Systems and methods for implementing network functionality

Abstract

This invention provides a system and method for placing network functions, including a non-real-time radio access network (NRWAN) intelligent controller. The NRWAN intelligent controller comprises a graph model, a ground shadow prediction application, and a network function placement application. The ground shadow prediction application determines a ground shadow prediction result based on the latest ephemeris information and sends the result to the graph model. The graph model receives performance measurement data. The graph model constructs a time-varying graph based on the ground shadow prediction result, the latest ephemeris information, and the performance measurement data, and sends the time-varying graph to the network function placement application. The placement application determines the placement information for the network function based on the time-varying graph. This invention solves the problem that network function placement methods in related technologies are difficult to apply to satellite network environments.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of cloud-network convergence, and more specifically, to a system and method for placing network functions. Background Technology

[0002] Currently, wireless networks rely on cloud-native architectures for construction, enabling more efficient and flexible network management and resource scheduling. The cloud-native concept emphasizes building and managing network functions (NFs) through technologies such as microservices, containerization, and dynamic scheduling, supporting automated network deployment, elastic scaling, and efficient operation. The Open Radio Access Network (O-RAN) organization has established a complete standards framework for integrating cloud-native network functions into terrestrial radio access networks.

[0003] With the development of wireless networks, satellite networks play a vital role in global connectivity and intelligent communication. Satellite networks possess characteristics distinct from terrestrial networks, particularly in terms of dynamic space topology and energy constraints. Existing O-RAN architectures rely on relatively static terrestrial network topologies for resource management and NF placement. Satellites, however, are highly dynamic, with their orbital positions and coverage areas changing over time. Furthermore, satellite networks lack cloud-native network functionality placement solutions.

[0004] There is currently no effective solution to the above problems. Summary of the Invention

[0005] This invention provides a system and method for placing network functions, thereby at least solving the problem that network function placement methods in related technologies are difficult to apply to satellite network environments.

[0006] According to an embodiment of the present invention, a system for placing network functions is provided, comprising: a non-real-time radio access network intelligent controller, the non-real-time radio access network intelligent controller including a graph model, a ground shadow prediction application, and a network function placement application, wherein the ground shadow prediction application determines a ground shadow prediction result based on the latest ephemeris information and sends the ground shadow prediction result to the graph model; the graph model receives performance measurement data; the graph model constructs a time-varying graph based on the ground shadow prediction result, the latest ephemeris information, and the performance measurement data, and sends the time-varying graph to the network function placement application; the network function placement application determines placement information for placing the network function based on resource information required for placing the network function and the time-varying graph.

[0007] In one exemplary embodiment, the service further includes: Open Radio Access Network (RAN) cloud resource management and orchestration services, wherein the RAN cloud resource management and orchestration services include joint RAN cloud orchestration and management, wherein the joint RAN cloud orchestration and management receives performance measurement data of the RAN cloud through a first interface and sends the performance measurement data to the graph model.

[0008] In an exemplary embodiment, before the Joint Open Radio Access Network Cloud Orchestration and Management receives performance measurement data from the Open Radio Access Network Cloud through the first interface, the system further includes: the graph model sending a performance measurement request to the Joint Open Radio Access Network Cloud Orchestration and Management; the Joint Open Radio Access Network Cloud Orchestration and Management sending the performance measurement request to the Open Radio Access Network Cloud through the first interface; wherein the performance measurement request is used to request the Open Radio Access Network Cloud to perform a performance test and return the performance measurement data.

[0009] In an exemplary embodiment, before the shadow prediction application determines the shadow prediction result based on the latest ephemeris information, the system includes: the graph model receiving the latest ephemeris information and sending the latest ephemeris information to the shadow prediction application.

[0010] In one exemplary embodiment, the network function placement application determines placement information for placing a network function by: the network function placement application receiving a placement task for a network function, and determining the placement information based on resource information required for placing the network function included in the placement task, and the time-varying graph.

[0011] In one exemplary embodiment, the performance measurement data includes at least one of the following: available computing resources, available storage resources, battery status, mirror resources, and link status of satellite nodes.

[0012] In an exemplary embodiment, the time-varying graph includes node features and edge features. The node features include at least one of the following: available computing resources, available storage resources, power status, mirror resources, and orbit status. The edge features include at least one of the following: satellite node link status, connectable time, available link bandwidth resources, and link health status.

[0013] In one exemplary embodiment, the resource information includes at least one of the following: computing resource requirements, storage resource requirements, latency requirements, image requirements, task execution duration, and estimated energy consumption.

[0014] In an exemplary embodiment, determining the placement information based on resource information required for placing the network function included in the placement task and the time-varying graph includes: the network function placement application extracting features from the resource information to obtain a first feature vector; the network function placement application extracting features from the node and edge features of the time-varying graph to obtain a second feature vector; the network function placement application performing preliminary screening of satellites in the open radio access network cloud based on the first and second feature vectors to obtain multiple first target satellites; the network function placement application extracting features from the node and edge features of the first target satellites in the time-varying graph to obtain a third feature vector; and the network function placement application inputting the first and third feature vectors into an agent to obtain the placement information.

[0015] In an exemplary embodiment, the agent is obtained through intelligent learning. During the learning process, the agent is guided to learn in the direction of high reward value through a reward function. During the learning process, the estimated placement information output by the agent is used to place network function samples to obtain environmental parameters, which are then input into the reward function to obtain the reward value. The environmental parameters include: load differences between satellite nodes, task execution delay, average power consumption, and whether the network function samples are successfully deployed and achieve the corresponding functions.

[0016] In one exemplary embodiment, the Open Radio Access Network (RAN) cloud resource management and orchestration service further includes an RAN cloud network function organization. After the network function placement application determines the placement information, the service further includes: the network function placement application sending the placement information to the RAN cloud network function organization; and the RAN cloud network function organization placing the network function through a second interface.

[0017] According to an embodiment of the present invention, a method for placing network functions is provided, applied to service management orchestration, comprising: receiving a placement task for a network function, wherein the placement task includes resource information required for placing the network function; determining placement information based on the resource information and a time-varying graph; and placing the network function according to the placement information.

[0018] In one exemplary embodiment, before determining placement information based on the resource information and the time-varying graph, the method further includes: sending a performance measurement request to the Open Radio Access Network (ORN) cloud via a graph model, wherein the performance measurement request is used to instruct the ORN cloud to perform performance measurements on each satellite; and receiving performance measurement data sent by the ORN cloud via the graph model, wherein the performance measurement data includes power status information of each of the satellites.

[0019] In one exemplary embodiment, initiating a performance measurement request to the Open Radio Access Network (ORN) cloud via a graph model includes: sending the performance measurement request to a first interface for orchestration and management of the ORN cloud via the graph model; receiving performance measurement data sent by the ORN cloud via the graph model includes: receiving the performance measurement data via the first interface for orchestration and management of the ORN cloud.

[0020] In an exemplary embodiment, before determining the placement information based on the resource information and the graph model, the method further includes: receiving the latest ephemeris information through the graph model; sending the latest ephemeris information to a shadow prediction application through the graph model, so that the shadow prediction application outputs a shadow prediction result based on the latest ephemeris information; and receiving the shadow prediction result sent by the shadow prediction application through the graph model.

[0021] In an exemplary embodiment, after receiving the shadow forecast result sent by the shadow forecast application through the graph model, the method further includes constructing the time-varying graph through the graph model based on the performance measurement data, the ephemeris information, and the shadow forecast result.

[0022] In an exemplary embodiment, determining placement information based on the resource information and the time-varying graph includes: sending the time-varying graph to a network function placement application via the graph model; extracting features from the resource information using the network function placement application to obtain a first feature vector; extracting features from the node and edge features of the time-varying graph using the network function placement application to obtain a second feature vector; performing preliminary screening of satellites in the open radio access network cloud using the network function placement application based on the first and second feature vectors to obtain multiple first target satellites; extracting features from the node and edge features of the first target satellites in the time-varying graph using the network function placement application to obtain a third feature vector; and inputting the first and third feature vectors into an agent using the network function placement application to obtain the placement information.

[0023] In an exemplary embodiment, the agent is obtained through intelligent learning. During the learning process, the agent is guided to learn in the direction of high reward value through a reward function. During the learning process, the estimated placement information output by the agent is used to place network function samples to obtain environmental parameters, which are then input into the reward function to obtain the reward value. The environmental parameters include: load differences between satellite nodes, task execution delay, average power consumption, and whether the network function samples are successfully deployed and achieve the corresponding functions.

[0024] In an exemplary embodiment, after determining the placement information based on the resource information and the characteristic information of each of the first target satellites through the network function placement application, the method further includes: sending the placement information to the Open Radio Access Network Cloud Network Function Code through the network function placement application, and placing the network function using a second interface through the Open Radio Access Network Cloud Network Function Code.

[0025] In one exemplary embodiment, the performance measurement data includes at least one of the following: available computing resources, available storage resources, battery status, mirror resources, and link status of satellite nodes.

[0026] In an exemplary embodiment, the time-varying graph includes node features and edge features. The node features include at least one of the following: available computing resources, available storage resources, power status, mirror resources, and orbit status. The edge features include at least one of the following: satellite node link status, connectable time, available link bandwidth resources, and link health status.

[0027] In one exemplary embodiment, the resource information includes at least one of the following: computing resource requirements, storage resource requirements, latency requirements, image requirements, task execution duration, and estimated energy consumption.

[0028] According to yet another embodiment of the present invention, a computer-readable storage medium is also provided, wherein a computer program is stored therein, wherein the computer program, when executed by a processor, implements the steps of the method described in any of the preceding claims.

[0029] According to yet another embodiment of the present invention, an electronic device is also provided, including a memory and a processor, wherein the memory stores a computer program and the processor is configured to run the computer program to perform the steps in any of the above method embodiments.

[0030] According to yet another embodiment of the present invention, a computer program product is also provided, comprising a computer program that, when executed by a processor, implements the steps of the method described in any of the preceding claims.

[0031] This invention provides a system for placing network functions, comprising: a non-real-time radio access network intelligent controller, which includes a graph model, a ground shadow prediction application, and a network function placement application. The ground shadow prediction application determines a ground shadow prediction result based on the latest ephemeris information and sends the result to the graph model. The graph model receives performance measurement data. Based on the ground shadow prediction result and the performance measurement data, the graph model constructs a time-varying graph and sends it to the network function placement application. The network function placement application determines placement information for the network function based on the time-varying graph. This achieves the goal of placing network functions within a satellite network. Therefore, it solves the problem that network function placement methods are difficult to apply to satellite network environments. Attached Figure Description

[0032] Figure 1 This is a schematic diagram of a system for placing network functions according to an embodiment of the present invention;

[0033] Figure 2 This is a flowchart of a method for placing network functions according to an embodiment of the present invention;

[0034] Figure 3 This is a schematic diagram of a satellite ground shadow according to an embodiment of the present invention;

[0035] Figure 4 This is a flowchart of the network function placement application according to an embodiment of the present invention;

[0036] Figure 5 This is a structural block diagram of a network function placement application according to an embodiment of the present invention;

[0037] Figure 6 This is an information interaction diagram of a network function placement method according to an embodiment of the present invention. Detailed Implementation

[0038] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings and examples.

[0039] It should be noted that the terms "first," "second," etc., in the specification, claims, and drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0040] The resource management and network function (NF) placement in the existing O-RAN architecture are based on a relatively static terrestrial network topology. However, satellites are highly dynamic; their orbital positions and coverage areas change over time, and communication link quality is affected by inter-satellite distances and obstructions. The existing O-RAN architecture struggles to adapt to the complex space-based environment, easily leading to service quality degradation or even service interruption. Therefore, capturing the dynamic spatiotemporal characteristics of the satellite open radio access network cloud (O-RAN Cloud, or O-Cloud) to optimize NF deployment and ensure service quality is a critical issue that urgently needs to be addressed.

[0041] The limited energy resources of satellite nodes complicate NF (Network Functions) deployment. Satellite systems rely on solar panels to convert solar energy into electricity. When the satellite is in sunlight, solar energy powers the payload and charges the batteries; however, when in the shadow or during eclipse, the satellite relies on limited battery power. Deploying too many NFs during the shadow period can lead to low power levels, causing partial network function interruptions and severely impacting service quality. Therefore, for O-Cloud, the management system must be highly adaptable, capable of real-time sensing and dynamic deployment of NF locations based on satellite battery power, power supply, and other information. Existing O-RAN architectures are primarily designed for relatively stable ground environments and lack real-time monitoring of power status and flexible NF deployment mechanisms.

[0042] The method embodiments provided in this application are applied to Figure 1 The system shown has network functionality, such as Figure 1 As shown, the system that houses network functions includes: Service Management Orchestration (SMO), Near-Real-Time RAN Intelligent Controller (Near-RT RIC), Open RAN Centralized Unit (O-CU), Open RAN Distributed Unit (O-DU), Open RAN Radio Unit (O-RU), and Open RAN Cloud (O-Cloud).

[0043] Service Management Orchestration (SMO) includes: Open Radio Access Network Cloud Resource Management and Orchestration Service (O-Cloud Resource Management and Orchestration Service) and Non-Real-Time RAN Intelligent Controller (Non-RT RIC). O-Cloud Resource Management and Orchestration Service includes: Open Radio Access Network Cloud Network Function Orchestration (NFO) and Joint Open Radio Access Network Cloud Orchestration and Management (FOCOM). Non-RT RIC includes: Network Function Placement Applications, Ground Shadow Prediction Applications, Other Applications, R1 Termination, and Graph Models.

[0044] The Service Management Orchestration (SMO) and O-Cloud are connected via the O2 Deployment Management Services (O2DMS) and O2 Infrastructure Management Services (O2IMS) interfaces. The SMO sends placement information to the Open Radio Access Network Cloud (O-Cloud) through the O2DMS interface, and the O-Cloud Network Functions Orchestration (NFO) instantiates network functions for O-Cloud through the O2DMS interface. The SMO also sends performance measurement requests to the O-Cloud through the O2IMS interface, and the O-Cloud returns performance measurement data to the SMO through the O2IMS interface.

[0045] The non-real-time wireless access network intelligent controller for placing network functions includes a graph model, a ground shadow prediction application, and a network function placement application. The ground shadow prediction application determines a ground shadow prediction result based on the latest ephemeris information and sends the result to the graph model. The graph model receives performance measurement data. The graph model constructs a time-varying graph based on the ground shadow prediction result, the latest ephemeris information, and the performance measurement data, and sends the time-varying graph to the network function placement application. The network function placement application determines the placement information for the network function based on the resource information required for placement and the time-varying graph.

[0046] In one exemplary embodiment, the Open Radio Access Network (ORN) cloud resource management and orchestration service includes joint ORN cloud orchestration and management, wherein the joint ORN cloud orchestration and management receives performance measurement data of the ORN cloud through a first interface and sends the performance measurement data to the graph model.

[0047] In one exemplary embodiment, before the Joint Open Radio Access Network Cloud Orchestration and Management receives performance measurement data from the Open Radio Access Network Cloud through the first interface, the method further includes: the graph model sending a performance measurement request to the Joint Open Radio Access Network Cloud Orchestration and Management; the Joint Open Radio Access Network Cloud Orchestration and Management sending the performance measurement request to the Open Radio Access Network Cloud through the first interface; wherein the performance measurement request is used to request the Open Radio Access Network Cloud to perform a performance test and return the performance measurement data.

[0048] In one exemplary embodiment, before the shadow prediction application determines the shadow prediction result based on the latest ephemeris information, the method further includes: the graph model receiving the latest ephemeris information and sending the latest ephemeris information to the shadow prediction application.

[0049] In one exemplary embodiment, the network function placement application determines placement information for placing a network function by: the network function placement application receiving a placement task for a network function, and determining the placement information based on resource information required for placing the network function included in the placement task, and the time-varying graph.

[0050] In one exemplary embodiment, the performance measurement data includes at least one of the following: available computing resources, available storage resources, battery status, mirror resources, and link status of satellite nodes.

[0051] In an exemplary embodiment, the time-varying graph includes node features and edge features. The node features include at least one of the following: available computing resources, available storage resources, power status, mirror resources, and orbit status. The edge features include at least one of the following: satellite node link status, connectable time, available link bandwidth resources, and link health status.

[0052] In one exemplary embodiment, the resource information includes at least one of the following: computing resource requirements, storage resource requirements, latency requirements, image requirements, task execution duration, and estimated energy consumption.

[0053] In an exemplary embodiment, determining the placement information based on resource information required for placing the network function included in the placement task and the time-varying graph includes: the network function placement application extracting features from the resource information to obtain a first feature vector; the network function placement application extracting features from the node and edge features of the time-varying graph to obtain a second feature vector; the network function placement application performing preliminary screening of satellites in the open radio access network cloud based on the first and second feature vectors to obtain multiple first target satellites; the network function placement application extracting features from the node and edge features of the first target satellites in the time-varying graph to obtain a third feature vector; and the network function placement application inputting the first and third feature vectors into an agent to obtain the placement information.

[0054] In an exemplary embodiment, the agent is obtained through intelligent learning. During the learning process, the agent is guided to learn in the direction of high reward value through a reward function. During the learning process, the estimated placement information output by the agent is used to place network function samples to obtain environmental parameters, which are then input into the reward function to obtain the reward value. The environmental parameters include: load differences between satellite nodes, task execution delay, average power consumption, and whether the network function samples are successfully deployed and achieve the corresponding functions.

[0055] In one exemplary embodiment, the Open Radio Access Network (RAN) cloud resource management and orchestration service further includes an RAN cloud network function organization. After the network function placement application determines the placement information, the service further includes: the network function placement application sending the placement information to the RAN cloud network function organization; and the RAN cloud network function organization placing the network function through a second interface.

[0056] This embodiment provides a method for placing network functions in the system described above for applying network placement functions. Figure 2 This is a flowchart of a method for placing network functions according to an embodiment of the present invention, such as... Figure 2 As shown, the process includes the following steps:

[0057] Step S202: Receive a network function placement task, wherein the placement task includes resource information required to place the network function;

[0058] Specifically, the network function placement application receives the network function placement task. The placement task includes resource information required for placing the network function. Specifically, features can be extracted from information such as the deployment descriptor. The extracted resource information includes: computing resource requirements, storage resource requirements, latency requirements, image requirements, task execution duration, and estimated energy consumption, resulting in a feature vector. .

[0059] Step S204: Determine placement information based on the resource information and the time-varying diagram;

[0060] A new graph model is added to Non-RT RIC, which models satellite nodes and their communication relationships as time-varying graphs to better capture the spatiotemporal dependencies between satellite nodes and provide a management view of satellite nodes for the O-RAN architecture.

[0061] Step S206: Place the network function according to the placement information.

[0062] In one exemplary embodiment, before determining placement information based on the resource information and the graph model, the method further includes: sending a performance measurement request to the Open Radio Access Network (ORN) cloud via the graph model, wherein the performance measurement request is used to instruct the ORN cloud to perform performance measurements on each satellite; and receiving performance measurement data sent by the ORN cloud via the graph model, wherein the performance measurement data includes power status information of each of the satellites.

[0063] Specifically, initiating a performance measurement request to the Open Radio Access Network (ORN) cloud via the graph model includes: sending the performance measurement request to a first interface for orchestration and management of the ORN cloud via the graph model; and receiving performance measurement data sent by the ORN cloud via the graph model includes: receiving the performance measurement data via the first interface for orchestration and management of the ORN cloud.

[0064] The first interface is Figure 1The O2 Infrastructure Management Services (O2IMS) interface is shown. The graph model initiates a performance measurement request to the Joint Open Radio Access Network Cloud Orchestration and Management (FOCOM) via Service Management Orchestration Service (SMOS). FOCOM initiates the same performance measurement request to the Open Radio Access Network Cloud via the first interface. The Open Radio Access Network Cloud performs the performance measurement as requested. The Open Radio Access Network Cloud returns performance measurement data to FOCOM via the first interface, including information such as available computing resources, available storage resources, battery status, mirror resources, and satellite node link status. FOCOM returns the performance measurement data to the graph model via SMOS communication.

[0065] In the above embodiments, by receiving the satellite's power status information in the first interface of the O-RAN architecture, multi-dimensional dynamic factors such as ephemeris information, power status information, and link health status are fully incorporated, which can more comprehensively reflect the actual operating status of the satellite node. In particular, the monitoring of power and ephemeris information effectively improves the adaptability of network function placement strategies in complex space-based environments and helps to ensure the stability of service quality.

[0066] In an exemplary embodiment, before determining the placement information based on the resource information and the graph model, the method further includes: receiving the latest ephemeris information through the graph model; sending the latest ephemeris information to a shadow prediction application through the graph model, so that the shadow prediction application outputs a shadow prediction result based on the latest ephemeris information; and receiving the shadow prediction result sent by the shadow prediction application through the graph model.

[0067] Specifically, operators can inject ephemeris information into the Non-RT RIC through the operation and control system, enabling it to have power and satellite position awareness and monitoring capabilities.

[0068] Ephemeris information is injected into the graphical model of the Non-RT RIC. This ephemeris information can come from the satellite control system and is used to represent the orbital state of satellite nodes. For example, it can be described using orbital six-factor notation, including:

[0069] 1) Semi-major axis a: describes the size and shape of the satellite orbit.

[0070] 2) Eccentricity e: describes the degree of orbital eccentricity and affects the change in distance between the satellite and the Earth.

[0071] 3) Orbital inclination i: Defines the angle of inclination of the satellite's orbital plane relative to the Earth's equatorial plane.

[0072] 4) Right ascension of the ascending node Defines the astronomical coordinates of the intersection point of the satellite's orbital plane and the Earth's equatorial plane.

[0073] 5) Perigee angle : Defines the angle of perigee (the point where the satellite is closest to Earth) in the satellite's orbit.

[0074] 6) True nearest point angle : Describes the satellite's position in its orbit, used to determine the satellite's specific location at any given time.

[0075] By adding ephemeris information, the specific location and orbital status of satellites can be monitored in real time, thus providing the necessary basis for the rational placement of network functions.

[0076] In this embodiment, the graph model sends the latest ephemeris data to the Earth Shadow Prediction Application and requests the Earth Shadow Prediction result; the Earth Shadow Prediction Application performs model inference based on the ephemeris data (which can use existing model inference techniques, such as neural networks), and outputs the Earth Shadow Prediction result. The Earth Shadow Prediction Application transmits the Earth Shadow Prediction result through a third interface ( Figure 1 The R1 interface in the graph and SMOS communication return to the graph model.

[0077] In an exemplary embodiment, after receiving the shadow forecast result sent by the shadow forecast application through the graph model, the method further includes: constructing the time-varying graph through the graph model based on the performance measurement data, the ephemeris information and the shadow forecast result.

[0078] Add a graph model to the Non-RT RIC. Receive performance measurement data from the O2 Infrastructure Management Service Interface (O2IMS interface) via the Joint Open Radio Access Network Cloud (FOCOM) and through a third interface ( Figure 1 The R1 interface in the cloud obtains the shadow forecast results and constructs a time-varying map based on the above information to accurately capture the state changes of the open wireless access cloud in different time periods, providing environmental information for the placement of network functions.

[0079] Time-varying graph This is used to represent the state of each satellite node and its communication link in the open radio access network cloud at time t, where It is a collection of satellite nodes, E It is a collection of communication links between nodes. The system maintains and updates this graph model at different times t to capture the dynamic changes of nodes and links. The time-varying graph includes node features and edge features, where,

[0080] (1) Node characteristics:

[0081] Nodes in a time-varying graph This represents a satellite node. Each satellite node has several attributes that change over time, reflecting its dynamic, time-varying state. For example, each satellite node... It can contain the following attributes:

[0082] Available computing resources, storage resources : This characterizes the types of computing resources possessed by satellite nodes (such as CPU, DPU, FPGA, etc.) and the available resource quantity for each type of computing resource. The amount of free storage resources a node possesses.

[0083] Image resources (list of cached image files) ): The image file resources contained in the satellite node. The system prioritizes allocating network functions to satellite nodes containing the required image files to reduce image retrieval latency and improve network function startup efficiency.

[0084] Orbital status (satellite ephemeris information) ): Records satellite orbital status information, used to define the satellite's position and motion status in orbit, which has an important impact on node availability, communication link establishment and maintenance, can help predict satellite communication windows and connectivity, and provide data support for land shadow forecasting.

[0085] Satellite power status This includes satellite battery power, predicted power consumption of onboard payloads, and solar power supply status. For example, using... Represents the satellite node at time t The battery capacity varies depending on the satellite's workload and solar charging status; the power consumption prediction of the onboard payload. This represents the predicted average power consumption of the spaceborne payload over the next T time slots; This indicates the solar power supply status, and this parameter depends on the relative positions of the sun, satellite, and earth.

[0086] like Figure 3 As shown, when the satellite enters the Earth's shadow region, it cannot charge and the satellite payload is powered by a battery. When the satellite is in the illuminated region, it can charge the battery and power the onboard payload. Whether the satellite is in the Earth's shadow region can be determined by the satellite shadow prediction algorithm. , This indicates that the satellite is currently in the illuminated area, and will remain in the illuminated area for a duration of n; conversely, This indicates that the satellite is currently in the Earth's shadow region, and will remain in the shadow region for a duration of n. Therefore, considering all the above factors, the satellite node... The power state can be represented as

[0087] (2) Edge features:

[0088] Satellite node link status: In the time-varying graph, edges Represents a node and nodes The communication links between nodes are represented by each edge, which reflects the connectivity between the nodes.

[0089] The state of communication links changes dynamically with time and satellite location, affecting network function placement information. For example, an edge may contain the following key attributes:

[0090] Connectivity time Communication links between satellites and between satellites and ground gateways are not permanent. Due to satellite movement, links between nodes may be lost at some point. Connectivity time reflects how long inter-satellite or satellite-to-ground links can maintain communication within a future period.

[0091] Link available bandwidth resources This indicates the maximum data transmission rate that can be supported between nodes. Link bandwidth may fluctuate as satellite positions change. Bandwidth resources are affected not only by the relative positions of the satellite and ground equipment, but also by other link traffic and network load, impacting data transmission efficiency and speed.

[0092] Link health status Link health is a comprehensive indicator of overall link performance, taking into account multiple factors such as signal interference, link congestion, and channel quality. Link health reflects the stability and quality of the link, determining whether it can maintain efficient and stable data transmission. In satellite networks, link health dynamically changes with environmental factors (such as weather conditions and space irradiance) and the operational status of the satellite.

[0093] To enhance adaptability to satellite mobility and power state changes during NF placement, power state information is added to the O2IMS interface, and ephemeris information is injected into the graphical model in the Non-RT RIC, including:

[0094] 1) The remaining power of the batteries on the satellite node reflects the satellite's energy reserve status.

[0095] 2) Open the wireless access network cloud energy consumption to help determine the rate of energy consumption.

[0096] 3) Other relevant information.

[0097] By introducing power status information and obtaining it in real time through the O2IMS interface, the system can make intelligent placement of NFs based on the power status information, preventing low-power satellite nodes from overloading and ensuring service quality stability.

[0098] In an exemplary embodiment, determining placement information based on the resource information and the graph model includes: sending the time-varying graph to a network function placement application via the graph model; extracting features from the resource information using the network function placement application to obtain a first feature vector; extracting features from the node and edge features of the time-varying graph using the network function placement application to obtain a second feature vector; performing preliminary screening of satellites in the open radio access network cloud using the network function placement application based on the first and second feature vectors to obtain multiple first target satellites; extracting features from the node and edge features of the first target satellites in the time-varying graph using the network function placement application to obtain a third feature vector; and inputting the first and third feature vectors into an agent using the network function placement application to obtain the placement information.

[0099] The graphical model updates the time-varying map based on performance measurement data and shadow prediction results; the graphical model sends the latest time-varying map to the network function placement application via SMOS communication and the R1 interface. The network function placement application determines the placement information based on the latest time-varying map and the resource information required to place the network function.

[0100] In one exemplary embodiment, after determining the placement information based on the resource information and the characteristic information of each of the first target satellites through the network function placement application, the method further includes: placing the network function according to the placement information.

[0101] The network function placement application sends the placement information to the Open Radio Access Network Cloud Network Function Orchestration (NFO). Specifically, the network function placement application sends the placement information to the Open Radio Access Network Cloud Network Function Orchestration (NFO) via the R1 interface and SMOS communication. The Open Radio Access Network Cloud Network Function Orchestration (NFO) performs NF instantiation operation on O-Cloud through the O2DMS interface.

[0102] like Figure 4 This is a flowchart of the application deployment of network functions, including:

[0103] The network function placement application receives the network function placement task. Feature extraction is performed on information related to the resources required for network functions, including computing resource requirements, storage resource requirements, latency requirements, mirroring requirements, task execution duration, and estimated energy consumption, to obtain feature vectors. (First eigenvector).

[0104] like Figure 5 This is a structural diagram of the network function placement application. The graph neural network extracts features of satellite nodes and edges based on time-varying graphs to generate high-dimensional feature vectors. (Second feature vector) contains information on various aspects such as node resources, power status, link health status, and cache status.

[0105] Node pre-screening. Based on the specific requirements for network function deployment, available satellite nodes are preliminarily screened to ensure that the screened nodes meet the basic conditions for network function deployment. Satellites that meet the basic conditions are identified as the primary target satellites.

[0106] For example, if a requirement is for Graphics Processing Unit (GPU) computing resources and a transmission bandwidth of at least Y Mbps (where Y can be set according to actual conditions), nodes with sufficient GPU computing power are first selected, and it is ensured that the communication link bandwidth of these nodes can meet the minimum transmission requirements of the task. This pre-screening process effectively reduces the computational complexity of subsequent information and ensures that only nodes that meet the requirements participate in further network function placement.

[0107] A deep reinforcement learning framework is employed to learn optimal network function placement information through the interaction between the agent and the environment. This reinforcement learning problem is modeled in the following form:

[0108] State: The input state of the agent includes high-dimensional environmental features obtained through graph neural network processing. and task characteristics Therefore, the state at time t can be expressed as:

[0109] ;

[0110] Action: i.e., the output of the intelligent agent It contains placement information for each node. .in, If network functions are included... If placed on node i, then ;on the contrary, .

[0111] ;

[0112] Reward function: After an agent performs an action, the environment provides reward feedback, guiding the agent to learn in directions with higher rewards. Consider a reward function that is applicable to multiple tasks. :

[0113]

[0114] in: This indicates the load balancing status and measures the load difference between different satellite nodes. The smaller the load difference, the more balanced the system load. This indicates the task execution delay, including task data transmission delay and task computation delay. This represents the average battery charge of the satellite during mission execution (the average battery charge of the satellite during mission execution after the network function is installed), used to incentivize agents to optimize node power usage to enhance system reliability. This indicates whether the network function has been successfully deployed and is performing its intended purpose, thus optimizing system reliability. If the network function fails to execute due to factors such as communication interruption or low satellite battery power, then... ;on the contrary, . It represents the weight of each reward item, used to balance the needs of different tasks and adjusted according to the characteristics of the task.

[0115] The training objective of a reinforcement learning agent is to learn a... This allows the action chosen at each time step to maximize the cumulative reward. For example, a model trained on the Proximal Policy Optimization (PPO) algorithm can be used for inference.

[0116] Based on the network function placement information generated by the reinforcement learning agent, instructions are issued to deploy network functions through FOCOM control of the corresponding satellite nodes.

[0117] like Figure 6 The information interaction diagram of the network function placement method shown includes:

[0118] S1, the operation and control system periodically injects ephemeris data into the graph model;

[0119] S2, the network function placement application sends a graph model request to the graph model through the R1 interface;

[0120] S3, the graphical model checks whether the time-varying graph has been updated. If not, the time-varying graph is updated using the following steps:

[0121] S31, the graph model initiates an O-Cloud performance measurement request to FOCOM via SMOS communication, including information such as computing resources, storage resources, battery power, image cache status, connectivity between nodes, and link status;

[0122] S32, FOCOM initiates a performance measurement request to O-Cloud via the O2IMS interface;

[0123] S33, O-Cloud performs performance measurements as required;

[0124] S34, O-Cloud returns performance measurement data to FOCOM via the O2IMS interface;

[0125] S35, FOCOM returns measurement results to the graph model via SMOS communication;

[0126] S36, the graphical model sends the latest ephemeris data to the Earth Shadow Prediction application and requests the Earth Shadow Prediction results;

[0127] S37, Earth shadow prediction application performs model inference based on ephemeris data and outputs earth shadow prediction results;

[0128] S38, the shadow prediction application returns the shadow prediction results to the graph model through the R1 interface and SMOS communication;

[0129] S39, the graphical model updates the time-varying map based on performance measurement results and ground shadow prediction results;

[0130] Steps S31 to S39 above are optional steps. If the time-varying graph is not updated, the graph model updates the time-varying graph through steps S31 to S39 above.

[0131] S4, the graph model sends the latest time-varying graph to the network function placement application through SMOS communication and R1 interface;

[0132] S5, the network function placement application determines the placement information based on the time-varying graph and the resource information required to place the network function;

[0133] S6, the network function placement application sends placement information to the NFO via the R1 interface and SMOS communication;

[0134] S7, NFO performs the instantiation of the placement network function on the Open Radio Access Network cloud via O2DMS.

[0135] In the above embodiments, the adaptability of the O-RAN architecture to the satellite environment is increased, which can effectively improve the quality of service, specifically including the following two aspects:

[0136] 1) Existing O-RAN architectures provide construction solutions for terrestrial networks, which are difficult to apply to satellite network environments. This solution is geared towards the 6G integrated space-ground trend and, considering the characteristics of the space-based environment, fully incorporates multi-dimensional dynamic factors such as ephemeris information, power status, and link health status. It can more comprehensively reflect the actual operating status of satellite nodes, especially in monitoring power and ephemeris information. This effectively improves the adaptability of network function placement strategies in complex space-based environments and helps ensure the stability of service quality.

[0137] 2) This solution adds a graph model to capture the spatiotemporal characteristics of highly dynamic satellite nodes. In the time dimension, it accurately describes the dynamic time-varying behaviors such as satellite orbital motion, link establishment and connection; in the spatial dimension, it describes the distribution characteristics of satellite nodes and communication links in space, captures the interaction relationships between different nodes, and provides O-RAN with a management view of open radio access network cloud nodes, enhancing its adaptability to the satellite environment.

[0138] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0139] Embodiments of the present invention also provide a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the steps of the method described in any of the preceding claims.

[0140] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard disk, magnetic disk, or optical disk.

[0141] Embodiments of the present invention also provide an electronic device including a memory and a processor, the memory storing a computer program and the processor being configured to run the computer program to perform the steps in any of the above method embodiments.

[0142] In one exemplary embodiment, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor and the input / output device is connected to the processor.

[0143] Specific examples in this embodiment can be found in the examples described in the above embodiments and exemplary implementations, and will not be repeated here.

[0144] Embodiments of the present invention also provide a computer program product, including a computer program that, when executed by a processor, implements the steps of the methods described in various embodiments of the present application.

[0145] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. They can be implemented using computer-executable program code, and thus can be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those described herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0146] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A system for incorporating network functionality, characterized in that, include: A non-real-time wireless access network intelligent controller, comprising a graph model, a ground shadow prediction application, and a network function placement application, wherein... The ground shadow prediction application determines the ground shadow prediction result based on the latest ephemeris information and sends the ground shadow prediction result to the graph model; The graph model receives performance measurement data; The graph model constructs a time-varying graph based on the ground shadow prediction results, the latest ephemeris information, and the performance measurement data, and sends the time-varying graph to the network function placement application; The network function placement application determines the placement information of the network function based on the resource information required to place the network function and the time-varying graph.

2. The system according to claim 1, characterized in that, Also includes: Open Radio Access Network (RAN) cloud resource management and orchestration services, wherein the open RAN cloud resource management and orchestration services include joint open RAN cloud orchestration and management, wherein... The joint Open Radio Access Network (ORN) cloud orchestration and management receives performance measurement data from the ORN cloud through the first interface and sends the performance measurement data to the graph model.

3. The system according to claim 2, characterized in that, Before the joint open radio access network cloud orchestration and management receives performance measurement data from the open radio access network cloud through the first interface, the following is also included: The graph model sends a performance measurement request to the Joint Open Radio Access Network cloud orchestration and management; The joint open radio access network cloud orchestration and management sends the performance measurement request to the open radio access network cloud through the first interface; The performance measurement request is used to request the open radio access network cloud to perform performance testing and return the performance measurement data.

4. The system according to claim 1, characterized in that, Before the shadow prediction application determines the shadow prediction result based on the latest ephemeris information, it also includes: The graph model receives the latest ephemeris information and sends the latest ephemeris information to the Earth Shadow Prediction Application.

5. The system according to claim 1, characterized in that, The network function placement application determines the placement information for the network function, including: The network function placement application receives a network function placement task and determines the placement information based on the resource information required to place the network function included in the placement task and the time-varying graph.

6. The system according to claim 1, characterized in that, The performance measurement data includes at least one of the following: available computing resources, available storage resources, battery status, mirror resources, and satellite node link status.

7. The system according to claim 1, characterized in that, The time-varying graph includes node features and edge features. The node characteristics include at least one of the following: available computing resources, available storage resources, power status, image resources, and track status; The edge features include at least one of the following: satellite node link status, connectable time, available link bandwidth resources, and link health status.

8. The system according to claim 1, characterized in that, The resource information includes at least one of the following: computing resource requirements, storage resource requirements, latency requirements, image requirements, task execution duration, and estimated energy consumption.

9. The system according to claim 5, characterized in that, The placement information is determined based on the resource information required for placing the network function included in the placement task, and the time-varying graph, including: The network function placement application performs feature extraction on the resource information to obtain a first feature vector; The network function placement application extracts features from the node and edge features of the time-varying graph to obtain a second feature vector; The network function placement application performs preliminary screening of satellites in the open radio access network cloud based on the first feature vector and the second feature vector to obtain multiple first target satellites; The network function placement application extracts features from the first target satellite node features and edge features in the time-varying graph to obtain a third feature vector; The network function placement application inputs the first feature vector and the third feature vector into the agent to obtain the placement information.

10. The system according to claim 9, characterized in that, The agent is obtained through intelligent learning. During the learning process, the agent is guided to learn in the direction of high reward value through a reward function. The environmental parameters obtained after the agent outputs the estimated placement information to place the network function sample are input into the reward function to obtain the reward value. The environmental parameters include: load difference between satellite nodes, task execution delay, average power consumption, and whether the network function sample is successfully deployed and achieves the corresponding function.

11. The system according to claim 2, characterized in that, The Open Radio Access Network (ORN) cloud resource management and orchestration service also includes ORN cloud network function orchestration, and after the network function placement application determines the placement information, it further includes: The network function placement application sends the placement information to the open wireless access network cloud network function orchestration. The open wireless access network cloud network function orchestration places the network functions through the second interface.

12. A method for placing network functionality, characterized in that, The system applied to any one of claims 1 to 11 comprises: Receive a network function placement task, wherein the placement task includes resource information required to place the network function; The placement information is determined based on the resource information and the time-varying diagram; The network function is placed according to the placement information.

13. The method according to claim 12, characterized in that, Before determining placement information based on the resource information and the time-varying diagram, the method further includes: A performance measurement request is sent to the Open Radio Access Network cloud via a graph model, wherein the performance measurement request is used to instruct the Open Radio Access Network cloud to perform performance measurements on each satellite; The performance measurement data sent by the open radio access network cloud is received through the graph model, wherein the performance measurement data includes the power status information of each of the satellites.

14. The method according to claim 13, characterized in that, Initiating a performance measurement request to the Open Radio Access Network Cloud via a graph model includes: sending the performance measurement request to a first interface orchestrated and managed by the Joint Open Radio Access Network Cloud via a graph model; Receiving performance measurement data sent by the Open Radio Access Network (ORN) cloud through the graph model includes: receiving the performance measurement data through a first interface jointly orchestrated and managed by the ORN cloud.

15. The method according to claim 13 or 14, characterized in that, Before determining the placement information based on the resource information and the graph model, the method further includes: The latest ephemeris information is received through the graph model; The graph model sends the latest ephemeris information to the shadow prediction application, so that the shadow prediction application outputs shadow prediction results based on the latest ephemeris information. The graph model receives the shadow prediction results sent by the shadow prediction application.

16. The method according to claim 15, characterized in that, After receiving the shadow prediction result sent by the shadow prediction application through the graphical model, the method further includes, The time-varying map is constructed using the graphical model based on the performance measurement data, the ephemeris information, and the ground shadow prediction results.

17. The method according to claim 12, characterized in that, The placement information is determined based on the resource information and the time-varying diagram, including: The time-varying graph is sent to the network function placement application via the graph model; The application uses the network function to extract features from the resource information to obtain a first feature vector. The network function is used to place applications to extract features from the node and edge features of the time-varying graph, thereby obtaining a second feature vector; The network function placement application performs preliminary screening of satellites in the open radio access network cloud based on the first feature vector and the second feature vector to obtain multiple first target satellites; The network function is used to place applications to extract features from the first target satellite node features and edge features in the time-varying graph, and a third feature vector is obtained. The first feature vector and the third feature vector are input into the agent through the network function placement application to obtain the placement information.

18. The method according to claim 17, characterized in that, The agent is obtained through intelligent learning. During the learning process, the agent is guided to learn in the direction of high reward value through a reward function. The environmental parameters obtained after the agent outputs the estimated placement information to place the network function sample are input into the reward function to obtain the reward value. The environmental parameters include: load difference between satellite nodes, task execution delay, average power consumption, and whether the network function sample is successfully deployed and achieves the corresponding function.

19. The method according to claim 17, characterized in that, After the application determines the placement information based on the resource information and the characteristic information of each of the first target satellites through the network function placement application, the method further includes: The network function placement application sends the placement information to the Open Radio Access Network Cloud Network Function Orchestration service, which includes the Open Radio Access Network Cloud Resource Management and Orchestration service. The open wireless access network cloud network function orchestration places the network functions through the second interface.

20. The method according to claim 12, characterized in that, The resource information includes at least one of the following: computing resource requirements, storage resource requirements, latency requirements, image requirements, task execution duration, and estimated energy consumption.

21. The method according to claim 12, characterized in that, The time-varying graph includes node features and edge features. The node characteristics include at least one of the following: available computing resources, available storage resources, power status, image resources, and track status; The edge features include at least one of the following: satellite node link status, connectable time, available link bandwidth resources, and link health status.

22. The method according to claim 13, characterized in that, The performance measurement data includes at least one of the following: available computing resources, available storage resources, battery status, mirror resources, and satellite node link status.

23. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein the computer program, when executed by a processor, implements the steps of the method described in any one of claims 12 to 22.

24. An electronic device comprising a memory and a processor, characterized in that, The memory stores a computer program, and the processor is configured to run the computer program to perform the method as described in any one of claims 12 to 22.

25. 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 described in any one of claims 12 to 22.

Images