A core network on-path compute service system and communication method

Abstract

The application provides a core network on-the-fly computing service system and a communication method, wherein the system comprises: a UPF entity; the UPF entity comprises an on-the-fly computing service module; the on-the-fly computing service module comprises a computing task offloading interface, a computing resource pool, a data cache and preprocessing unit and a service function interface; the forwarding component transmits a data packet to the data cache and preprocessing unit; the computing task offloading interface is used for dynamically scheduling computing resources to execute computing tasks; the data cache and preprocessing unit is used for caching and preprocessing the data packet, and transmitting the preprocessed data to the heterogeneous computing unit; the heterogeneous computing unit is used for computing the processed data, and transmitting the computing result to a destination address through the forwarding component. The system can reduce the bandwidth pressure and transmission delay of the satellite-ground link, and improve the real-time performance and reliability of data processing and services.

Description

Technical Field

[0001] This invention relates to the field of mobile communications, and more specifically to a core network-as-a-network computing service system and communication method. Background Technology

[0002] With the development of 6G communication technology, new service scenarios such as integrated sensing, computing, artificial intelligence (AI), and V2X will generate massive amounts of data. In scenarios such as satellite internet, non-terrestrial networks, and space-ground core network collaboration, the traditional core network architecture faces severe challenges. Currently, especially its User Plane Function (UPF), the primary design goal remains high-speed, low-latency packet forwarding. For example, data perceived by the UE or data generated by terminal computing tasks is uploaded to the core network without processing, and then transmitted to the data center via the bearer network for further computation and processing. The communication method has the following problems: 1) Huge network bandwidth pressure: In particular, in the scenario of space-ground collaboration, massive amounts of raw data (such as remote sensing images, video streams, sensor data, etc.) are transmitted to the ground through space links, occupying valuable space-ground communication spectrum resources and causing network congestion; 2) High processing latency and poor real-time performance: Data needs to go through a long path of "on-board -> space-ground link -> bearer network -> data center", and the end-to-end latency cannot meet the ultra-low latency requirements of services such as autonomous driving, remote control, and disaster early warning; 3) Reliability bottleneck: Long-distance, cross-atmosphere space-ground links are susceptible to interference, resulting in low data transmission reliability and affecting the continuity of critical services.

[0003] To reduce bandwidth pressure and transmission latency of links (such as satellite-to-ground links) and improve the real-time performance and reliability of data processing and services, it is particularly important to design a core network-as-a-network computing service system. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art and provide a core network in-network computing service system and communication method.

[0005] This invention provides an in-network computing service system for a core network, comprising: a UPF entity; the UPF entity includes a basic user plane function module and an in-network computing service module; the basic user plane function module includes a forwarding component and a control component; the in-network computing service module includes a computing task offloading interface, a computing resource pool, a data caching and preprocessing unit, and a service-oriented function interface; wherein, the computing resource pool includes heterogeneous computing units; the control component is used to receive policy rules, parse them, and transmit the computing task descriptors in the policy rules to the computing task offloading interface; the forwarding component is used to receive user data packets and transmit the data packets... The data is transmitted to the data cache and preprocessing unit; the computing task offloading interface is used to parse the computing task descriptor and dynamically schedule the computing resources of the heterogeneous computing unit to execute the computing task; the data cache and preprocessing unit is used to cache the data packet, preprocess it according to the computing task requirements, and transmit the preprocessed data to the heterogeneous computing unit; the heterogeneous computing unit is used to perform calculations on the data preprocessed by the data cache and preprocessing unit, and forward the calculation results to the destination address through the forwarding component; the service function interface is used to register, update, and deregister the computing service type of the UPF entity with the core network.

[0006] According to one embodiment of the present invention, the system further includes the core network control plane; the core network control plane includes NEF, PCF and SMF; the NEF is used to receive user computing service requests, perform authentication and authorization; the PCF is used to generate specific policy rules according to the computing service requests; and the SMF is used to distribute the policy rules to the control components.

[0007] According to one embodiment of the present invention, it further includes an application server; the application server is used to initiate a computing service request to the NEF.

[0008] According to one embodiment of the present invention, the computing resource pool is virtualized into multiple containers or lightweight virtual machines.

[0009] On the other hand, the present invention provides a communication method based on an on-the-go computing service system, using the aforementioned on-the-go computing service system for communication; comprising: the control component receiving policy rules, parsing them, and sending the computing task descriptors in the policy rules to the computing task unloading interface; the computing task unloading interface parsing the computing task descriptors and dynamically scheduling the corresponding computing resources of the heterogeneous computing unit; the forwarding component receiving user data packets and transmitting the data packets to the data caching and preprocessing unit; the data caching and preprocessing unit caching the data packets and preprocessing them according to task requirements; the data caching and preprocessing unit forwarding the preprocessed data to the heterogeneous computing unit, the heterogeneous computing unit executing computing tasks according to task requirements; and the heterogeneous computing unit forwarding the computing results to the destination address through the forwarding component.

[0010] According to one embodiment of the present invention, the in-network computing service system further includes a core network control plane and an application server; the core network control plane includes NEF, PCF, and SMF; before the forwarding component receives user data packets, the system further includes: the in-network computing service module registering its computing service type and available resources with the core network control plane through its service-oriented functional interface; the application server initiating a request to the core network control plane to enable the registered in-network computing service through the NEF API; the core network control plane authenticating and authorizing the application server, generating specific policy rules, and distributing the policy rules to the control component.

[0011] According to one embodiment of the present invention, the core network control plane authenticates and authorizes the application server, generates specific policy rules, and distributes the policy rules to the control component, including: the NEF receiving and verifying the application server's request, and authenticating and authorizing the application server; the PCF and NEF jointly generating specific policy rules; the SMF sending a request to the PCF to obtain policy rules related to the current session, and the PCF distributing the corresponding policy rules to the SMF; and the SMF distributing the policy rules to the control component through control plane signaling.

[0012] According to one embodiment of the present invention, the control component receives policy rules, parses them, and sends the computation task descriptor in the policy rules to the computation task unloading interface. The method further includes: the control component configuring forwarding control rules in the policy rules to the forwarding component; the forwarding component receives user data packets and transmits the data packets to the data cache and preprocessing unit, including: the forwarding component matching according to the forwarding control rules; if there is no match, the data packet is forwarded to the destination address according to the original rules; if there is a match, the data packet is transmitted to the data cache and preprocessing unit.

[0013] According to one embodiment of the present invention, after the computing task offloading interface parses the computing task descriptor and dynamically schedules the corresponding computing resources of the heterogeneous computing units, the following steps are included: the computing task offloading interface of the original satellite's UPF entity monitors its computing resource status in real time and reports the computing resource status to the core network control plane; the core network control plane detects whether the task is overloaded; if the task is not overloaded, the original satellite continues to execute the task; if the task is overloaded, the core network control plane selects another satellite to initiate a collaborative computing request; after receiving the request, the control component of the other satellite's UPF entity evaluates its own computing resources through its computing task offloading interface, and after confirming that its own resources are available, returns an agreement response to the core network control plane; the core network control plane distributes policy rules to the other satellite's UPF entity; the accompanying computing service module of the other satellite's UPF entity receives the data packet, and its computing task offloading interface schedules the heterogeneous computing units to execute the computing task; the forwarding component of the other satellite's UPF entity returns the computing result to the original satellite; the original satellite's UPF entity sends the computing result to the destination address.

[0014] According to one embodiment of the present invention, while the computing resource pool executes computing tasks according to task type, it reports the task execution status to the core network control plane; the core network control plane charges for the task execution status reported by the on-path computing service module and performs QoS control in coordination with PCF.

[0015] The in-network computing service system of the core network according to the present invention solves the problem that the existing UPF only has data forwarding function and cannot meet the needs of 6G new services for real-time, reliable and local processing of massive data by deeply integrating the in-network computing service module inside the traditional UPF.

[0016] It should be understood that the above general description and the following specific embodiments are merely exemplary and illustrative, and do not limit the scope of the invention. Attached Figure Description

[0017] The accompanying drawings, which are part of the specification of this invention, illustrate exemplary embodiments of the invention. The drawings, together with the description in the specification, serve to illustrate the principles of the invention.

[0018] Figure 1 This is a schematic diagram of the overall architecture of the core network's on-the-go computing service system according to an embodiment of the present invention;

[0019] Figure 2 This is a flowchart illustrating the data packet processing method of a communication method based on an in-path computing service system according to an embodiment of the present invention. Detailed Implementation

[0020] The features and exemplary embodiments of various aspects of the present invention will now be described in detail. To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only configured to explain the present invention and to exemplify the principles of the present invention, and are not configured to limit the present invention. In addition, the structural components in the drawings are not necessarily drawn to scale. For example, the dimensions of some structural components or regions in the drawings may be enlarged for other structural components or regions to aid in the understanding of the embodiments of the present invention.

[0021] The directional terms used in the following description refer to the directions shown in the figures and are not intended to limit the specific structure of the embodiments of the present invention. In the description of the present invention, it should be noted that, unless otherwise stated, the terms "installation," "connection," and "joining" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms in the present invention according to the specific circumstances.

[0022] Furthermore, the terms "comprising," "including," "having," or any other variations thereof are intended to cover non-exclusive inclusion, such that a structure or component that includes a list of elements includes not only those elements but also other structural elements that are not expressly listed or inherent to the structure or component. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of other identical elements in the article or apparatus that includes the element.

[0023] Spatial relation terms such as "below," "under," "under," "low," "above," "on," and "high" are used for descriptive convenience to explain the positioning of one element relative to a second element, indicating that these terms are intended to cover different orientations of the device, in addition to those different from those shown in the figure. Furthermore, phrases such as "one element on / below another element" can indicate that two elements are in direct contact, or that there are other elements between the two elements. In addition, terms such as "first" and "second" are also used to describe individual elements, areas, parts, etc., without specifically indicating order or sequence, and should not be considered restrictive. Similar terms are used throughout the description to represent similar elements.

[0024] It will be apparent to those skilled in the art that the present invention can be practiced without requiring some of these specific details. The following description of embodiments is merely intended to provide a better understanding of the invention by illustrating examples of the invention.

[0025] Terminology Explanation:

[0026] UPF: User-Face Functionality;

[0027] NEF: Network Open Functionality;

[0028] PCF: Policy Control Function;

[0029] SMF: Session Management Function;

[0030] AMF: Access and Mobility Management Functions;

[0031] NRF: Network Functions Repository;

[0032] QoS: Quality of Service;

[0033] PDU: Protocol Data Unit Session;

[0034] PCC Rules: Policy and Charging Control;

[0035] PDR: Packet Detection Rule.

[0036] Figure 1 This is a schematic diagram of the overall architecture of the core network's on-the-go computing service system according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating the data packet processing method of a communication method based on an in-path computing service system according to an embodiment of the present invention.

[0037] like Figure 1As shown, this invention provides an in-network computing service system for a core network, comprising: a UPF entity. The UPF entity includes a basic user plane function module and an in-network computing service module. The basic user plane function module includes a forwarding component and a control component. The in-network computing service module includes a computing task offloading interface, a computing resource pool, a data caching and preprocessing unit, and a service-oriented function interface. The computing resource pool includes heterogeneous computing units. The control component receives policy rules, parses them, and transmits the computing task descriptors in the policy rules to the computing task offloading interface. The forwarding component receives user data packets and transmits them to the data caching and preprocessing unit. The computing task offloading interface parses the computing task descriptors and dynamically schedules the computing resources of the heterogeneous computing units to execute computing tasks. The data caching and preprocessing unit caches data packets, preprocesses them according to computing task requirements, and transmits the preprocessed data to the heterogeneous computing units. The heterogeneous computing units perform calculations on the data preprocessed by the data caching and preprocessing unit and forward the calculation results to the destination address through the forwarding component. The service-oriented function interface registers, updates, and deregisters the computing service type of the UPF entity with the core network control plane.

[0038] Specifically, in the standard 3GPP 5G core network architecture, the UPF is responsible for packet routing and forwarding, policy enforcement, etc., but does not have general computing service capabilities. Although the standard 3GPP 5G core network architecture has mobile edge computing (MEC) technology, it still acts as a data pipeline and has not been adapted for special constraints (such as high dynamism, high latency, intermittent connectivity, and resource constraints).

[0039] The in-network computing service system provided in this embodiment deeply integrates a general-purpose, service-oriented in-network computing service module within the traditional UPF (User-Defined Frame). This deeply integrates intelligence with network communication, breaking away from the traditional UPF design which is merely a forwarding plane. It creates a new type of network entity integrating sensing, communication, and computing, achieving architectural innovation. This system can process data locally through the UPF entity, only transmitting valuable results or simplified data back to the user, significantly reducing transmission bandwidth and latency. Experimental verification shows that this in-network computing service system can reduce link load by more than 70%, and end-to-end latency can be reduced by more than 50%.

[0040] This system reduces reliance on unstable links (such as satellite-to-ground links) for data transmission, allowing critical business processes (such as the perception-decision-control closed loop) to be completed closer to the user, significantly improving service real-time performance and reliability, and saving on expensive satellite-to-ground bandwidth leasing costs, thus reducing operating costs. The system efficiently utilizes network resources, achieving deep integration of communication, perception, and computing resources, enabling service providers to offer integrated "connectivity + computing" services and providing value-added services to vertical industries (such as agriculture, environmental protection, and transportation). Furthermore, the system provides feasible network architecture support for applications in the 6G era that are extremely sensitive to latency and bandwidth (global AR / VR, space internet, and global collaboration of unmanned systems).

[0041] For example, heterogeneous computing units can include hardware computing units such as CPUs, GPUs, NPUs, and FPGAs. Data caching and preprocessing units can perform preprocessing on data packets, such as data extraction and format conversion. In-band computing service modules can interact with the core network control plane (e.g., static configuration or NRF service discovery) through their service-oriented function interfaces to register the types of computing services they can provide (e.g., image recognition, data aggregation, model inference) and available resources. Registration information can include UPF identifiers, DNNs, slices, location information, and computing service information.

[0042] Computing services can include on-the-go data processing and networking services. For example, providing video transcoding services, data encryption / decryption services, or serving as a light node in a blockchain directly on satellite.

[0043] like Figure 1 As shown, according to one embodiment of the present invention, in addition to the UPF entity, the system also includes a core network control plane. The core network control plane includes control plane network elements such as NEF, PCF, and SMF. The NEF is used to receive user computing service requests and perform authentication and authorization. The PCF is used to generate specific policy rules based on the computing service requests. The SMF is used to distribute the policy rules to the control components.

[0044] In this embodiment, for example, the service-oriented function interface can register, update, and deregister the computing service capabilities of the UPF entity with the core network statically or via NRF. Specific policy rules can instruct the UPF entity's in-band computing service module to provide computing services for specific data flows. For example, the SMF can distribute policy rules to the control component of the UPF entity via the N4 interface. The control component receives the policy rules, parses them, and configures the forwarding control rules in the policy rules into the forwarding component of the UPF entity. The forwarding component guides the data flow to the in-band computing service module for processing according to the forwarding control rules in the policy rules, and the processing result is then forwarded by the forwarding component to the destination address.

[0045] This system integrates an in-network computing service module within the UPF entity and coordinates it with core network control plane elements (NEF, PCF, SMF). The computing capabilities of this in-network computing service module are then securely and controllably opened to third-party applications as a policy-controllable service.

[0046] According to one embodiment of the present invention, the core network of the in-network computing service system further includes a management plane (OAM) and an intelligent plane. The management plane (OAM) is responsible for the daily operation and maintenance and fault handling of the in-network computing service system. The intelligent plane generates optimization suggestions through AI models and outputs these suggestions to network elements such as the PCF to optimize resource scheduling and enhance fault handling. The core network control plane also includes control plane network elements such as the AMF, used for user terminal access authentication and mobility management. The core network adopts an SBA (Service-Based Architecture) to achieve modularization of network element functions.

[0047] According to one embodiment of the present invention, in addition to the UPF entity and the core network control plane, the system also includes an application server. The application server is used to initiate computing service requests to the NEF.

[0048] In this embodiment, the application server (e.g., a third-party application server) can initiate a computing service request to the NEF of the core network control plane through a standard API (such as CAPIF or 3GPP NEF API) to request the activation of a specific computing service (such as on-board) for a specific user or data flow. For example, operators can manually submit computing tasks through the application server interface, or the application server can automatically trigger the computing request.

[0049] According to one embodiment of the present invention, the computing resource pool is virtualized into multiple containers or lightweight virtual machines.

[0050] In this embodiment, the computing resource pool is virtualized into multiple containers or lightweight virtual machines, which can provide isolated computing environments for different tenants or applications. The virtualization technology mainly provides collaborative management and scheduling of computing resources for the core network and provides computing resource isolation for dedicated slices.

[0051] like Figure 2 As shown, on the other hand, the present invention provides a communication method based on an on-the-go computing service system, using the aforementioned on-the-go computing service system for communication; including:

[0052] S010: The control component receives the policy rules, parses them, and sends the computation task descriptor in the policy rules to the computation task unloading interface.

[0053] S020: The computing task offloading interface parses the computing task descriptor and dynamically schedules the corresponding computing resources of heterogeneous computing units;

[0054] S030: The forwarding component receives user data packets and transmits the data packets to the data buffer and preprocessing unit;

[0055] S040: The data caching and preprocessing unit caches data packets and preprocesses them according to task requirements;

[0056] S050: The data caching and preprocessing unit forwards the preprocessed data to the heterogeneous computing unit, which then executes the computing tasks according to the task requirements.

[0057] S060: The heterogeneous computing unit forwards the computation results to the destination address through the forwarding component.

[0058] In this embodiment, for example, the computing task can be initiated by the application server, or by the user terminal (UE) through a specific marker or signaling in the uplink data packet. The in-path computing service module identifies the marker and triggers the corresponding computing. The destination address can be another terminal, the application server, or the simplified result data can be sent to the destination (such as the ground).

[0059] According to one embodiment of the present invention, the on-the-go computing service system further includes a core network control plane and an application server. The core network control plane includes NEF, PCF, and SMF;

[0060] Before the forwarding component receives user data packets, it also includes:

[0061] S001 (Capability Registration): The in-line computing service module registers its computing service type and available resources with the core network control plane through its service-oriented function interface;

[0062] S002 (Service Request): The application server initiates a request to the core network control plane via the NEF API to enable the registered on-board computing service.

[0063] S003 (Policy Generation and Distribution): The core network controller authenticates and authorizes the application server, generates specific policy rules, and distributes the policy rules to the control components.

[0064] In this embodiment, for example, when the UPF entity starts, its on-the-fly computing service module automatically registers with the NEF on the core network control plane, reporting the types of services it can provide (e.g., image recognition) and its currently available computing resources. The application server initiates a request to the core network control plane to enable specific computing (e.g., on-board) services for a data stream of a certain terminal, a certain DNN, or a certain slice. This request may include: the terminal identifier requesting the service (e.g., SUPI or IP address), a description of the data stream for which computing needs to be performed (e.g., source / destination IP, port, protocol), the type of computing service requested (e.g., real-time target detection, "data aggregation"), specific service quality parameters (e.g., recognition model ID, aggregation frequency, maximum latency requirement), and the destination address of the computing results (e.g., the server IP of a disaster monitoring center). The core network authenticates and authorizes the application server, generates specific policy rules, and distributes the policy rules to the control component during the session establishment or modification signaling process.

[0065] This communication method uses the Core Network Capability Open Framework (NEF) to uniformly abstract, authenticate, authorize, and control computing (such as on-board) capabilities, enabling third-party applications (e.g., ground-based disaster monitoring centers) to access computing resources (such as on-board computing resources) on demand and securely, just like using network bandwidth.

[0066] According to an embodiment of the present invention, step S003: the core network controller performs authentication and authorization on the application server, generates specific policy rules, and distributes the policy rules to the control component, including:

[0067] S003.1: NEF receives and verifies the application server's request, and authenticates and authorizes the application server;

[0068] S003.2: PCF and NEF work together to generate specific policy rules;

[0069] S003.3: The SMF sends a request to the PCF to obtain the policy rules related to the current session, and the PCF distributes the corresponding policy rules to the SMF; the SMF distributes the policy rules to the control component through control plane signaling.

[0070] In this embodiment, after receiving a request from the application server, the NEF verifies whether the application server has permission to use this service. After authenticating the application's identity, the NEF initiates a Policy Authorization Request to the PCF. This request translates and maps the business requirements in the ComputeServiceRequest into network-understandable policy rule information. Based on the authorization information and network policy rules received from the NEF, the PCF generates a dedicated Compute-Aware PCC Rule (aware computing policy and billing rule). The rule information mainly includes: a unique identifier for this computing service session, the service type and parameters authorized after NEF authentication, and service quality requirements based on the subscription and request (such as latency and reliability).

[0071] After generating policy rules, the PCF associates them with the SMF and sends the association process to the SMF. The association between the PCF and the SMF includes: associating the service interfaces (such as NPCF) between the PCF and the SMF, and associating the PCC rules between the SMF and the PCF.

[0072] After receiving the policy rules sent by the PCF, the SMF interacts with the UPF entity (generation and distribution of policy rules). The interaction between the SMF and the UPF includes: interaction between the SMF and the UPF via the N4 interface; the SMF translates and encapsulates the Compute-Aware PCCRule distributed by the PCF into an Enhanced PDR / FAR (Enhanced Packet Inspection Rule / Forwarding Action Rule) executable by the UPF entity, and distributes it to the control components of the UPF entity.

[0073] The forwarding component, based on the enhanced PDR / FAR issued by SMF, guides the matched data packets to the in-band computing service module via the Compute TaskTrigger internal signaling. After the in-band computing service module completes the computation, it encapsulates the result data into a new data packet and sends it back to the forwarding component via the Compute Result Notification internal signaling. The forwarding component then forwards the data according to the calculated FAR.

[0074] According to an embodiment of the present invention, step S010, the control component receives the policy rule, parses it, and sends the computation task descriptor in the policy rule to the computation task unloading interface, and further includes:

[0075] S011: The control component configures the forwarding control rules in the policy rules to the forwarding component;

[0076] Step S030: The forwarding component receives the user data packet and transmits it to the data buffer and preprocessing unit, including:

[0077] The S030.1 forwarding component matches the data packet according to the forwarding control rules. If there is no match, the data packet is forwarded directly to the destination address according to the original rules (i.e., the traditional rules). If there is a match, the data packet is transmitted to the data buffer and preprocessing unit.

[0078] In this embodiment, the policy rules may include data flow filters and associated computing tasks. Data flow filters define matching rules for data packets. The forwarding component transmits data packets matching the data flow filters to the data caching and preprocessing unit of the in-network computing service module for preprocessing. The computing resource pool executes associated computing tasks on the data flow, for example, performing calculations on the computing flow according to different computing tasks. This communication method, based on the "computation diversion" mechanism of the data flow policy, enables user plane data to be dynamically and flexibly diverted to the in-network computing service module for processing, realizing a "network-computing integration" business process.

[0079] According to an embodiment of the present invention, after the computing task offloading interface parses the computing task descriptor and dynamically schedules the corresponding computing resources of the heterogeneous computing units in step S020, it includes:

[0080] S021: The original satellite's computing task offloading interface monitors its computing resource status in real time and reports the computing resource status to the core network control plane; the core network control plane detects whether the task is overloaded; if the task is not overloaded, the original satellite continues to execute the task; if the task is overloaded, steps S022-S026 are executed.

[0081] S022: The core network control plane selects another satellite (e.g., a neighboring satellite) to initiate a collaborative computing request;

[0082] S023: After receiving the request, the control component of another satellite evaluates its own computing resources through its computing task offloading interface. After confirming that its own resources are available, it returns an agreement response to the core network control plane.

[0083] S024: The core network control plane distributes policy rules to the UPF entity of another satellite;

[0084] S025: The UPF entity of another satellite receives data packets, and its computing task offloading interface schedules heterogeneous computing units to perform computing tasks.

[0085] S026: The relay component of another satellite returns the calculation results to the original satellite; the UPF entity of the original satellite sends the calculation results to the destination address.

[0086] In this embodiment, for example, the computational task offloading interface of the original satellite's UPF entity monitors its computational resource status in real time and reports the computational resource status to the core network control plane through the N4 interface (the control plane interface between the SMF and UPF). The core network control plane detects whether the task is overloaded (e.g., detects the load of the UPF entity's computing resources). If the task is not overloaded, the original satellite's UPF entity continues to execute the task. If the task is overloaded, the core network control plane, through unified collaborative scheduling and management of UPF entities (e.g., based on changes in the location area of ​​the UPF entity and the load status of the UPF entity), selects neighboring satellites and initiates collaborative computing requests to their UPF entities (e.g., the selection of UPF entities for collaborative computing follows the I-UPF or UPF (PSA) role definition). After receiving the request, the control component of the neighboring satellite's UPF entity evaluates its own computing resources through its computing task offloading interface. After confirming that its resources are available, subsequent user data packets are sent to the in-band computing service module of the neighboring satellite's UPF entity. The in-band computing service module of the neighboring satellite's UPF entity receives the data packets, and its computing task offloading interface schedules heterogeneous computing units to execute computing tasks. The forwarding component of the neighboring satellite's UPF entity returns the computing results to the original satellite's UPF entity, which then sends the computing results to the destination address.

[0087] In a satellite constellation scenario with multiple satellites equipped with the aforementioned core network and on-the-path computing service system, when the computing tasks of the on-the-path computing service module of the current satellite are overloaded, the task can be offloaded to the on-the-path computing service module of a neighboring satellite via inter-satellite links, forming inter-satellite computing collaboration. This communication method improves the computing service capability of the entire satellite constellation through task collaboration and load balancing among multiple onboard on-the-path computing service modules.

[0088] According to one embodiment of the present invention, while the computing resource pool executes computing tasks according to task type, it reports the task execution status to the core network control plane; the core network control plane performs billing based on the task execution status reported by the in-path computing service module, and performs QoS control in coordination with the PCF.

[0089] The communication method provided in this embodiment adds billing and quality of service assurance. The core network control plane can charge according to the task execution status (e.g., computing time, resource consumption) reported by the on-path computing service module, and cooperate with PCF to ensure the QoS of computing services (e.g., maximum processing latency).

[0090] Taking real-time satellite remote sensing image recognition as an example, a ground-based disaster monitoring center (as a third-party application) needs to monitor ships in a certain sea area in real time. The remote sensing satellite carrying the on-the-go computing service system of the core network of this invention enters its working orbit, and the communication method of this invention is as follows:

[0091] S1: System initialization (capability registration).

[0092] When the UPF entity starts up, its on-path computing service module automatically initiates a registration service capability with the core network, reporting that it can provide target detection services, real-time target detection services based on target detection models such as YOLO, and reporting its current available GPU computing power as 10 TFLOPS and other computing power indicators.

[0093] S2: Application server and NEF associated service request (service request).

[0094] The application server of the disaster monitoring center sends service request messages to the core network through the standard NEF API. This request may include:

[0095] UE_Identifier: The identifier of the terminal requesting its service (such as a mobile phone number or IP address);

[0096] DataFlowDescription: A description of the data flow for which computation needs to be performed (e.g., source / destination IP, port, protocol);

[0097] ComputeServiceType: The type of computing service requested (such as real-time target detection, "data aggregation", etc.);

[0098] ServiceParameters: Specific service quality parameters (such as identification model ID, aggregation frequency, maximum latency requirement);

[0099] ResultDestination: The destination address of the calculation results (such as the server IP of a disaster monitoring center).

[0100] S3: NEF is associated with PCF to generate policy rules (i.e., policy rules are generated according to user needs).

[0101] Upon receiving the request, NEF verifies whether the disaster monitoring center has the authority to use this service. After authenticating the application's identity, NEF sends a PolicyAuthorizationRequest to PCF. This request translates and maps the business requirements in the ComputeServiceRequest into network-understandable policy rule information. Based on the authorization information received from NEF, user subscription data, and network policy rules, PCF generates a dedicated Compute-AwarePCC Rule (aware computing policy and billing rule). The main content of the rule information includes:

[0102] ServiceIdentifier: A unique identifier for this computing service session;

[0103] AuthorizedComputeService: The service type and parameters authorized after NEF authentication;

[0104] SubscribedQoS: Service quality requirements (such as latency and reliability) based on subscription and request.

[0105] S4: PCF and SMF association (generation and distribution of policy rules), including:

[0106] Association Interface: Associating service interfaces (such as NPCF) between PCF and SMF;

[0107] Core interaction content: In the PDU session flow, SMF and PCF associate PCC rules, such as SM PolicyAssociation Establishment / Modification, and PCF sends the association process to SMF.

[0108] This PCC rule is an extension of the existing PCC rule, and the core new content includes:

[0109] Compute Action Trigger: In addition to the traditional "data stream filter", a ComputeAction field is added. This field indicates what computation task the UPF performs on the matched data stream (e.g., TaskID: "Ship_Detection").

[0110] Result forwarding rules: In traditional "forwarding behavior", the next-hop address of the calculation result is explicitly specified (which may be different from the destination of the original data stream).

[0111] S5: SMF and UPF entity interaction (generation and distribution of policy rules), including:

[0112] Related interface: The SMF and UPF interact via the N4 interface;

[0113] Core interaction content: SMF translates and encapsulates the Compute-Aware PCC Rule issued by PCF into Enhanced PDR / FAR (Enhanced Packet Inspection Rule / Forwarding Action Rule) executable by UPF entities and issues it.

[0114] Enhanced PDR: In addition to matching packet characteristics, it also includes a Compute Indication flag; when a packet matches this PDR, the UPF entity not only performs forwarding, but also triggers the in-path computation service module;

[0115] Enhanced FAR: Its forwarding parameters not only include the destination IP, but may also include Pre-Compute FAR and Post-Compute FAR. For example, the pre-compute action is to copy the data packet to the in-band compute service module, and the post-compute action is to forward the result packet returned by the in-band compute service module to the Result Destination.

[0116] In this step, a combined policy of PDR (Packet Detection Rule) and FAR (Forwarding Action Rule) is generated in the session flow. This policy is issued to the UPF entity via the N4 interface by the Session Management Function (SMF). The policy content is: "Direct the image data stream from the satellite terminal ID, with the destination IP being the original ground station IP, to the UPF entity's in-path computing service module for processing; the computing task ID is vessel detection; after processing, the result is forwarded to the disaster monitoring center IP."

[0117] S6: SMF and UPF entity interaction.

[0118] The relationship between the control components, forwarding components, and in-band computing service modules within a UPF entity:

[0119] Associated interface: internal bus or high-speed interconnect interface;

[0120] Core interaction: Based on the enhanced PDR / FAR issued by SMF, the forwarding component guides the matched data packet to the in-band computing service module via the Compute Task Trigger internal signaling; after the in-band computing service module completes the calculation, it encapsulates the result data into a new data packet and sends it back to the forwarding component via the Compute Result Notification internal signaling, and the forwarding component forwards it according to the calculated FAR.

[0121] S7: In-band computing execution (association and collaboration between the forwarding component and the in-band computing service module).

[0122] The sea surface image data captured by the user terminal's sensors is encapsulated into data packets and flows through the UPF entity. The forwarding component of the UPF entity checks each data packet to see if its source and destination addresses match the policy issued in step S3. For data packets that match the policy issued in step S3, the forwarding component or hardware mirror copies these image data packets and directs them to the computing task offloading interface of the in-path computing service module. The data caching and preprocessing unit of the in-path computing service module reassembles the data packets into complete image frames and transmits the complete image frames to the computing resource pool. The GPU / NPU in the computing resource pool loads a pre-trained YOLO model, performs inference on the image frames, identifies the ships in them, and generates structured data (JSON format) containing ship location and category information. The computing resource pool of the in-path computing service module sends the result data (only a few KB in size) back to the forwarding component. The forwarding component forwards the result data to the IP address of the disaster monitoring center according to FAR.

[0123] Traditional systems require the transmission of tens of gigabytes or even hundreds of terabytes of raw images. Using the on-the-go computing service system of the core network of this invention, the raw images can be processed, producing data in the form of only a few kilobytes. This reduces satellite-to-ground bandwidth consumption and transmission latency to extremely low levels, enabling disaster monitoring centers to receive monitoring results almost in real time.

[0124] The above embodiments of the present invention can be combined with each other and have corresponding technical effects.

[0125] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A core network-as-a-network computing service system, characterized in that, include: UPF entity and core network control plane; The core network control plane includes NEF, PCF, and SMF; The NEF is used to receive user computing service requests, perform authentication and authorization; the PCF is used to generate specific policy rules based on the computing service requests; the policy rules include data flow filters and associated computing tasks; the data flow filters are used to define matching rules for data packets; the SMF is used to distribute the policy rules to the control component; the UPF entity includes a basic user plane function module and an in-path computing service module. The basic user plane functional modules include forwarding components and control components; The on-the-go computing service module includes a computing task offloading interface, a computing resource pool, a data caching and preprocessing unit, and a service-oriented function interface; wherein, the computing resource pool includes heterogeneous computing units; The control component is used to receive policy rules, parse them, and transmit the computation task descriptor in the policy rules to the computation task unloading interface; the forwarding component is used to receive user data packets and transmit data packets that match the data stream filter to the data cache and preprocessing unit. The computing task offloading interface is used to parse computing task descriptors and dynamically schedule the computing resources of the heterogeneous computing units to execute associated computing tasks. The data caching and preprocessing unit is used to cache data packets, preprocess them according to the computing task requirements, and transmit the preprocessed data to the heterogeneous computing unit. The heterogeneous computing unit is used to perform calculations on the data preprocessed by the data cache and preprocessing unit, and forward the calculation results to the destination address through the forwarding component; The service-oriented interface is used to register, update, and deregister the computing service type of the UPF entity with the core network control plane.

2. The system according to claim 1, characterized in that, It also includes an application server; the application server is used to initiate computing service requests to the NEF.

3. The system according to claim 1, characterized in that, The computing resource pool is virtualized into multiple containers or lightweight virtual machines.

4. A communication method based on an on-the-go computing service system, characterized in that, Communication is performed using the in-network computing service system as described in claim 1; including: The control component receives the policy rules, parses them, and sends the computation task descriptor in the policy rules to the computation task unloading interface. The computing task offloading interface parses the computing task descriptor and dynamically schedules the corresponding computing resources of the heterogeneous computing unit. The forwarding component receives user data packets and transmits the data packets to the data buffer and preprocessing unit; The data caching and preprocessing unit caches data packets and preprocesses them according to task requirements; The data caching and preprocessing unit forwards the preprocessed data to the heterogeneous computing unit, and the heterogeneous computing unit executes computing tasks according to task requirements. The heterogeneous computing unit forwards the calculation results to the destination address through the forwarding component.

5. The communication method according to claim 4, characterized in that, The on-the-go computing service system also includes a core network control plane and an application server; the core network control plane includes NEF, PCF, and SMF. Before the forwarding component receives user data packets, it also includes: The in-line computing service module registers its computing service type and available resources with the core network control plane through its service-oriented function interface; The application server initiates a request to the core network via the NEF API to enable the registered on-board computing service. The core network authenticates and authorizes the application server, generates specific policy rules, and distributes the policy rules to the control component.

6. The communication method according to claim 5, characterized in that, The core network authenticates and authorizes the application server, generates specific policy rules, and distributes the policy rules to the control component, including: The NEF receives and verifies the application server's request, and authenticates and authorizes the application server. The PCF and NEF work together to generate specific policy rules; The SMF sends a request to the PCF to obtain the policy rules related to the current session, and the PCF sends the corresponding policy rules to the SMF. The SMF distributes policy rules to the control component via control plane signaling.

7. The communication method according to claim 6, characterized in that, The control component receives policy rules, parses them, and sends the computation task descriptor in the policy rules to the computation task unloading interface. It also includes: The control component configures the forwarding control rules in the policy rules to the forwarding component; The forwarding component receives user data packets and transmits the data packets to the data buffer and preprocessing unit, including: The forwarding component matches the data packet according to the forwarding control rules. If there is no match, the data packet is forwarded to the destination address according to the original rules. If there is a match, the data packet is transmitted to the data cache and preprocessing unit.

8. The communication method according to claim 4, characterized in that, The computation task offloading interface parses the computation task descriptor and dynamically schedules the corresponding computational resources of the heterogeneous computing unit, including: The computing task offloading interface of the original satellite's UPF entity monitors its computing resource status in real time and reports the computing resource status to the core network control plane; the core network control plane detects whether the task is overloaded; if the task is not overloaded, the original satellite continues to execute the task; if the task is overloaded, the core network control plane selects a neighboring satellite to initiate a collaborative computing request. After receiving the request, the control component of the UPF entity of the neighboring satellite evaluates its own computing resources through its computing task offloading interface. After confirming that its own resources are available, it returns an agreement response to the core network control plane. The core network control plane distributes policy rules to the UPF entities of neighboring satellites; The UPF entity of the neighboring satellite receives the data packet, and its computing task offloading interface schedules the heterogeneous computing unit to perform computing tasks. The forwarding component of the UPF entity of the neighboring satellite returns the calculation result to the original satellite; the UPF entity of the original satellite sends the calculation result to the destination address.

9. The communication method according to claim 6, characterized in that, While executing computing tasks according to task type, the computing resource pool reports the task execution status to the core network control plane; The core network control plane performs billing based on the task execution status reported by the in-path computing service module and performs QoS control in coordination with the PCF.

Images