guarantee requirements.

[0130] For example, the edge side 104 obtains the KQI indicator that matches the computing power service request based on the current network status of the network to which the edge side 104 belongs and the terminal computing power indicator. If the KQI indicator does not meet the SLA guarantee requirements, the edge side 104 generates the computing power service request indicator of the target user and sends it to the cloud 102. The cloud 102 receives the computing power service request indicator of the target user.

[0131] Step S304: Based on the user intent, the terminal computing power indicator, and the computing power service request indicator, work with the terminal to process the computing power service request and provide the terminal with computing power support for the cloud-based large model.

[0132] Optionally, Cloud102 Cloud, based on user intent, terminal computing power indicator, and computing power service request indicator, collaborates with the terminal to process computing power service requests and provides computing power support for the cloud-based large model to the terminal.

[0133] In one embodiment, based on the terminal computing power indicator, the terminal's geographical location information, and the target user's SLA guarantee requirements, the terminal is instructed to interact with the edge side to process computing power service requests. This includes: under the instruction of the terminal computing power indicator, determining the corresponding anchor UPF of the terminal based on the terminal's geographical location information and SLA guarantee requirements, so as to instruct the terminal to interact with the edge side to process computing power service requests based on the anchor UPF.

[0134] For example, under the instruction of the terminal computing power symbol, the cloud 102 obtains the geographical location information of the terminal 106, and selects the anchored UPF corresponding to the terminal 106 from the candidate UPF pool based on the geographical location information of the terminal 106 and the SLA guarantee requirements, so as to instruct the terminal 106 to interact with the edge side 104 to process computing power service requests based on the anchored UPF. By using geographically proximate anchored UPFs and edge computing resources, end-to-end latency is significantly reduced, real-time performance and interactive experience are improved, thereby ensuring the timeliness of processing computing power service requests.

[0135] In one embodiment, based on user intent, terminal computing power indicator, and computing power service request indicator, the system collaborates with the terminal to process computing power service requests and provides computing power support for cloud-based large models to the terminal. This includes: obtaining the KQI guarantee required by the target user based on the computing power service request indicator; if the KQI guarantee representation requires anchor UPF replacement, determining a new anchor UPF from the candidate UPFs based on the real-time cell location information of the candidate UPFs; and, based on the new anchor UPF and user intent, collaboratively processing the computing power service request with the terminal and providing computing power support for cloud-based large models to the terminal; if the KQI guarantee representation does not require anchor UPF replacement, using the anchor UPF as the new anchor UPF, and subscribing to SLA guarantee policy recommendations from the NWDAF based on the terminal computing power indicator and computing power service request indicator; and sending user session update recommendations to the PCF to establish an association between the PCF and the new anchor UPF.

[0136] Optionally, after obtaining the specific KQI indicators that the target user is not meeting based on the computing power service request indicator, cloud 102 determines whether to perform anchor UPF replacement based on the specific non-compliant KQI indicators. If it is determined that anchor UPF replacement is required, a new anchor UPF is selected from the candidate UPFs based on the real-time cell location information of the candidate UPFs. Based on the new anchor UPF and the user's intent, it collaborates with terminal 106 to process the computing power service request and provides computing power support from the cloud-based large model to terminal 106. If it is determined that anchor UPF replacement is not required, the anchor UPF is used as the new anchor UPF. Based on the terminal's computing power indicator and computing power service request indicator, it subscribes to SLA guarantee policy suggestions from NWDAF and sends user session update suggestions to PCF to establish an association between PCF and the new anchor UPF. By combining anchor UPFs closer to the terminal with edge computing power resources, transmission latency and queue waiting time are reduced, and dynamic scheduling and anchor switching respond quickly when indicators are not met, accelerating the achievement of key SLA indicators.

[0137] In an exemplary embodiment, based on the new anchor UPF and user intent, the system collaboratively processes computing power service requests with the terminal and provides computing power support from the cloud-based large model to the terminal. This includes: subscribing to the terminal's experience data report via NWDAF to the new anchor UPF, with the subscription information carrying a user identifier, so that the new anchor UPF uploads the experience data of the target user corresponding to the user identifier to the NWDAF; the NWDAF predicts the network support service status based on the experience data, and generates an initial SLA guarantee strategy suggestion based on the predicted network support service status and the terminal's needs, and returns it to the cloud; receiving the initial SLA guarantee strategy suggestion, and planning a solution based on the initial SLA guarantee strategy suggestion and user intent to obtain a corresponding execution plan; and processing the computing power service request according to the execution plan.

[0138] For example, cloud 102 subscribes to the NWDAF for initial SLA guarantee policy recommendations. The NWDAF, based on the subscription information, reports the experience data of the subscribed terminal to the new anchor UPF. The subscription information carries a user identifier, enabling the new anchor UPF to upload the experience data of the target user corresponding to the user identifier to the NWDAF. The NWDAF predicts the network support service status based on the experience data, wireless cell load, and previous network quality. Based on the predicted network support service requests and the terminal's requirements, it determines whether to generate initial SLA guarantee policy recommendations. If generation is determined, initial SLA guarantee policy recommendations are generated and returned to the cloud. Cloud 102 receives the initial SLA guarantee policy recommendations and plans a solution based on the recommendations and user intent, obtaining a corresponding execution plan. The cloud then processes the computing power service requests according to the execution plan. By predictively adjusting resources, the impact of sudden congestion on the experience is reduced, improving SLA achievement rate. Experience data-driven strategies are elevated to session-level dynamic adjustments, improving the availability and reliability of critical tasks (low-latency / high-throughput tasks). Data aggregation and analysis by geographical region and cell dimension facilitates compliance with local regulations and privacy requirements.

[0139] In one embodiment, the computing power service request is processed according to the execution plan, including: performing uplink / downlink CSI channel switching at the physical layer, performing bandwidth adaptation at the network layer, and establishing a QoS dedicated load policy at the application layer; performing service requirement decomposition, estimation, and coordination that matches the service characteristics of the computing power service request; generating SLA guarantee policy recommendations; sending SLA guarantee policy recommendations and policy conflict priorities to the PCF; the PCF generates corresponding SLA guarantee rules based on the SLA guarantee policy recommendations, policy conflict priorities, and terminal location information, and sends them to the SMF, so that the SMF can establish a dedicated load according to the SLA guarantee rules; and if the dedicated load recommendation is successful, sending a dedicated load establishment success indication to the NWDAF via the cloud.

[0140] Optionally, the cloud 102 performs uplink / downlink CSI channel switching at the physical layer, performs bandwidth sub-adaptation at the network layer, and suggests QoS dedicated load policies at the application layer. It performs service requirement decomposition, estimation, and coordination that matches the service characteristics of computing power service requests, generates SLA guarantee policy suggestions, and sends SLA guarantee policy suggestions and policy conflict priorities to the PCF. The PCF makes further decisions based on the SLA guarantee policy suggestions, policy conflict priorities, and terminal location information, generates corresponding SLA guarantee rules, and sends them to the SMF, so that the SMF can suggest dedicated load according to the SLA guarantee rules. If the dedicated load suggestion is successful, the cloud sends a feedback indication of successful dedicated load suggestion to the NWDAF.

[0141] By executing corresponding policies at the physical and network layers, the cloud updates and adjusts the user's intent and the initial SLA policy recommendations received, making the SLA policy recommendations sent to the PCF more closely aligned with user needs, thereby ensuring the user's service experience.

[0142] In one embodiment, the method further includes: receiving an updated initial SLA guarantee policy suggestion sent by the NWDAF when the SLA guarantee policy suggestion cannot meet the SLA guarantee requirements; the updated initial SLA guarantee policy suggestion is obtained by the NWDAF receiving new experience data of the target user reported by the new anchor UPF, predicting the network support service status based on the new experience data to obtain a new predicted network support service status, and updating the initial SLA guarantee policy suggestion based on the new predicted network support service status and the terminal's needs; planning a scheme based on the updated initial SLA guarantee policy suggestion and the user's intent to obtain a corresponding new execution scheme; and processing the computing power service request according to the new execution scheme.

[0143] For example, the NWDAF receives new experience data of the target user reported by the new anchor UPF, predicts the network support service status based on the new experience data, obtains a new predicted network support service status, updates the initial SLA guarantee policy recommendation based on the new predicted network support service status and the terminal's needs, and sends the updated initial SLA guarantee policy recommendation to the cloud 102. If the SLA guarantee policy recommendation cannot meet the SLA guarantee requirements, the cloud 102 receives the updated initial SLA guarantee policy recommendation sent by the NWDAF, and plans a solution based on the updated initial SLA guarantee policy recommendation and the user's intent to obtain a corresponding new execution plan; and processes the computing power service request according to the new execution plan. A clear division of labor and collaboration has been formed between the cloud and NWDAF. When the SLA guarantee policy recommendation cannot meet the SLA guarantee requirements, NWDAF will update the initial SLA guarantee policy recommendation and upload the updated initial SLA guarantee policy recommendation to the cloud. The cloud will then determine whether to update the updated initial SLA guarantee policy recommendation based on the user's intent. This makes the SLA guarantee policy recommendation issued to PCF more suitable for the current application environment, thereby ensuring the processing effect and timeliness of the target user's computing power service requests.

[0144] In one exemplary embodiment, such as Figure 4 As shown, a method for processing computing power service requests is provided, which is applied to... Figure 1 The edge side 104 in the middle includes steps S401 to S403, wherein:

[0145] Step S401: Receive the terminal computing power indicator forwarded by the cloud; the terminal computing power indicator is obtained by the terminal when the target user initiates a computing power service request and is sent to the cloud, and the terminal computing power indicator carries the user intent of the target user.

[0146] Optionally, when a target user who has signed an SLA initiates a computing power service request, the terminal 106 obtains the target user's user intent and obtains the terminal computing power indicator of the terminal 106. The terminal computing power indicator carries the user intent and is uploaded to the cloud 102. The cloud 102 forwards the terminal computing power indicator to the edge side 104, and the edge side 104 receives the terminal computing power indicator forwarded by the cloud.

[0147] Step S402: Based on the terminal's computing power indicator, the terminal's geographical location information, and the target user's SLA guarantee requirements, the cloud instructs the terminal to interact with the edge side to process the computing power service request. Then, based on the current network status of the network to which the edge side belongs and the terminal's computing power indicator, the cloud obtains the KQI indicator that matches the computing power service request.

[0148] For example, in the cloud 102, the anchor UPF corresponding to the terminal 106 is determined based on the terminal computing power indicator, the geographical location information of the terminal 106, and the SLA guarantee requirements of the target user. After instructing the terminal 106 to interact with the edge side 104 to process the computing power service request based on the anchor UPF, the edge side 104 obtains the KQI indicator that matches the computing power service request based on the current network status of the network to which the edge side belongs and the terminal computing power indicator.

[0149] Step S403: If the KQI indicator does not meet the SLA guarantee requirements, generate a computing power service request indicator for the target user and send it to the cloud. The cloud is used to process the computing power service request in collaboration with the terminal and provide computing power support for the cloud-based large model to the terminal based on the user's intent and the terminal's computing power indicator and computing power service request indicator.

[0150] Optionally, if the KQI indicator does not meet the SLA guarantee requirements, the edge side 104 generates a computing power service request indicator for the target user and sends it to the cloud 102. The cloud 102, based on the user intent carried by the terminal computing power indicator, the terminal computing power indicator, and the computing power service request indicator, works with the terminal 106 to process the computing power service request.

[0151] In one embodiment, the computing power service request is processed in interaction with the terminal, provided that the KQI metrics meet the SLA guarantee requirements. The KQI metrics include at least one of the following: number of stutters, stutter duration, bit rate, resolution, response latency, setup latency, jitter, and bit error rate.

[0152] For example, if at least one of the KQI metrics, including the number of stutters, stutter duration, bitrate, resolution, response latency, setup latency, jitter, and bit error rate, meets the pre-set threshold requirements in the SLA guarantee requirements, the edge device 104 and the terminal 106 interact to process computing power service requests. When the KQI metrics meet the SLA guarantee requirements, the edge device directly interacts with the terminal to process computing power service requests, ensuring processing efficiency and effectiveness for the requests.

[0153] In one exemplary embodiment, a computing power service request processing method is provided, applied to... Figure 1 Terminal 106, wherein:

[0154] When a target user initiates a computing power service request, the system acquires a terminal computing power indicator and uploads it to the cloud. The terminal computing power indicator carries the target user's intent. The cloud forwards the terminal computing power indicator to the edge side and, based on the terminal computing power indicator, the terminal's geographical location information, and the target user's SLA guarantee requirements, instructs the terminal to interact with the edge side to process the computing power service request. The edge side acquires a KQI indicator that matches the computing power service request based on the current network status of the network to which the edge side belongs and the terminal computing power indicator. If the KQI indicator does not meet the SLA guarantee requirements, the system generates a computing power service request indicator for the target user and sends it to the cloud. The cloud, based on the user intent, the terminal computing power indicator, and the computing power service request indicator, collaborates with the terminal to process the computing power service request and provides the terminal with computing power support from the cloud's large-scale model.

[0155] Optionally, when a target user who has signed an SLA initiates a computing power service request, terminal 106 obtains the target user's user intent and terminal computing power indicator, uploads the terminal computing power indicator carrying the user intent to the cloud, and cloud 102 forwards the terminal computing power indicator to edge 104. Based on the terminal computing power indicator, the geographical location information of terminal 106, and the target user's SLA guarantee requirements, cloud 102 determines the anchor UPF corresponding to terminal 106 and instructs terminal 106 to interact with edge 104 to process the computing power service request based on the anchor UPF. Edge 104 obtains the KQI indicator matching the computing power service request based on the current network status of the network to which edge 104 belongs and the terminal computing power indicator. If the KQI indicator does not meet the SLA guarantee requirements, edge 104 generates a computing power service request indicator for the target user and sends it to cloud 102. Cloud 102, based on the user intent, terminal computing power indicator, and computing power service request indicator, collaborates with terminal 106 to process the computing power service request.

[0156] In one embodiment, obtaining a terminal computing power indicator includes: obtaining the terminal's communication data, user behavior data, personalized preference data, terminal computing power information, load information, and link quality information, as well as the session scenario information of the terminal; generating a terminal computing power indicator based on the session scenario information, terminal computing power information, load information, link quality information, and computing power service requests; and obtaining user intent through the following steps: performing local processing on the communication data, user behavior data, and personalized preference data to obtain user intent; the local processing includes personal privacy data protection encryption, semantic query, and intent recognition.

[0157] For example, terminal 106 acquires call data, user behavior data, personalized preference data, terminal computing power information, load information, and link quality information, as well as the session scenario information of terminal 106. Based on the session scenario information, terminal computing power information, load information, link quality information, and computing power service requests, it generates a terminal computing power indicator. It also performs personal privacy data protection encryption, semantic querying, and intent recognition on the communication data, user behavior data, and personalized preference data to identify the target user's intent. By performing data processing and data integration, the terminal computing power indicator and user intent are generated, laying the data foundation for subsequent processing of computing power service requests.

[0158] In one embodiment, local processing of communication data, user behavior data, and personalized preference data to obtain user intent includes: filtering malicious and sensitive content in the communication data, user behavior data, and personalized preference data to obtain cleaned communication data, user behavior data, and personalized preference data; and performing local data processing on the cleaned communication data, user behavior data, and personalized preference data to obtain user intent.

[0159] Optionally, terminal 106 filters malicious and sensitive content from communication data, user behavior data, and personalized preference data to obtain cleaned communication data, user behavior data, and personalized preference data. It then performs privacy-protecting encryption, semantic querying, and intent recognition on the cleaned communication data, user behavior data, and personalized preference data to identify the target user's intent. By performing corresponding data cleaning operations on each data item, the accuracy of the identified user intent is improved.

[0160] In one exemplary embodiment, a method and system for efficient collaborative edge-cloud computing resources are proposed to address issues such as edge-cloud size model collaboration and edge-cloud task offloading.

[0161] After the terminal that has signed the SLA agreement installs the small model on the terminal side and connects to the network, the user initiates the initial registration and completes the authentication process.

[0162] The PCRF (4G) / PCF (5G) application HSS / UDM network element database stores SLA data in the user's subscribed package, distinguishes the protection policies configured by the protection application (SubAppID) and the corresponding protection application category (AppID), and makes real-time policy decisions.

[0163] When a user initiates a computing power service request, the terminal mini-model filters sensitive and malicious content, performs local data processing and reasoning such as personal privacy data protection encryption, semantic query, intent recognition, and tool invocation, and realizes low-latency, lightweight intelligent decision-making. It generates indicators such as the terminal's current situation, its own computing power and load, current link quality, and computing power service (single / continuous multiple requests) and uploads them to the edge computing gateway server.

[0164] The large-scale communication industry model deployed on the edge computing gateway server executes global policies based on the received instructions and the current network conditions, performs basic calculations that match the SLA guarantee of the current target user, realizes network orchestration and configuration execution, alleviates cloud computing pressure, and reduces latency;

[0165] The edge gateway server big model is based on a comprehensive analysis of cell congestion, computing load, time-frequency domain resource utilization, computing power service requirements of target service users, and their computing power and load status. When it is found that the KQI indicator matching the computing power service request does not meet the SLA guarantee requirements, a computing power service request indicator for the target user is generated, and massive computing power support is requested from the cloud-side big model to achieve intelligent resource scheduling and precise scenario response.

[0166] After receiving the request instruction, the cloud-side big model autonomously plans, remembers, acts and reflects, and coordinates the entire process of tool invocation, data aggregation and sharing, breaking down data silos between terminals and networks. Based on the business big model, it analyzes the communication intent and issues global policies to achieve intelligent on-demand scheduling of network computing power, on-demand expansion of application bandwidth, and on-demand guarantee of scenario latency.

[0167] In one embodiment, such as Figure 5 The diagram shows a flowchart of a computing power service request processing method, wherein:

[0168] Key network element configurations:

[0169] SMF: Configure SLA protection function and select intelligent UPF according to the instructions issued by PCF.

[0170] Intelligent UPF: Configure the application to be protected (SubAppID) and the category to which the application belongs (AppID); configure the quality detection strategies that should be provided for protection, including latency, bandwidth, packet loss rate, etc.

[0171] NWDAF: Configure the protection application (SubAppID) and its corresponding protection application category (AppID); configure the GFBR / MFBR information of the protection application.

[0172] PCF: Configure corresponding quality assurance schemes based on application category (AppID); NWDAF service discovery based on TA and event type.

[0173] In one embodiment, such as Figure 6 As shown, the specific implementation of the computing power service request processing method is provided, wherein:

[0174] Initial phase: When a user UE with a signed SLA goes online, the SMF obtains the session policy from the PCF and selects the smart UPF according to the instructions issued by the PCF.

[0175] The terminal-side AI mini-model collects user link / business layer data (including prediction data) from multiple dimensions such as communication data, user behavior, and personalized preferences obtained from the terminal.

[0176] For example, if computing power suddenly drops (due to changes in power consumption, battery level, location, movement speed, etc. / sudden increase or decrease in computing power business demand, etc.), the system can use its own AI algorithms to filter sensitive and malicious content, perform local data processing and reasoning such as personal privacy data protection encryption, semantic query, intent recognition, and tool calls, to achieve low-latency, lightweight intelligent decision-making, generate indicators such as the terminal's current situation, its own computing power and load, current link quality, and computing power business (single / continuous multiple requests), and report them to the edge gateway communication industry big model;

[0177] 1. The UE detected a sudden temperature rise.

[0178] 2. Proactively send emergency status updates.

[0179] 3. NWDAF will immediately notify all SMF subscribers.

[0180] 4. Ongoing tasks are being migrated to MEC.

[0181] Details of the terminal computing power indicator:

[0182] According to 3GPP TR 23.700-60 (6G Network Architecture Study), the computing power indicator reported by the terminal includes the following dimensions:

[0183] 1. Basic calculation ability:

[0184]

[0185] 2. Real-time status information:

[0186]

[0187] The large-scale model of the edge gateway communication industry is based on edge domain policies, performing basic calculations that match the SLA guarantee of the current target user, and realizing network orchestration and configuration execution.

[0188] Once the edge gateway communication industry big model detects that the KQI indicator matching the computing power service request does not meet the SLA guarantee requirements, it will generate a computing power service request indicator for the target user and continue to request massive computing power support from the cloud-side big model to achieve intelligent resource scheduling and precise scenario response.

[0189] After receiving the request indicator, the cloud-side big model subscribes to the SLA guarantee policy recommendation from NWDAF, carrying user identification information;

[0190] NWDAF's computing power database:

[0191]

[0192] NWDAF sends user session update suggestions to PCF based on subscription information, recommending that users migrate their data sessions to Smart UPF.

[0193] PCF sends an update message to SMF based on NWDAF's policy recommendations, and SMF disconnects the user's data session. When the user's data session reconnects, it selects Smart UPF for access.

[0194] NWDAF reports subscription experience data to the user's anchor point UPF, carrying user identification information; (in cases such as session release, subscription cancellation, or exceeding the supported location area, NWDAF cancels the SLA guarantee subscription to the UPF).

[0195] UPF reports the experience data of the specified user to NWDAF, along with real-time cell location information;

[0196] NWDAF predicts the network support for services based on the wireless cell load, the previous network quality, and the previous terminal service experience. It then determines whether to generate an SLA guarantee strategy based on the terminal's requirements and feeds back the initial guarantee strategy suggestion to the cloud model large-scale computing power management platform.

[0197] The cloud-based large-scale computing power management platform, based on the NWDAF's dedicated load strategy recommendations and combined with the user intent perceived on the edge, autonomously plans, remembers, acts, and reflects, with tool calls coordinating the entire process: by executing uplink / downlink CSI channel switching at the physical layer, performing bandwidth adaptation at the network layer based on network status information, and establishing QoS dedicated loads at the application layer, it executes business requirement decomposition, calculation, and coordination that matches the user's business characteristics; based on the large business model, it analyzes communication intent and issues corresponding first-phase SLA guarantee strategy recommendations to the PCF;

[0198] PCF makes further decisions based on the SLA guarantee strategy recommendations, policy conflict priorities, and location information from the cloud-based model large-scale computing power management platform. It then issues corresponding dynamic SLA guarantee rules to SMF, establishes / updates dedicated loads, and simultaneously sends a successful dedicated load establishment indication to NWDAF through the cloud-based model large-scale computing power management platform.

[0199] With the support of large-scale cloud computing power, users can obtain computing power services that meet SLA guarantees: low power consumption, high real-time processing computing power tasks, and obtain information with high confidence.

[0200] If the strategy in one phase still cannot meet the user's SLA guarantee requirements:

[0201] The terminal continues to report computing power service requirements. NWDAF predicts the network support service status based on the wireless cell load, the previous network quality, and the previous terminal service experience. It further determines whether to upgrade the protection strategy based on the terminal's requirements and feeds back the second-stage protection strategy recommendations to the cloud model large-scale computing power management platform.

[0202] Compared with the prior art, this application has the following technical advantages:

[0203] 1. The three-party model of edge cloud and device, and the multi-level computing power are coordinated throughout the entire process to ensure the fairness and transparency of AI decision-making, enhance its security and controllability, and avoid bias and unfairness in the decision-making process that may be caused by the deviation of a certain model training data or the defects of the algorithm.

[0204] 2. Cloud-based large-scale model ensures efficient and flexible SLA methods: By implementing uplink / downlink CSI channel switching at the physical layer, bandwidth adaptation at the network layer based on network status information, and establishing QoS dedicated load at the application layer, the system executes service requirement decomposition, calculation, and coordination that matches user service characteristics, thereby dynamically adjusting network resources and offloading tasks in real time.

[0205] 3. NWDAF distinguishes between a user's coverage application (SubAppID) and the coverage application category (AppID), thus differentiating between SLA-contracted users and contracted services and improving user experience.

[0206] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0207] Based on the same inventive concept, this application also provides a computing power service request processing apparatus for implementing the computing power service request processing method described above. The solution provided by this apparatus is similar to the implementation scheme described in the above method; therefore, the specific limitations in one or more embodiments of the computing power service request processing apparatus provided below can be found in the limitations of the computing power service request processing method described above, and will not be repeated here.

[0208] In one exemplary embodiment, such as Figure 7 As shown, a computing power service request processing device 700 is provided, applied in the cloud, including: a first receiving module 701, a first processing module 702, a second receiving module 703, and a second processing module 704, wherein:

[0209] The first receiving module 701 is used to receive the terminal computing power indicator uploaded by the terminal; the terminal computing power indicator is obtained by the terminal after the target user initiates a computing power service request; the terminal computing power indicator carries the user intent of the target user;

[0210] The first processing module 702 is used to forward the terminal computing power indicator to the edge side, and instruct the terminal to interact with the edge side to process computing power service requests based on the terminal computing power indicator, the terminal's geographical location information and the target user's SLA guarantee requirements.

[0211] The second receiving module 703 is used to receive the computing power service request indicator of the target user sent by the edge side. The computing power service request indicator is obtained by the edge side based on the current network status of the network to which the edge side belongs and the terminal computing power indicator, and is generated when the KQI indicator does not meet the SLA guarantee requirements.

[0212] The second processing module 704 is used to process computing power service requests in collaboration with the terminal and provide computing power support for cloud-based large models to the terminal based on user intent, terminal computing power indicator and computing power service request indicator.

[0213] In one embodiment, the first processing module 702 is further configured to determine the anchor UPF corresponding to the terminal based on the terminal's geographical location information and SLA guarantee requirements, under the instruction of the terminal computing power symbol, so as to instruct the terminal to process computing power service requests in interaction with the edge side based on the anchor UPF.

[0214] In one embodiment, the second processing module 704 is further configured to obtain the KQI guarantee required by the target user based on the computing power service request indicator; if the KQI guarantee representation requires anchor UPF replacement, determine a new anchor UPF from the candidate UPFs based on the real-time cell location information of the candidate UPFs, and, based on the new anchor UPF and the user intent, coordinate with the terminal to process the computing power service request and provide the terminal with computing power support for the cloud-based large model; if the KQI guarantee representation does not require anchor UPF replacement, use the anchor UPF as the new anchor UPF, subscribe to SLA guarantee policy recommendations from the NWDAF based on the terminal computing power indicator and the computing power service request indicator; and send a user session update recommendation to the PCF to establish an association between the PCF and the new anchor UPF.

[0215] In an exemplary embodiment, the second processing module 704 is further configured to report experience data of the subscription terminal to the new anchor UPF via NWDAF. The subscription information carries a user identifier, so that the new anchor UPF uploads the experience data of the target user corresponding to the user identifier to NWDAF. NWDAF predicts the network support service status based on the experience data, and generates an initial SLA guarantee strategy suggestion based on the predicted network support service status and the terminal's needs, and returns it to the cloud. It receives the initial SLA guarantee strategy suggestion, and plans a solution based on the initial SLA guarantee strategy suggestion and the user's intent to obtain the corresponding execution plan. It processes the computing power service request according to the execution plan.

[0216] In one embodiment, the second processing module 704 is further configured to perform uplink / downlink CSI channel switching at the physical layer, perform bandwidth adaptation at the network layer, and establish a QoS dedicated load policy at the application layer; perform service requirement decomposition, estimation, and coordination that matches the service characteristics of computing power service requests; generate SLA guarantee policy recommendations; send SLA guarantee policy recommendations and policy conflict priorities to the PCF; the PCF is configured to generate corresponding SLA guarantee rules based on the SLA guarantee policy recommendations, policy conflict priorities, and terminal location information, and send them to the SMF, so that the SMF can establish a dedicated load according to the SLA guarantee rules; and if the dedicated load recommendation is successful, send a dedicated load establishment success indication to the NWDAF via the cloud.

[0217] In one embodiment, the second processing module is further configured to receive an updated initial SLA guarantee strategy suggestion sent by the NWDAF when the SLA guarantee strategy suggestion cannot meet the SLA guarantee requirements; the updated initial SLA guarantee strategy suggestion is obtained by the NWDAF receiving new experience data of the target user reported by the new anchor UPF, predicting the network support service status based on the new experience data, obtaining a new predicted network support service status, updating the initial SLA guarantee strategy suggestion based on the new predicted network support service status and the terminal's needs; planning a scheme based on the updated initial SLA guarantee strategy suggestion and the user's intent to obtain a corresponding new execution scheme; and processing the computing power service request according to the new execution scheme.

[0218] In one exemplary embodiment, such as Figure 8 As shown, a computing power service request processing device 800 is provided, applied at the edge, including: a third receiving module 801, a first acquiring module 802, and a generating module 803, wherein:

[0219] The third receiving module 801 is used to receive the terminal computing power indicator forwarded by the cloud. The terminal computing power indicator is obtained by the terminal when the target user initiates a computing power service request and is sent to the cloud. The terminal computing power indicator carries the user intent of the target user.

[0220] The first acquisition module 802 is used to instruct the terminal to interact with the edge side to process computing power service requests based on the terminal computing power indicator, the terminal's geographical location information, and the target user's SLA guarantee requirements in the cloud, and then obtain the KQI indicator that matches the computing power service request based on the current network status of the network to which the edge side belongs and the terminal computing power indicator.

[0221] The generation module 803 is used to generate a computing power service request indicator for the target user and send it to the cloud when the KQI indicator does not meet the SLA guarantee requirements. The cloud is used to process the computing power service request in collaboration with the terminal and provide computing power support for the cloud big model to the terminal based on the user's intent and the terminal's computing power indicator and computing power service request indicator.

[0222] In one embodiment, the computing power service request processing device 800 is further configured to interact with the terminal to process computing power service requests when the KQI indicators meet the SLA guarantee requirements; the KQI indicators include at least one of the following: number of stutters, stutter duration, bit rate, resolution, response latency, setup latency, jitter, and bit error rate.

[0223] In one exemplary embodiment, such as Figure 9 As shown, a computing power service request processing device 900 is provided, applied to a terminal, including a second acquisition module 901, wherein:

[0224] The second acquisition module 901 is used to acquire a terminal computing power indicator and upload it to the cloud when the target user initiates a computing power service request. The terminal computing power indicator carries the user intent of the target user. The cloud is used to forward the terminal computing power indicator to the edge side and instruct the terminal to interact with the edge side to process the computing power service request based on the terminal computing power indicator, the geographical location information of the terminal, and the SLA guarantee requirements of the target user.

[0225] The edge side is used to obtain KQI indicators that match the computing power service request based on the current network status of the network to which the edge side belongs and the terminal computing power indicator; if the KQI indicators do not meet the SLA guarantee requirements, it generates the computing power service request indicator for the target user and sends it to the cloud.

[0226] The cloud is used to process computing power service requests in collaboration with the terminal based on user intent, terminal computing power indicator, and computing power service request indicator, and to provide computing power support for the cloud's large model to the terminal.

[0227] In one embodiment, the second acquisition module 901 is further configured to acquire the terminal's communication data, user behavior data, personalized preference data, terminal computing power information, load information, and link quality information, as well as the session scenario information in which the terminal is located; generate a terminal computing power indicator based on the session scenario information, terminal computing power information, load information, link quality information, and computing power service requests; and perform local processing on the communication data, user behavior data, and personalized preference data to obtain user intent; the local processing includes personal privacy data protection encryption, semantic query, and intent recognition.

[0228] In one embodiment, the second acquisition module 901 is further configured to filter malicious and sensitive content in communication data, user behavior data, and personalized preference data to obtain cleaned communication data, user behavior data, and personalized preference data; and to perform local data processing on the cleaned communication data, user behavior data, and personalized preference data to obtain user intent.

[0229] Each module in the aforementioned computing power service request processing device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0230] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 10As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The database stores terminal computing power indicators, terminal geographic location information, target user SLA guarantee requirements, and target user computing power service request indicators. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements a computing power service request processing method.

[0231] Those skilled in the art will understand that Figure 10 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0232] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the computing power service request processing method of various embodiments.

[0233] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the computing power service request processing method of various embodiments.

[0234] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the computing power service request processing method of various embodiments.

[0235] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0236] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0237] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0238] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A computing power service request processing system, characterized in that, The system includes a terminal, an edge device, and a cloud. The terminal is used to obtain a terminal computing power indicator and upload it to the cloud when a target user initiates a computing power service request; the terminal computing power indicator carries the user intent of the target user; The cloud is used to forward the terminal computing power indicator to the edge side, and instruct the terminal to interact with the edge side to process the computing power service request based on the terminal computing power indicator, the geographical location information of the terminal, and the SLA guarantee requirements of the target user. The edge side is used to obtain a KQI indicator that matches the computing power service request based on the current network status of the network to which the edge side belongs and the terminal computing power indicator; if the KQI indicator does not meet the SLA guarantee requirements, it generates a computing power service request indicator for the target user and sends it to the cloud. The cloud is used to process the computing power service request in collaboration with the terminal and provide computing power support for the cloud-based large model to the terminal based on the user intent, the terminal computing power indicator, and the computing power service request indicator.

2. The system according to claim 1, characterized in that, The system also includes: The cloud is also used to determine the anchor UPF corresponding to the terminal based on the terminal's geographical location information and the SLA guarantee requirements, under the instruction of the terminal computing power symbol, so as to instruct the terminal to interact with the edge side to process the computing power service request based on the anchor UPF.

3. The system according to claim 2, characterized in that, The system also includes: The cloud platform is also used to obtain the KQI guarantee required by the target user based on the computing power service request indicator. If the KQI guarantee indicator requires anchor UPF replacement, a new anchor UPF is determined from the candidate UPFs based on the real-time cell location information of the candidate UPFs. Based on the new anchor UPF and the user's intent, the cloud platform works with the terminal to process the computing power service request and provides the terminal with computing power support from the cloud-based large model. If the KQI guarantee indicator does not require anchor UPF replacement, the anchor UPF is used as the new anchor UPF, and based on the terminal computing power indicator and the computing power service request indicator, the cloud platform subscribes to the NWDAF for SLA guarantee policy recommendations. The cloud is also used to send user session update suggestions to the PCF so that the PCF can establish an association with the new anchor UPF.

4. The system according to claim 3, characterized in that, The system also includes: The cloud is further used to subscribe to the terminal's experience data reporting to the new anchor UPF through the NWDAF. The subscription information carries a user identifier, so that the new anchor UPF uploads the experience data of the target user corresponding to the user identifier to the NWDAF. The NWDAF predicts the network support service status based on the experience data, and generates an initial SLA guarantee strategy suggestion based on the predicted network support service status and the terminal's needs, and returns it to the cloud. The cloud is further configured to receive the initial SLA guarantee strategy suggestion, and plan a solution based on the initial SLA guarantee strategy suggestion and the user intent to obtain a corresponding execution plan; and process the computing power service request according to the execution plan.

5. The system according to claim 3, characterized in that, The system also includes: The cloud platform is further used to perform uplink / downlink CSI channel switching at the physical layer, bandwidth adaptation at the network layer, and establish QoS dedicated load policies at the application layer. It also performs service requirement decomposition, estimation, and coordination that matches the service characteristics of the computing power service request, generating SLA guarantee policy recommendations. The cloud platform sends the SLA guarantee policy recommendations and policy conflict priorities to the PCF. The PCF generates corresponding SLA guarantee rules based on the SLA guarantee policy recommendations, the policy conflict priorities, and the terminal's location information, and sends them to the SMF, enabling the SMF to establish a dedicated load according to the SLA guarantee rules. If the dedicated load recommendation is successful, the cloud platform sends a successful dedicated load establishment indication to the NWDAF.

6. The system according to claim 4, characterized in that, The system also includes: The cloud is further configured to, when the SLA guarantee strategy recommendation cannot meet the SLA guarantee requirements, receive an updated initial SLA guarantee strategy recommendation sent by the NWDAF, perform scheme planning based on the updated initial SLA guarantee strategy recommendation and the user intent, and obtain a corresponding new execution scheme; process the computing power service request according to the new execution scheme; the updated initial SLA guarantee strategy recommendation is obtained by the NWDAF receiving new experience data of the target user reported by the new anchored UPF, predicting the network support service status based on the new experience data, obtaining a new predicted network support service status, and updating the initial SLA guarantee strategy recommendation based on the new predicted network support service status and the terminal's needs.

7. The system according to claim 1, characterized in that, The system also includes: The edge side is also used to interact with the terminal to process the computing power service request when the KQI indicator meets the SLA guarantee requirements; the KQI indicator includes at least one of the following indicators: number of stutters, stutter duration, bit rate, resolution, response latency, setup latency, jitter, and bit error rate.

8. The system according to claim 1, characterized in that, The system also includes: The terminal is further configured to acquire the terminal's communication data, user behavior data, personalized preference data, terminal computing power information, load information, and link quality information, as well as the session scenario information in which the terminal is located; generate the terminal computing power indicator based on the session scenario information, the terminal computing power information, the load information, the link quality information, and the computing power service request; and perform local processing on the communication data, the user behavior data, and the personalized preference data to obtain the user intent; the local processing includes personal privacy data protection encryption, semantic query, and intent recognition.

9. The system according to claim 8, characterized in that, The system also includes: The terminal is further configured to filter malicious and sensitive content in the communication data, user behavior data, and personalized preference data to obtain cleaned communication data, user behavior data, and personalized preference data; and to perform local data processing on the cleaned communication data, user behavior data, and personalized preference data to obtain the user intent.

10. A method for processing computing power service requests, characterized in that, Applied to the cloud, the method includes: The terminal receives a terminal computing power indicator uploaded by the receiving terminal; the terminal computing power indicator is obtained by the terminal after the target user initiates a computing power service request; the terminal computing power indicator carries the user intent of the target user. The terminal computing power indicator is forwarded to the edge side, and based on the terminal computing power indicator, the geographical location information of the terminal, and the SLA guarantee requirements of the target user, the terminal is instructed to interact with the edge side to process the computing power service request. The edge side receives the computing power service request indicator of the target user sent by the edge side; the computing power service request indicator is obtained by the edge side based on the current network status of the network to which the edge side belongs and the terminal computing power indicator, and generates the indicator if the KQI indicator does not meet the SLA guarantee requirements. Based on the user intent, the terminal computing power indicator, and the computing power service request indicator, the system works in collaboration with the terminal to process the computing power service request and provide computing power support for the cloud-based large model to the terminal.

11. The method according to claim 10, characterized in that, The step of instructing the terminal to interact with the edge side to process the computing power service request based on the terminal computing power indicator, the terminal's geographical location information, and the target user's SLA guarantee requirements includes: Under the instruction of the terminal computing power symbol, the anchor UPF corresponding to the terminal is determined according to the geographical location information of the terminal and the SLA guarantee requirements, so as to instruct the terminal to process the computing power service request in interaction with the edge side based on the anchor UPF.

12. The method according to claim 11, characterized in that, The step of processing the computing power service request in coordination with the terminal based on the user intent, the terminal computing power indicator, and the computing power service request indicator includes: The KQI guarantee required by the target user is obtained based on the computing power service request indicator; When the KQI guarantee representation requires the anchor UPF to be replaced, a new anchor UPF is determined from the candidate UPFs based on the real-time cell location information of the candidate UPFs. Based on the new anchor UPF and the user intent, the computing power service request is processed in collaboration with the terminal, and computing power support for the cloud big model is provided to the terminal. If the KQI guarantee representation does not require anchor UPF replacement, the anchor UPF will be used as the new anchor UPF, and SLA guarantee policy recommendations will be subscribed to from NWDAF based on the terminal computing power indicator and the computing power service request indicator. Send a user session update suggestion to the PCF to establish an association between the PCF and the new anchored UPF.

13. The method according to claim 12, characterized in that, The step of processing the computing power service request in collaboration with the terminal based on the new anchored UPF and the user intent includes: The NWDAF subscribes to the new anchor UPF for the terminal's experience data reporting. The subscription information carries a user identifier, so that the new anchor UPF uploads the experience data of the target user corresponding to the user identifier to the NWDAF. The NWDAF predicts the network support service status based on the experience data, and generates an initial SLA guarantee strategy suggestion based on the predicted network support service status and the terminal's needs, and returns it to the cloud. Receive the initial SLA guarantee strategy suggestion, and plan the solution according to the initial SLA guarantee strategy suggestion and the user intent to obtain the corresponding execution plan; The computing power service request is processed according to the execution plan.

14. The method according to claim 12, characterized in that, The step of processing the computing power service request according to the execution plan includes: At the physical layer, uplink / downlink CSI channel switching is performed; at the network layer, bandwidth adaptation is performed; and at the application layer, QoS dedicated load policies are established. Service requirements are decomposed, calculated, and coordinated to match the service characteristics of the computing power service requests, and SLA guarantee policy recommendations are generated. The PCF sends the SLA guarantee policy suggestion and policy conflict priority to the PCF; the PCF generates corresponding SLA guarantee rules based on the SLA guarantee policy suggestion, the policy conflict priority and the location information of the terminal and sends them to the SMF, so that the SMF establishes a dedicated load according to the SLA guarantee rules. If the dedicated load suggestion is successful, the PCF sends a dedicated load establishment success indication to the NWDAF through the cloud.

15. The method according to claim 13, characterized in that, The method further includes: If the SLA guarantee policy recommendation cannot meet the SLA guarantee requirements, the NWDAF receives an updated initial SLA guarantee policy recommendation. The updated initial SLA guarantee policy recommendation is obtained by the NWDAF receiving new experience data of the target user reported by the new anchored UPF, predicting the network support service status based on the new experience data, obtaining a new predicted network support service status, and updating the initial SLA guarantee policy recommendation based on the new predicted network support service status and the terminal's needs. Based on the updated initial SLA guarantee strategy recommendations and the user intent, a new execution plan is developed to obtain the corresponding new execution plan. The computing power service request is processed according to the new execution plan.

16. A method for processing computing power service requests, characterized in that, Applied to the edge side, the method includes: Receive terminal computing power indicator forwarded by the cloud; the terminal computing power indicator is obtained by the terminal when the target user initiates a computing power service request and is sent to the cloud, and the terminal computing power indicator carries the user intent of the target user; After the cloud instructs the terminal to interact with the edge side to process the computing power service request based on the terminal computing power indicator, the terminal's geographical location information, and the target user's SLA guarantee requirements, it obtains the KQI indicator that matches the computing power service request based on the current network status of the network to which the edge side belongs and the terminal computing power indicator. If the KQI indicator does not meet the SLA guarantee requirements, a computing power service request indicator for the target user is generated and sent to the cloud. The cloud is used to process the computing power service request in collaboration with the terminal based on the user's intent, the terminal computing power indicator, and the computing power service request indicator, and to provide the terminal with computing power support for the cloud-based large model.

17. The method according to claim 16, characterized in that, The method further includes: If the KQI metric meets the SLA guarantee requirements, the computing power service request is processed in interaction with the terminal.

18. The method according to any one of claims 16-17, characterized in that, The KQI metrics include at least one of the following: number of stutters, stutter duration, bit rate, resolution, response latency, setup latency, jitter, and bit error rate.

19. A method for processing computing power service requests, characterized in that, Applied to a terminal, the method includes: When a target user initiates a computing power service request, a terminal computing power indicator is obtained and uploaded to the cloud; the terminal computing power indicator carries the user intent of the target user; the cloud is used to forward the terminal computing power indicator to the edge side, and instruct the terminal to interact with the edge side to process the computing power service request based on the terminal computing power indicator, the geographical location information of the terminal, and the SLA guarantee requirements of the target user. The edge side is used to obtain a KQI indicator that matches the computing power service request based on the current network status of the network to which the edge side belongs and the terminal computing power indicator; if the KQI indicator does not meet the SLA guarantee requirements, it generates a computing power service request indicator for the target user and sends it to the cloud. The cloud is used to process the computing power service request in collaboration with the terminal based on the user intent, the terminal computing power indicator, and the computing power service request indicator, and to provide the terminal with computing power support for the cloud-based large model.

20. The method according to claim 19, characterized in that, The acquisition of the terminal computing power indicator includes: The system acquires the terminal's communication data, user behavior data, personalized preference data, terminal computing power information, load information, and link quality information, as well as the session scenario information of the terminal. Based on the session scenario information, the terminal computing power information, the load information, the link quality information, and the computing power service request, the terminal computing power indicator is generated; The user intent is obtained through the following steps: The communication data, user behavior data, and personalized preference data are processed locally to obtain the user intent; the local processing includes encryption for personal privacy data protection, semantic query, and intent recognition.

21. The method according to claim 20, characterized in that, The step of locally processing the communication data, user behavior data, and personalized preference data to obtain the user intent includes: Malicious and sensitive content in the communication data, user behavior data, and personalized preference data is filtered to obtain cleaned communication data, user behavior data, and personalized preference data. The cleaned communication data, user behavior data, and personalized preference data are processed locally to obtain the user intent.

22. A computing power service request processing device, characterized in that, The device, applied in the cloud, includes: The first receiving module is used to receive a terminal computing power indicator uploaded by the terminal; the terminal computing power indicator is obtained by the terminal after the target user initiates a computing power service request; the terminal computing power indicator carries the user intent of the target user. The first processing module is used to forward the terminal computing power indicator to the edge side, and instruct the terminal to interact with the edge side to process the computing power service request based on the terminal computing power indicator, the geographical location information of the terminal, and the SLA guarantee requirements of the target user. The second receiving module is used to receive the computing power service request indicator of the target user sent by the edge side; the computing power service request indicator is obtained by the edge side based on the current network status of the network to which the edge side belongs and the terminal computing power indicator, and generates a KQI indicator that matches the computing power service request, and generates the indicator if the KQI indicator does not meet the SLA guarantee requirements. The second processing module is used to process the computing power service request in collaboration with the terminal and provide computing power support for the cloud-based large model to the terminal based on the user intent, the terminal computing power indicator and the computing power service request indicator.

23. A computing power service request processing device, characterized in that, Applied to the edge side, the device includes: The third receiving module is used to receive the terminal computing power indicator forwarded by the cloud; the terminal computing power indicator is obtained by the terminal when the target user initiates a computing power service request and is sent to the cloud, and the terminal computing power indicator carries the user intent of the target user. The first acquisition module is used to, in the cloud, instruct the terminal to interact with the edge side to process the computing power service request based on the terminal computing power indicator, the geographical location information of the terminal, and the SLA guarantee requirements of the target user, and then acquire the KQI indicator that matches the computing power service request based on the current network status of the network to which the edge side belongs and the terminal computing power indicator. The generation module is used to generate a computing power service request indicator for the target user and send it to the cloud when the KQI indicator does not meet the SLA guarantee requirements; the cloud is used to process the computing power service request in collaboration with the terminal and provide computing power support for the cloud-based large model to the terminal according to the user intent, the terminal computing power indicator, and the computing power service request indicator.

24. A computing power service request processing device, characterized in that, Applied to a terminal, the device includes: The second acquisition module is used to acquire a terminal computing power indicator and upload it to the cloud when a target user initiates a computing power service request; the terminal computing power indicator carries the user intent of the target user; the cloud is used to forward the terminal computing power indicator to the edge side, and instruct the terminal to interact with the edge side to process the computing power service request based on the terminal computing power indicator, the geographical location information of the terminal, and the SLA guarantee requirements of the target user. The edge side is used to obtain a KQI indicator that matches the computing power service request based on the current network status of the network to which the edge side belongs and the terminal computing power indicator; if the KQI indicator does not meet the SLA guarantee requirements, it generates a computing power service request indicator for the target user and sends it to the cloud. The cloud is used to process the computing power service request in collaboration with the terminal based on the user intent, the terminal computing power indicator, and the computing power service request indicator, and to provide the terminal with computing power support for the cloud-based large model.

25. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 10 to 21.

26. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 10 to 21.

27. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 10 to 21.

Images